6 Aug 2025

Roles, Goals, and Limitations – Prompting Your Large Language Model Leads to Better Responses

by | | 0

By Marc McCarty (assisted by several LLMs)

Ever since OpenAI burst on the scene in late 2022 with its free version of ChatGPT many of us have had a love-hate relationship with large language models (LLMs). On the one hand we have been blown away by the capabilities of these chatty, helpful, bots, but on the other we find they sometimes wildly misinterpret what we are asking, or produce results that, while they appear to be accurate and useful, bear no relationship to reality!

In an earlier post I explained how LLMs differ from traditional search engines like Google or Bing, and why this happens.

In this post, I’ll discuss an important tool you can use to improve your interactions with LLMs, and increase the likelihood they will provide relevant, accurate responses. You can do this by developing thoughtful habits that govern how you ask the LLM for assistance. This is called “prompting.”  

What Do We Mean by Prompting?

“Prompting” is just the term used to describe the instructions we provide to an LLM when we initiate an interaction.  Understanding the anatomy of a good prompt can appreciably increase your chance of getting an accurate and relevant information from the LLM.

I believe the three core characteristics of a good prompt can be summarized in three words —Roles, Goals, and Limitations.

Roles: Defining Who’s Who in the Conversation

Assigning clear roles in your prompt guides the LLM to adopt the right tone, expertise, and style when responding, and it can also help the model search out the best training data to use when formulating its response. Defining the roles involves two perspectives:

  • The LLM’s Role: Specify what persona or expertise you want the model to embody. For example, “Act as a financial advisor,” or “You are a high school science teacher.” This helps the model tailor its language, depth, and focus, producing responses that are contextually appropriate and that are aligned with your expectations.
  • The User’s Role: Describe your own background or needs. For example, if you are looking for ideas to plan a vacation, what are the ages and interests of the folks who will be making the trip? This information can help guide the LLM to areas of training more closely aligned to your objectives.

Goals: Clarifying the Desired Outcome

Every effective prompt is anchored in a clear goal. Before you ask, consider:

  • What do you want to achieve? Are you seeking a summary, a detailed explanation, a list of pros and cons, a creative story, or something else? Articulate your objectives explicitly.
  • What does success look like? Describe the ideal output format—should it be a bulleted list, a step-by-step guide, or a concise paragraph? Do you want the response to include specific hyperlinks to websites supporting the response? – Ask for it! Remember, the more specific you are, the more likely the LLM will align its response to fit your needs.

Limitations: Setting the Rules and Boundaries

Outlining the acceptable limits you want to impose on the LLM is important as well, and it can help control the quality and reliability of the output. Think in terms of how you would advise a “human” assistant who was doing research to provide your desired response (your “goals”).

For example, maybe it is very important that the LLM seek out the most current information available. If the model (like most do today) can perform an internet search to update its response, be sure to reinforce the need for it to do that when crafting an answer for you. If you want each assertion provided in the response to contain links to reliable internet-based source material; you can instruct the model to provide those.

It’s also a good idea to ask your LLM not to hallucinate responses. That might sound a bit odd, after if you are asking for advice on the best restaurant for Indian food in Little Rock, why would the LLM ever think you wanted it to make up a dining location to provide that experience? 

The answer is that sometimes users want a creative response, not necessarily an accurate “real world” example. Sometimes users want the LLM to hallucinate – to be “creative” and take existing data and extrapolate and arrange it in new format.  Asking the LLM not to do this (assuming you need factual responses), apparently does reduce the tendency the model has to guess or make up a source information source. Positive requests such as “provide sources for any conclusion that does not involve common knowledge” or “if you have inadequate data to state a conclusion, say so — do not speculate” can help.

You might also insist that the LLM take on your task one step at a time, proceeding to the next step after it has displayed results and received your permission to continue.  This can be particularly useful in helping to direct the model, so it doesn’t waste time and energy going down a path that isn’t relevant or useful.          

A Practical Example: Putting It All Together

Here’s an example of how you might structure a prompt using these three characteristics outlined above:

“You are an AI language model acting as a career counselor. I am a college student with no prior work experience, seeking advice on entering the tech industry. Please provide a step-by-step guide detailing how I can determine what options are available and how I can determine if it would be a good “fit” for me.  Consider what are the most relevant interests, character traits, course selection and life goals. Your response should include major bullet points and subcategories. It should be written in a conversational tone. You should include reputable sources for each step and avoid speculation—stick to verifiable information only.” Limit your response to approximately 750 words.”

  • Roles: AI as career counselor; user as college student.
  • Goals: Step-by-step guide, bullet points, reputable sources.
  • Limitations: No speculation, only verifiable information, citations required, limit length

Experiment, Practice, and Don’t Expect Perfection

We tend to forget that we are at the very earliest stages of learning how best to interact with LLMs. There are many experts out there with ideas about how to work with these cheerful bots. At the same time programmers at each of the companies that created LLMs are constantly working to improve them, to make them more intuitive, responsive and less prone to hallucination. 

For now the best strategy seems to be err on the side of providing LLMs too much, rather than too little, guidance and at least from my experience focusing on the three  characteristics of a good prompting: “roles, goals and limitations” is a simple, yet effective way to produce better, more accurate results from your interactions with LLMs.

However, “better” is not “perfect.” Even if you tell prompt an LLM not to hallucinate a response, it will still do so from time to time. Even if you ask it to provide you with authoritative references, you’ll from time to find it provides you broken website links, or a citation to a source that does not exist. For now, this is an issue we will have to accept. When accuracy is important – as it often is – you simply must check and in some cases double check an LLM’s response to make sure it isn’t “making stuff up.”

Yet, even with these trade-offs, I found that I use a LLMs more each day. While no LLM wrote this Post for me, it did shorten the time needed to spend working on it, and it offered several suggestions to improve on what I wrote. The net result, at least I believe, is a better product, produced in less time.

For me, that is not a bad trade off.

Happy Prompting!

6 Aug 2025

Trust But Verify — How LLMs Differ from Search Engines—and Why They Sometimes Hallucinate

by | | 0

Large Language Models (LLMs) like ChatGPT and traditional search engines such as Google may both help you find information, but they work in fundamentally different ways. Understanding these differences is key to using LLMs effectively—and to recognizing why LLMs sometimes “hallucinate,” or make up information.

How a Search Engine Works

A traditional search engine such as Google or Microsoft Edge is essentially a giant, constantly updated database of web pages. When you enter a query, the search engine:

  • Scans its index of the web for pages containing your keywords.
  • Ranks those pages by relevance, popularity, and other factors.
  • Shows you a list of links to real, existing web pages for you to explore.

Search engines don’t generate new information—they retrieve what already exists, helping you find the most relevant sources for your query. It’s possible, of course that the information retrieved may be inaccurate, or that your search request may have been incomplete and missed some important information, but a “search engine” is just that, an engine designed to seek out and retrieve information contained somewhere on the internet.

How an LLM Works

An LLM, by contrast, is trained on huge amounts of text data (books, articles, websites) to learn the patterns and relationships between words and ideas. When you ask it a question or request (this is called a “prompt”), an LLM:

  • Analyzes your prompt to understand context and intent.
  • Predicts and generates a sequence of words that best fits your request, based on its training data and learned patterns.

Initially LLMs did not look up answers live on the web (although many now use your prompt to generate an internet search and adds that information to its database). But even in this case the response is generated by “guessing” the most likely next word in a sentence, drawing on what the model was exposed to during training.

Why LLMs Hallucinate

Because LLMs generate text based on statistical patterns—not by checking facts—they can sometimes produce information that sounds plausible, but isn’t true. This is called “hallucination.” For example, if asked about a recent event that happened after its last training update, the LLM might invent details, unless it is able to access real-time data.

Hallucinations can also happen if:

  • The model’s training data was incomplete or contained errors.
  • The prompt is unclear or ambiguous.
  • The model tries to fill in gaps with its best “guess,” even if there’s no factual basis.

Remember, LLMs have no concept of facts/knowledge and just learn statistical patterns of how likely it is one word follows another (given a context).

LLMs with Web Search: A Step Forward (But Not Perfect)

Newer LLMs can now perform real-time web searches to supplement their responses. When you ask a question, these LLMs:

  • Runs an internet search to find relevant, up-to-date information.
  • Incorporates this information into its answer, often providing citations or links.

This hybrid approach—sometimes called Retrieval-Augmented Generation (RAG)—helps the LLM provide more accurate and current answers, especially for recent events or facts outside its training data.

However, even with web search capabilities, LLMs can still hallucinate:

  • They might misinterpret search results or combine information incorrectly.
  • They can still generate plausible sounding but false statements if the search doesn’t return clear or accurate data.
  • The process of merging search results with generated text can introduce new errors or fabrications.

Key Takeaways for Users

  • Search engines retrieve: They show you real, existing web pages that match your keywords.
  • LLMs generate: They create new text based on patterns in their training data, not by looking up facts in real time.
  • LLMs can hallucinate: Because they generate responses, they sometimes make up information that sounds real but isn’t.
  • Web search integration helps: LLMs that use live web search can provide more current, accurate information—but they’re still not perfect and can hallucinate.

Understanding these differences will help you use both tools wisely—and always double-check important facts, especially when using LLMs for critical information.

18 Jun 2025

The New 2025 BEAD Restructuring Policy Notice: What Missouri Community Stakeholders Need to Know

by | | 0

The long-delayed effort to begin spending nearly $42.5 billion in broadband infrastructure funding authorized by Congress nearly four years ago is still unfolding. This blog is intended to help community stakeholders—including local government officials, chambers of commerce, nonprofits, and engaged citizens—understand the current situation and work productively to ensure the best possible broadband service is deployed as quickly as possible in their community.

The latest chapter in the BEAD funding saga is the release of a newly revised set of requirements that states must follow to access their share of available BEAD funds. This policy—titled the “Broadband Equity, Access, and Deployment (BEAD) Program: BEAD Restructuring Policy Notice”(the “Policy Notice”) was released June 6, 2025, and it represents a significant shift in federal broadband funding priorities, eliminating or revising many of the earlier requirements imposed by the Biden administration in the original BEAD NOFO, published three years ago, and supplemented with various updates that states had been working to fulfill up until a few weeks ago.

The new Policy Notice emphasizes swift action and sets an aggressive timeline for implementation. Much of the work previously completed by Missouri’s Office of Broadband Development (OBD), including the state’s Initial Proposal and the implementation efforts described in an earlier blog, will now need to be revised to comply with the new requirements outlined in the Policy Notice.

The number of Broadband Serviceable Locations (BSLs) that remain eligible for BEAD funding also may be reduced, because the Policy Notice now permits Unlicensed Fixed Wireless (ULFW) providers to assert they are now providing the broadband service at the level required by the BEAD Act. A new challenge process enabling ULFW providers to provide evidence to OBD is now underway and it should be completed within a few weeks.

All this comes against the backdrop of OBD’s highly successful first round of funding under the prior BEAD NOFO guidance. While specific winners of round one and its sub rounds were not announced, approximately 90% of the eligible locations received proposals and OBD did release information identifying the BSLs where a proposal had been preliminarily accepted. All those results will have to be thrown out. OBD and the ISPs that decide they still want to participate in the BEAD funding program will need to quickly adjust to a new set of rules and priorities and resubmit under a new set of scoring criteria. All this must happen in time for OBD to include the results in its revised final proposal to NTIA in early September.

The new required “Benefit of the Bargain” funding round (described below) is necessary because the Policy Notice, eliminates or substantially modifies criteria for awarding BEAD funding, and it opens up the process to a wider group of internet providers.

Here are some of the most significant changes:

New Definition of a “Priority Broadband Project.” The BEAD statute allowed states to prioritize certain projects over other types of technologies. These are referred to as “Priority Broadband Projects.” The Biden administration’s BEAD NOFO limited this definition to end to end fiber internet (fiber to the premises or FTTP). FTTP was favored over lower cost technologies so long as the amount of BEAD funding requested did not exceed the state’s Extreme High Cost Per Location Target (EHCPLT) for the location. The EHCPLT limit was imposed to make sure that all locations that were not adequately served could receive BEAD funding.

The Policy Notice broadens Priority Broadband Project definition to include any technology that the state (OBD in Missouri) determines can provide broadband service at speeds of no less than 100/20 Mbps with latency less than or equal to 100 ms., and that can easily scale speeds over time to meet the evolving connectivity needs of households and businesses and support the deployment of 5G, successor wireless technologies, and other advanced services.

This definition is important, because all Priority Broadband Projects will compete for BEAD funding using the same scoring matrix in the Benefit of the Bargain funding described below. (Policy Notice §3.1; pp.8-9).

ULFW Potentially Qualifies as a “Priority Project.” To qualify as a Priority Broadband Project, unlicensed fixed wireless (ULFW) must deliver reliable internet to all covered BSLs at speeds at 100/20 Mbps with latency under 100 ms. The BEAD Policy states that a ULFW provider’s network will be considered capable of reliably delivering service at the minimum levels required by the statute if it can demonstrate the ability to deliver download speeds of at least 5 Mbps assuming simultaneous connection at all locations. The BEAD Policy leaves the requirements and standards necessary for a ULFW provider to show its project is easily scalable over time to meet future needs to the discretion of each state. However, NTIA does reserve to itself the right to overrule the state’s determination in cases where it disagrees. (Policy Notice § 3.2; p.10; Appendix A)

LEO Satellite Potentially Qualifies as a “Priority Project.”  Low earth orbit satellite internet providers (LEO Satellite), such as Starlink, also could qualify as a Priority Broadband Project as well, if the provider commits to make sufficient bandwidth available on the network to serve all locations for a period of at least 10 years. States may adopt enforcement mechanisms—such as claw backs of BEAD grants—to ensure compliance; however, they may not take any ownership or security interest in the provider’s network equipment. (Policy Notice § 3.2; p.10; Appendix B)

New “Lowest Cost to the Government” Scoring Matrix Required. Generally, OBD is required to use a scoring matrix for all Priority Broadband Projects that favors the proposal or combination of proposals resulting in the lowest dollar outlay of BEAD funding.

However, so long as the EHCPLT is not exceeded, a higher-cost Priority Broadband Project proposal may be selected if it is within 15% of the amount requested by the lowest Priority Broadband Project  proposal.

In this case OBD may select among the Priority Broadband Project proposals based on a scoring matrix that takes only the following criteria into account:

  • The speed at which the project will be completed (with preference for faster deployment)
  • The technical capabilities and scalability of the proposed technology (for example, fiber’s ability to easily upgrade speeds in the future)
  • The reliability of the service, especially in challenging environments
  • Whether the applicant was previously a provisional subgrantee under the old rules (though this is only a minor consideration)

If all Priority Broadband Projects exceed the EHCPLT, OBD may accept a lower cost non-priority option if it meets the 100/20 Mbps and 100 ms. service level, even though it is not able to easily scale to meet future needs.(Policy Notice §§ 3.3–3.4; pp.10-13 )

NTIA Specific Oversight on Project Selection. While states retain broad authority over project selection, the BEAD Policy cautions states against setting EHCPLT at an unrealistically high level to preserve the possibility of fiber deployment. NTIA also reserves the right to disapprove project selections that it believes are inconsistent with BEAD program goals.
(Policy Notice p.9 and §3.4; pp.9 and 11-13)

Project Area Proposals Can Exclude BSLs. OBD must accept proposals that exclude one or more BSLs or provides for different delivery technologies where the provider determines the cost of deployment for a specific technology is too high. In such cases, OBD may solicit alternative solutions from other providers or consider a provider’s proposal to use different technologies for those BSLs.(Policy Notice §3.3; p.11)

Affordability Rules. OBD can no longer impose a specific Low-Cost Service Option (LCSO) that BEAD-funded projects must offer. Instead, providers must propose their own affordable plans, which must be made available to households eligible under the FCC’s Lifeline Program, rather than the broader criteria previously allowed under the Affordable Connectivity Program.
(Policy Notice §§ 2.6; 2.7; pp.6-9)

Open Access/Net Neutrality. Prohibitions on data caps and requirements to provide wholesale or open-access connections to BEAD-funded networks have been abolished. NTIA concluded these mandates discouraged participation and increased project costs.
(Policy Notice §2.3; p.5)

DEI, Workforce and Labor Requirements. Prior BEAD NOFO mandates concerning diversity, equity, and inclusion (DEI), fair labor practices, and related reporting have been eliminated. Instead, subgrantees are only required to certify compliance with applicable federal labor laws.(Policy Notice § 2.1; p.4)

Climate Resilience Factors.  The requirement that BEAD proposals be scored based on their contribution to climate resilience has been eliminated. Project participants will satisfy this statutory requirement by establishing risk management plans that account for technology infrastructure reliability and resilience to natural disasters (e.g., wildfires, flooding, tornadoes, hurricanes, etc.) as well as cybersecurity best practices..
(Policy Notice § 2.2; pp. 4-5)

No Credit for Community Input. OBD may no longer award scoring points or preference based on stakeholder engagement, demographic targeting, or other forms of community input. Public comment is still required for the Final Proposal, but it cannot affect project scoring.(Policy Notice § 2.4; pp. 5-6)

Where Do We Go From Here?

Admittedly more guidance will be issued in the coming weeks from OBD as it moves forward to implement the Policy Notice, and likely from NTIA as well. There also is the possibility that certain aspects of the Policy Notice will be challenged in court and the process will be further delayed based on a claim that it strays too far from BEAD’s statutory mandates.

However, it seems more likely that OBD will publish a new BSL map reflecting a revised set of BEAD funding eligible sites in a few weeks and a revised invitation for proposals under the Benefit of the Bargain funding round. That invitation also may provide further guidance (consistent with both the statute and the Policy Notice) detailing how ULFW and LEO Satellite proposals can qualify as a Priority Broadband Project.

For community stakeholders, the greatest risk may be that the providers that won preliminary allocations of BEAD funding under the BEAD NOFO process will simply give up, because they believe their FTTP proposal cannot overcome the construction and installation cost advantage of a ULFW solution. Even though many of these ISPs already have spent thousands of dollars on developing their proposals, after years of work, preparation and waiting, they may be at the point where they view further participation in the BEAD program as throwing good money after bad.

Yet the perceived cost advantage of these wireless technologies when compared to FTTP or hybrid fixed wired solutions may be illusory, depending on how OBD (and NTIA) hold UFLW, LEO Satellite or other technologies to the statute’s mandates for service that is easily scalable and future-proof. Much of Missouri’s unserved and underserved locations are in areas where terrain makes deployment of reliable wireless service challenging, and these challenges increase as future needs for higher connection speeds and lower latency require towers to be located closer to the end user and backhaul capacity to increase.

This is illustrated by the independent study commissioned by OBD in 2022. It is worth remembering that this study found that the overall cost of deploying wireless internet statewide was approximately the same as a 100% FTTP solution. While there likely are areas in Missouri, and throughout the United States, where wireless technologies have a substantially lower deployment cost and also can be easily scaled it meet future needs, the 2022 study appears to show that for many Missouri locations, if they are engineered properly to meet specific conditions and terrain challenges, wireless may cost more to deploy than FTTP.

The point here is that community stakeholders who want more than the minimum level of broadband service required by BEAD in their community, need to reach out to the local ISPs that participated in the aborted NOFO BEAD funding round, and urge them to move forward with a new proposal in the Benefit of the Bargain round this summer. They also should be receptive and at least consider joining with a provider in ways that allow them to reduce their  BEAD funding request, so it can fall within the limit of the new 15% rule imposed by the Policy Notice. For example, a local community might agree to fund a fixed a portion of a FTTP network provider’s cost to expand service to a few critical high cost BSLs in exchange for the provider’s agreement to reduce its BEAD funding request by a like amount. Since it is the amount of BEAD funding requested (rather than the cost of the network) that governs, the FTTP bid might be reduced to a point that it did not exceed 15% of a competing ULFW or LEO Satellite Priority Broadband Project proposal.

As the BEAD program enters this critical phase, efforts made by local leaders and  stakeholders still could help determine whether their community secures broadband infrastructure that just meets minimum federal thresholds, or instead infrastructure that is far more durable and future-proof. Now more than ever, communities need to support and encourage ISPs that have already submitted proposals that can meet this long-term vision.

13 May 2025

Missouri’s BEAD Program 2025

by | | 0

Just a Speed Bump or Is the Broadband Bus Running on Empty?

In June 2023, the National Telecommunications and Information Agency (NTIA) announced that Missouri had been allocated over $1.7 billion through the federal Broadband Equity, Access, and Deployment (BEAD) program, to expand high-speed internet access to more than 200,000 unserved and underserved locations across the state. This was the third highest amount of set aside for any state, with only California and Texas receiving larger allocations. The  BEAD funding was part of the Infrastructure Investment and Jobs Act (IIJA) enacted back in 2021, and the amount represented a historic opportunity to close Missouri’s digital divide and stimulate economic growth especially in rural unserved and underserved communities.

However “allocating” money isn’t the same thing as distributing it … either to the state or to the ISPs that will ultimately construct and operate new internet networks that ultimately will construct and operate new internet networks. Actually getting the money to where it could be used for an internet construction project has been painfully slow. At this point no money BEAD has been committed to a broadband project in Missouri, and it is looking less likely that any will be at this year.

This Blog describes why that’s happened, and suggests some steps communities might consider as we wait for the money.

BEAD Funding …. It’s a Process.

Even though BEAD was enacted in 2021 the “roll out” almost immediately encountered a substantial delay. It took a year and a half for the FCC to put together a set a map that NTIA could use to apportion the $44.24 billion Congress made available among the states. Much of that delay occurred because the FCC refused to begin working on the maps until Congress appropriate money specifically to fund the cost of preparing maps. Once the mapping process was completed, NTIA then required states to complete a multistage process that was roughly based on requirements outlined in the BEAD statute.

NTIA’s process required states to first determine locations eligible for funding and to create a procedure for selecting which ISPs would receive BEAD money. States were required to submit this procedure to NTIA for approval (Preliminary Plan Approval). Once Preliminary Plan Approval occurred, each state had one year to actually identify project locations eligible for funding and to conduct the  competitive process to select ISPs to build the broadband infrastructure using BEAD funds. The result of this work also had to be submitted to NTIA as the state’s Final Plan, and NTIA then would have an undetermined length of time to “approve it” (Final Plan Approval). Only after Final Plan Approval would ISPs selected in the state’s competitive grant process actually be assured they actually would receive the BEAD money, once the project was completed in accordance with the grant requirements.

Missouri Receives Preliminary Approval from NTIA and the Project Selection Process Begins

It took NTIA over a year (August 2, 2024) to give Missouri “Preliminary Plan Approval.” Thereafter, Missouri’s Office of Broadband Development (OBD) made significant progress moving toward a completion of a Final Plan for NTIA’s Final Plan Approval. Last year OBD conducted a rigorous challenge process to ensure that the locations designated as eligible for BEAD funding were accurate. It prequalified ISPs to streamline the process of reviewing ISP proposals to provide service in those areas, and it completed a first round of  BEAD funding applications, receiving 519 applications covering 192,284 locations-about 90% of the sites eligible for BEAD money.

A Short Pause…Or a Substantial Multi-Year Delay?

The acronym BEAD stands for Broadband Equity Access and Deployment, and for that reason alone, it’s not surprising that the incoming Trump administration would want some changes. NTIA is part of the Commerce Department, and  Commerce Secretary Howard Lutnick has consistently criticized the BEAD program for what he describes as “woke mandates,” including labor and climate-related requirements, and a “favoritism towards fiber-optic technology” that he believed needlessly inflated project costs and delayed infrastructure deployment.

During his January 2025 Senate confirmation hearing, Secretary Lutnick declined to commit to honoring NTIA’s existing Preliminary Plan Approvals, stating he would review plans for efficiency and alignment with “lowest cost” objectives. In March 2025, he announced a 90-day review focused on  Biden-era rules, prioritizing a tech-neutral approach (including satellite and fixed wireless) and streamlining infrastructure construction. The goal of this approach he argued was to accelerate deployment and reduce taxpayer costs. That has since been followed by an across the board 90 day extension in the date States must submit their Final Plan for NTIA approval.

Meanwhile, in March, congressional Republicans introduced the SPEED for BEAD Act to codify some of Secretary Lutnick’s priorities. Key provisions of this bill include:

               •Mandating technology neutrality, removing fiber-first preferences.

               •Eliminating labor (e.g., prevailing wage) and climate resilience requirements for broadband infrastructure.

               •Restricting BEAD funds to for infrastructure deployment only, blocking states from allocating funds for digital equity or workforce programs.

               •Requiring states to prioritize cost-effectiveness in subgrantee selection.

The legislation aligns with broader GOP efforts to redirect BEAD toward “shovel-ready” projects. Undoubtedly, it also puts fixed wireless providers and satellite providers in a much more competitive position as well, as they generally are less costly, at least in the short run, when compared to fiber to the premises broadband.

Of course, there are solid arguments supporting all of these proposals, and admittedly, the speed of implementing BEAD up to this point has been discouraging. However, the prospect of forcing states to throw out the work they’ve already done, and restart the planning and approval process under a new set of guidelines, is pretty disheartening, not only for the ISPs that proposed under Missouri’s approved Preliminary Plan, but to the communities hoping to finally get internet service as well. A recent opinion piece by Missouri Representative Louis Riggs, did a good job of expressing these concerns, and frustration with the delay and the change in approach.

Missouri stakeholders—including ISPs, local governments, and bipartisan state legislators-have expressed concern about mandatory federal changes forcing states to restart or revise their BEAD plans. A letter signed by 115 state legislators from 28 states urged Secretary Lutnick to make any changes to BEAD optional rather than mandatory to avoid undermining state authority and delaying broadband deployment, but even if changes are made optional at the state level, it seems likely that Missouri officials will want to carefully consider whether to go forward with the existing grant applications, or adopt the new standards.

While nothing can be said with certainty, it seems likely that even under comparatively optimistic timeframes, Missouri now will not be able submit a Final Plan for NTIA approval until late this year, and of course NTIA approval might take many months more. If the Trump administration chooses to focus efforts on passing the “Speed for BEAD Act” it seems likely that much of what OBD did this past year will need to be redone to allow new applications and proposals for funding that allow more wireless projects, and eliminate environmental and prevailing wage requirements.

Funding Under USDA Programs Seems Uncertain.

Although funding promised to Missouri under BEAD has dwarfed other federal programs, over the past several years the Department of Agriculture (USDA) has provided funding assistance through grants and loans for many rural communities as well. Two programs in particular administered by USDA are the Reconnect and the smaller, Telecommunications Loan Program. Unfortunately things don’t look promising here either. The Trump Administration’s Director of the Office of Management and Budget (OMB) has directed that funding under both of these eliminated, because they are duplicative.

Other Broadband Funding Options for Missouri Communities

Given the uncertainty around BEAD’s timeline and rules and the probable elimination of other federal programs, Missouri communities and ISPs may want to again explore other funding avenues to continue broadband expansion efforts. Missouri law authorizes several mechanisms to finance broadband infrastructure, including issuing tax-exempt bonds for Qualified Broadband Projects, Community Improvement Districts (CIDs), Neighborhood Improvement Districts (NIDs), Tax Increment Financing (TIF), and property and sales tax exemptions for broadband projects. These are outlined in a White Paper previously published on MoBroadband. You can find the White Paper [Here].

What’s next?

Well, the short answer is we don’t know. However, it is clear that many Missourians continue to lack a connection to reliable, affordable high speed internet, and the fact that Missouri received the third highest allocation of BEAD funding (even though the state ranks 19th in population), is compelling evidence that Missouri has a much larger connectivity problem than most other states. As long as this continues, it seems likely Missouri risks underperforming our peer states in the use of internet based applications that foster economic development, telehealth, and educational opportunity.

21 Mar 2025

Demystifying ChatGPT and Other AI Large Language Models

by | | 0

March 2025

Two years ago, I sat down in front of my computer for an unusual interview—not with a person, but with an AI large language model — ChatGPT. Back then, I was curious. Could this AI really hold a conversation? Did it understand what I was asking? And most importantly, where was it getting its information? The AI answered my questions smoothly—almost too smoothly. While I was impressed, I walked away with a nagging sense of mystery. I didn’t understand how it worked—just that it did. More importantly, I largely dismissed it as a useful tool due to highly publicized examples of how it had gotten things wrong and its tendency to hallucinate. Frankly, I was skeptical. (If you’re interested, you can find the original interview here.)

The blog turned out to be popular, far more so than many others I have written over the years. Much has happened in the last two years—not just with ChatGPT, but with many other large language models –or “LLMs”. Without question, LLMs have improved significantly since my “interview” with ChatGPT . They are faster, more accurate, better at citing sources, and in some cases able to access the internet for current data. 

LLMs have also have “gone mainstream,” becoming integral to our web searches on Meta, Google, and Bing. The same technology is now embedded in tools that offer “helpful” rewrites of our emails, texts, and documents, and it is a key component of most online and telephone help desk inquiries.

But how does an LLM actually work? Why is it so good at sounding human? And what’s really going on under the hood when it generates a response?

It’s time for an updated conversation—one that does a more complete job of explaining how these AI programs work, and what are the benefits and risks of incorporating them in our daily lives  

Tokens, Parameters, and the Magic of Predicting “What Comes Next”

One of the most common misconceptions about LLMs is that they “think” like we do. They don’t. There’s no “ghost in the machine,” no digital consciousness weighing options and forming opinions. Instead, it’s more accurate to think of LLMs as sophisticated probability engines that “guess” the next word based on a mathematical model programed and modified using vast amounts of text data.

To unpack the magic of an LLM, it helps to understand two key terms: tokens and parameters. Of course, there are many more—a cacophony of technical jargon thrown around by engineers—but these two will suffice for our purposes.

Tokens

Think of a token as a chunk of text. It could be a single word (like “and,” “to,” “from”), or more likely a syllable or part of a word (like “un-“, “dis-“, “-ly”). When you type a question into an LLM, it doesn’t “see” the words in the sentence as you do. It breaks them down into tokens (words and bits of words). This is critical because, in the next step, the model must analyze token relationships using probability weightings, to generate a response.

The latest versions of LLMs can handle over 100,000 tokens in a single session. That’s impressive, especially considering that a well-educated adult has a vocabulary of about 40,000 words. This expanded token limit allows for even more complex and extended conversations, enabling models to process and generate larger pieces of text in a single interaction.

A Trillion Parameters: The “Rules” Behind the Response

However, the real magic happens when the LLM selects the next word (or series of tokens) in response to our question or request. This is done using a set of mathematical probability weightings assigned to various token combinations that comprised the question or request you typed or ask. These weightings are called parameters.

If tokens are the building blocks—the “dictionary”—then parameters are the “rulebook” that governs how an LLM constructs coherent responses. Modern LLMs have a staggering number of parameters. For example, some of the most advanced models are estimated to have nearly 1.8 trillion parameters! To put that in perspective: if you looked at each parameter for just one second, it would take over 57,000 years to examine them all! 

“Training” the LLM

These parameters aren’t something an LLM invents on its own. They are shaped by human input— and analysis of word structures from massive datasets. These datasets incorporate a large percentage of human written knowledge and literature. Human feedback and carefully designed adjustments to parameters have been added to most LLM’s to keep responses accurate and safe. Over time, the responses generated by the LLM are refined, corrected, and guided by human oversight.

These parameters enable an LLM to generate understandable responses, that are almost always correct and accurate, because relative values assigned to the words used and their order, are based on vast amounts of analyzed data. 

This is one function of the massive data centers that have emerged worldwide, where models are trained and updated on enormous datasets containing text, and in some cases, audio and video data. By examining an extensive portion of human knowledge, the LLM produces outputs that mimic human speech. Over time, as these outputs are tested, refined, and corrected and updated through reinforcement and human feedback, they become more contextually appropriate and accurate as well.

For example, the LLM determines whether the word “bank” refers to a financial institution or the side of a river based on surrounding words and user context. It can even gauge the tone of a sentence—whether a phrase is humorous, serious, or sarcastic.

Here’s an example based on a real exchange with ChatGPT:

Me: What can cows jump over?

ChatGPT: Cows are not known for their jumping abilities, but they can jump if necessary. In general:

  • Cows can jump over low fences (typically up to 4-5 feet if motivated).
  • Younger, more agile cows may jump higher than older or heavier ones.
  • If startled or frightened, cows can clear obstacles they normally wouldn’t attempt.
  • The phrase “the cow jumped over the moon” from the nursery rhyme is, of course, just a fun exaggeration.

Me: Rhyme: What do cows jump over.

ChatGPT: Cows can jump, but not too high,
A fence, a log, if they really try.
Over a ditch, across some hay,
But not the moon—no, not today!

In the first response, the LLM assumed I wanted a factual answer. In the second, my input “rhyme” triggered a different set of parameters, generating a poetic and more light-hearted response.

Limitations: When AI Gets It Wrong

Understanding how LLMs work also explains why they sometimes get things very wrong. Since they rely on predicting statistically probable next words, they can hallucinate information—confidently stating incorrect facts because they “sound right” based on training data. This is why AI-generated content should always be double-checked, especially for accuracy-sensitive topics.

A striking example of this occurred when lawyers submitted a legal brief that was generated by an LLM. The brief cited multiple legal cases that simply did not exist. The lawyers, assuming the AI’s output was correct, did not verify the cases before submitting them to the court. When the judge reviewed the citations and discovered that none of them were real, the attorneys faced serious consequences, including fines and reputational damage. 

In addition, unless an LLM has real-time access to the internet and is updated regularly, its knowledge may be outdated. The old adage “garbage in, garbage out” applies. If an LLM is trained on flawed or biased data, its responses may reflect those inaccuracies. For example, if a model is trained on outdated medical research or biased historical accounts, its responses may reinforce those inaccuracies. 

Trust But Verify

So how can you verify the accuracy and reliability of information provided by an LLM?

  • Cross-check claims against reliable sources such as academic journals, news outlets, or expert databases.
  • Ask the LLM for its sources. While it doesn’t always provide direct citations, some models can generate references or indicate the general sources of their information, offering a starting point for verification.
  • Use models that cite sources. Some LLMs, like Perplexity AI, not only restate your query to ensure proper interpretation but also provide footnoted responses with references to specific sources, allowing for greater transparency and verification of information.

Take Time to Read and Consider AI Responses

One of the best habits to develop when using an LLM, is to slow down. Humans tend to trust authoritative responses, especially when delivered with certainty. But because they are a product of statistical probabilities rather than a conscious response, AI-generated answers should be carefully evaluated against your real-world knowledge and common sense. One method that can help in engaging critically, is to ask the LLM follow-up questions, request clarifications, and challenge responses. This will not only improve accuracy but it also ensures a deeper, more complete understanding of the topic.

A Powerful Tool

LLMs are powerful tools, but they remain just that—tools. They don’t replace human reasoning, intuition, or expertise, but they can significantly enhance them. By understanding how LLMs work, we can engage with them more effectively, apply them more appropriately, and develop a balanced perspective on their strengths, limitations, and best applications. Written with the assistance of ChatGPT and Perplexity AI

30 Apr 2024

BEAD — Let’s Not Throw Away Our Shot Part 3 – Adoption

by | | 0

By: Marc McCarty and Meredith Morrison

When COVID-19 was classified as a pandemic in March 2020, much of our normal in-person activity ceased, and was replaced by a new, “online normal.” Even though millions lacked a broadband connection, without question our ability to remain connected played a critical role in maintaining our workforce, healthcare, and educational institutions in the immediate crisis. For example,

  • By May of 2020, 35% of employees worked remotely full-time through online connection.
  • Within the pandemic’s first three months, telemedicine encounters increased by 766%.
  • In 2020, 77% of public schools moved their classes to a  distance-learning format, and 84% of college students reported either some or all of their classes shifted to online-only instruction.

Since the pandemic’s onset, many things have returned to pre-pandemic normalcy, but the increased reliance on broadband remains a permanent feature of our lives.

  • Approximately 13% of employees work remotely (online) full-time, and 28% maintain a hybrid work schedule.
  • Today digital medical consultations are chosen 38 times more often than before the pandemic. In 2023, 37% of mental health visits took place virtually, followed by infectious disease, obstetrics, and transplant consultations.
  • While students and teachers continue to prefer in-person instruction at least for elementary and secondary students, online learning has clearly become a critical component for post-secondary education. Over half of degree seekers take at least some of their classes online, and over a quarter study exclusively online.

Yet full adoption of the applications that rely on broadband to better our lives continues to be a concern, even in communities with access to high-speed internet infrastructure. This blog addresses some key barriers to broadband adoption and like the two preceding blogs, suggests ways Public Organizations can help increase adoption rates in all communities, especially in those now receiving BEAD funding.

The term broadband “adoption” refers to overcoming three main barriers that prevent communities and residents from using available broadband service to their benefit.

  • Convincing skeptics that they need a fixed high-speed internet connection in their home,
  • Overcoming the privacy concerns and the fear of criminal activity on the internet, and
  • Developing the skills needed to use broadband applications effectively.

Convincing the Skeptics

Some may be surprised that in 2024, there is still a need to make the case for fixed broadband. After all, approximately 4 out of 5  households are already connected with some form of fixed broadband at home. It doesn’t seem unreasonable to conclude that just like the voice calling to Kevin Costner in the movie “Field of Dreams” – “if you build it, they will come” – or in this case if you build out the broadband network, folks will quickly subscribe for the service.  

Unfortunately, the truth is a bit more complicated. The most recent NTIA study found that in the households that still lack a fixed internet connection in the home, nearly 6 out of 10 don’t want or feel they need broadband. That percentage dwarfs those who aren’t online because they think it is too expensive (18%) or those who say they would subscribe if service was available (4%).

Overcoming Privacy and Security Concerns

Digital privacy and security concerns also deter millions of Americans from engaging in online activities. The spread of emerging technologies and practices, such as smart home devices and online activity tracking, puts these concerns at the forefront of adoption.  A 2023 public survey of 7,500 Missouri households showed that eight out of ten respondents cited security of their personal information as their top concern of internet adoption, and over half expressed concern about seeing misleading information.

Developing Digital Skills

A high speed internet connection can enable individuals to receive online treatment from their healthcare provider at home; shop for products, pay bills and bank; apply for government benefits; work from home or start an online business. However, the same technology can sow disinformation and mistrust, steal personal financial data, or create an addictive dependency on social media. Whether the internet is used for good or for evil largely depends on whether the users are equipped with the knowledge and the skills to use broadband effectively.

Adoption Is Critical to the Success of the BEAD Program

The percentage of broadband skeptics is particularly concerning because many of those currently disconnected  reside in rural locations destined to receive the bulk of Missouri’s BEAD funding. Of course, how questions are asked can impact the response, but that should not blind us to the need for a concerted effort to make the case for the use of broadband-based applications, along with the practical digital skills training necessary to navigate the internet safely and securely.

Broadband adoption also is critical from a purely economic standpoint. If those most likely to be the beneficiaries of BEAD funding don’t think they need it, or are afraid to use broadband, ISPs face the prospect of low subscription rates in areas that already have far fewer potential subscribers per mile than urban or suburban areas.

Fortunately, these concerns can be overcome. Over half of the respondents to the 2023 public survey expressed an interest in internet training assistance. “Help finding information and resources I can trust” (33%) and “assistance with setting up or using new devices” (28%) were the top two areas.

Public Organizations’ Role in Broadband Adoption

Public Organizations are uniquely situated to provide adoption programs that will address these concerns for many reasons. First, the improvement of the health, education and economic opportunity of their constituencies is the primary mission of most every Public Organization. Second, in many cases Public Organizations have already established a local connection with the community – and specifically with members of the community most at risk of being unable to use broadband-based applications to better their lives.

Finally, for local governments and related nonprofits many of the beneficial aspects of broadband-based applications lie in the ability to deliver better services more efficiently. Those who have learned to use the internet to apply for a permit, paid taxes, reported problems with utility service or otherwise interacted with local government have found the process is faster and more convenient than making an in-person visit or filling out paper forms and submitting them by mail.

In addition to being more efficient, these same technologies can also result in significant cost savings through reduced personnel and processing. However, when many of the most vulnerable lack the skills necessary to use these technologies, much of the benefit is lost, because two processing systems must be maintained, one that operates online, and a second “paper” system that accommodates those who are unable to use the new technology.

Resources for Adoption

There are literally hundreds of Public Organizations that have developed programs designed to address one or more of the three barriers to digital adoption outlined earlier in this blog. Many can be located through Mobroadband’s Missouri’s Digital Asset Map. This resource, created by the UM System at the request of OBD, is designed to allow organizations that offer digital skills to help individuals and communities quickly locate organizations that offer digital literacy programs, such as computer classes, one-on-one technical assistance centers, and bilingual resources.

Each community may have a different set of needs, and not every program will be appropriate for every community. However, each likely will face three questions.

  • What digital adoption programs does our community need most?
  • How can we integrate and coordinate our efforts with programs like BEAD that focus on internet access?
  • How do we pay for digital adoption?   

There are tools available to answer these threshold questions. One is the Digitally Connected Community Guide offered through MU Extension. The Guide leads community stakeholders to develop a shared vision of ways broadband based applications can promote better health outcomes, online learning and education opportunities and economic opportunity in their community. With that vision, stakeholders are then provided tools to approach and engage with ISPs to create a workable written plan that focuses on using new or expanded networks in ways that implement the community’s vision.

But what programs should the community use? This will vary of course depending on the community’s specific needs. Yet ideally, in every case these adoption programs should be offered by a trusted resource, and ideally they should be capable of addressing each individual’s needs through one-on-one instruction when needed.

One idea that has been successfully piloted by MU Extension in select communities is the Digital Ambassador program. This program trains a permanent cadre of local volunteers called – “Digital Ambassadors.” Trained Digital Ambassadors share specialized knowledge with individuals in their community on topics such as telehealth, cybersecurity, social media, applications like Google Docs and Office 365. Additionally, Digital Ambassadors learn practical skills necessary to help individuals set up their internet connection at home and navigate financial obstacles to broadband use, by accessing programs that subsidize internet service or provide affordable hardware. MU Extension plans to obtain funding to expand the Digital Ambassador program to all communities throughout the state.

Funding Digital Adoption Programs

Of course, Public Organizations must determine how they will pay for digital adoption programs if they are to participate with ISPs to create “digitally connected” communities. Thankfully, one new source of funding are federal grants authorized by the Digital Equity Act. Beginning sometime late summer or early this fall, OBD  is expected to begin accepting applications from Public Organizations for up to $14.2 million in grants to pay for broadband adoption programs. Additional funding, expected to total at least $10 million, should be available over the next two fiscal years under this program.

At the same time, this summer the federal government is scheduled to begin awarding up to $1.25 billion of grants through a competitive digital adoption program administered by NTIA. These programs, along with others already in place, should offer every interested community the opportunity to obtain financial support for broadband adoption programs that can complement and support efforts to build and expand broadband infrastructure over the next five years.

Let’s Not Throw Away Our Shot

The common theme running across this series of blogs is that now is the time for action. There has never been this level of financial support at the federal government level for broadband, and there is every reason to believe this investment will not be repeated, at least in our generation. Missouri has received a significantly greater proportion of this federal funding than other states, primarily because we have a greater need. The percentage of unserved and underserved locations in our state has been among the highest in the country, and our economic progress and our quality of life have suffered because of it.

We now have an opportunity to change that. This is our shot – what will we do with it?

7 Mar 2024

BEAD — Let’s Not Throw Away Our Shot Part 2 –Affordability

by | | 0

By Marc McCarty

& Meredith Morrison

The last Blog described the Missouri Office of Broadband Development’s (OBD) proposal to spend up to $1.7 billion for broadband projects that will be funded by the federal government’s Broadband Equity Access Deployment (BEAD) program. We highlighted ways local government and nonprofit organizations (Public Organizations) can play an important role at the outset, by helping to make the list of locations eligible for BEAD funding as accurate as possible, and by initiating contact with internet service providers (ISPs) to express the community’s interest in partnering to “digitally connect” their community.

Three ingredients are essential for communities to digitally connect while holistically benefitting the health, education, and economic advancement of its residents: (1) access to internet service, (2) the skills necessary for service adoption, and (3) a pricing model that makes service affordable. BEAD only directly addresses internet access. Internet adoption and affordability will require Public Organizations to organize and enter into partnerships with existing and new ISPs.

This blog focuses on the challenge to make broadband service affordable; a challenge that has been made much more difficult because of the end of funding for the Affordable Connectivity Program (ACP).

Why Affordability Matters

Obviously, community residents and businesses cannot reap the benefits of broadband service if they can’t afford it! Like other utilities, broadband service comes with a monthly fee, and all ISPs generally must earn enough in monthly subscription revenue to cover the costs of capital investment, network maintenance, and service operations. In Missouri, most ISPs are “for-profit” companies, in which shareholders/investors must earn a reasonable return on their investments. For this reason, particularly, in hard to reach communities where there are fewer locations per square mile, obtaining a high percentage of paying subscribers can be critical to an ISP’s survival.

Most Missourians are understandably unwilling or unable to pay for high-cost broadband service. Multiple surveys show that a high monthly internet subscription cost is one of the biggest roadblocks to the adoption of broadband. Of course, in part the affordability of broadband service depends on whether potential subscribers feel confident they need and can use fixed broadband, or instead, can continue to rely on their internet-connected mobile phone or public hotspots. The next blog will cover how Public Organizations can address this issue by boosting internet adoption.

Many Missourians face a “would if I could” problem when monthly internet subscription costs compete with basic necessities, such as food, housing, water, and other utilities. Ironically, these households most desperately need broadband service to effectively access government and nonprofit assistance and improve their economic prospects for a better life. Clearly, some targeted program to make broadband affordable to low-income households is needed.

The Affordable Connectivity Program

The Affordable Connectivity Program was enacted along with BEAD as part of the 2021 infrastructure law to address this specific concern. In the two years it has been in place, 23 million households nationwide (nearly 400,000 in Missouri alone) received a $30 per month subsidy for internet service and a one-time discount they could apply to the purchase of a basic computer or tablet. Roughly 1 in 6  households across our state benefited from ACP, and in many rural counties 1 out of every 4 or 5 internet subscribers, received service subsidized by ACP.

The ACP became a victim of its own success. As structured, 42 million households were estimated to be eligible for the ACP benefit. Even though just a little more than half those families eventually applied and began participating in the ACP, the original $14 billion appropriation for the program is almost entirely depleted. The FCC has barred any new subscribers, and the program will end next month. While bipartisan legislation has been introduced to fund the ACP until year end, it is not expected to be enacted, and of course it is only a stopgap measure. In order to truly meet the needs of families, and provide a stable subscriber income base for ISPs, the program needs a permanent funding source, and this need is particularly evident as ISPs attempt to build out broadband in hard to reach unserved and underserved areas using the BEAD program.

Certainly absent a viable alternative, many participating households served by the ACP may be forced to drop service. While the federal government’s Lifeline program can offer some current ACP subscribers discounted broadband at connection speeds up to 25/3 Mbps, that is below the 100/20 Mbps mandated by the BEAD and ARPA programs. Additionally, the qualifying income levels for Lifeline are lower, and other eligibility requirements are more difficult to meet than those imposed by the ACP. Finally, many ISPs are not qualified to offer customers the Lifeline program benefit.

The Subscriber Revenue Gap Created by the End of ACP

Of course, the end of the ACP also creates some significant financial challenges for ISPs. ACP gave ISPs an incentive to build infrastructure in low-income communities, because the program increased their estimated take rate and subscriber revenues.  One study estimated that the loss of the ACP will translate into a decline in subscribers that will raise the cost of bringing broadband to unserved rural communities by 25%!

It is also unclear how ACP’s termination affects the sixty ISPs that were awarded grants as part of Missouri’s  $261 million Broadband Infrastructure Grant Program. These grant came from Missouri’s share of the American Recovery and Reinvestment Act — Coronavirus State and Local Fiscal Recovery Fund (SLFRF). Federal Regulations for the SLFRF program require that grant recipients offer low-income households affordable subscription options, commensurate with ACP’s benefits in areas served by grant-funded broadband infrastructure. While this requirement may soften the blow from the loss of the ACP in these isolated communities, the question remains, how will ISPs afford to implement a “commensurate” ACP benefit on their own?

Taken together, the loss of the ACP seems destined not only to make it harder for lower income households to subscribe for internet service, it also may make it significantly harder for Missouri to achieve the overall goals of the BEAD program – to extend broadband service to all unserved and to as many underserved areas of the state as possible.

How can Public Organizations Help?

In 2022, Governor Parson proposed using state funds to supplement broadband subscription costs for low income households.  However, that legislation ultimately was not enacted, and other than the Lifeline program previously discussed, there is no comprehensive federal or state program that directly subsidizes the monthly cost of broadband for lower income families. However, that does not mean that Public Organizations – particularly local government – are unable to indirectly help ISPs provide affordable broadband service to households that need financial assistance.

The key here is to recognize that ISPs can provide an ACP-type benefit to their customers so long as that cost is subsidized and that subsidy can take the form of either a direct payment to help the customer pay for service or, indirectly, by further reducing the ISP’s cost of building the broadband infrastructure. A BEAD grant can do that in part, but likely not if the ISP’s subsidy needs to be increased by 25%, because the ACP is not available! Yet it makes no sense to increase the government subsidy to the ISP for construction unless the ISP is willing to use the amount saved to provide an affordable broadband connection to its customers, through a program similar to the ACP.

For example, assume that an ISP was able to build and successfully operate a broadband network with a BEAD grant of $2 million, so long as it could rely on ACP-funded subscriber revenues to help pay for the infrastructure and cost of network operations. Based on the Commonsense study, one would conclude that the ISP would need 25% more ($2.5 million) because of the loss of the ACP. In that case Public Organizations might agree to provide the ISP the additional money needed to close the funding gap ($500,000), but only on the condition that the ISP agrees to self-fund an ACP-type benefit for the community in the future.

Unspent Local ARPA Funds

One possible source of the additional money needed, that some local governments may have available is the local government component of ARPA SLFRF money. In addition to money paid directly to the state, Missouri’s 114 Missouri counties and 15 cities received a separate SLFRF award. These local governments have the flexibility to use these funds for many purposes, one of which is “high-speed broadband infrastructure.” All local SLFRF money must be obligated by December 31, 2024 and spent by December 31, 2026. As of September 2023, Missouri’s local governments reported that they had obligated only $1,450,997,654 of the $2,419,661,436 in SLFRF Funds. This means that as of last Fall, just over 40 percent of the money awarded remained available.

Other possible local programs to access local money.

Many communities may already have appropriated their entire ARRA SLFRF award for other permitted purposes. In these cases, there are other options that can be used to provide financial support to ISPs that agree to provide affordable broadband service to lower income households. Again, the overall strategy is the same: use locally generated funds to reduce the ISP’s cost installing broadband infrastructure in unserved and underserved locations, in exchange for the ISP’s commitment to provide a targeted low-cost subsidy to qualified low income subscribers.

These options include special financing districts and government sponsored tax-exempt financing.

Special District Financing

Three  types of special financing districts are available to Public Organizations seeking to raise money to support an ISPs broadband infrastructure project. In each case, these special financing districts must support and fund a “public private partnership” (P3) that will provide broadband service within a specific geographic area or “District.” Each type of District may impose various new local taxes or special assessments for that purpose. The three types of districts are (1) a Community Improvement District (CID), (2) a Neighborhood Improvement District (NID) and a Broadband Infrastructure Improvement District (BIID).

The statutes set out the procedure and powers for each of these districts and while they are not identical, they do share some common elements:

  • The area served by the District can be specifically tailored to include all or a portion of a city or county, or in the case of CIDs and BIIDs multiple jurisdictions.
  • The District may fund a public private partnership that will bring broadband service to one or more unserved or underserved areas (as certified OBD). The statute defines unserved and underserved using similar criteria to that used for BEAD grant funding.
  • The District can impose a variety taxes (sales tax, property taxes, or a special property assessment to fund its contribution the P3). However in most cases, voter approval of the affected district residents is required to impose the tax.
  • In each case, the P3 agreement must provide that the ISP will own and operate the broadband network infrastructure (not the District).

A significant advantage of Missouri’s special financing district laws is that they are flexible and can be adapted to a variety of situations. Since Districts can impose local taxes and/or assessments that only apply in the District, it is possible to tailored them to include only those areas that need broadband. In this way they could be an ideal source of locally sourced funds to help ISPs close a funding gap and make it economically feasible to provide broadband service to residents of all income levels. This is important both because of the gap created by the loss of the ACP funding and, more generally because all BEAD – funded projects must have at least 25% of the project costs funded from some source other than the BEAD grant.

 

Example of a Special District Financing to Support Broadband Affordability

A community might decide to create a CID to assist in funding broadband infrastructure for 20 unserved farms and homes along a county road. Property owners along that road could petition the County to form and authorize funding for the CID. The District’s objective would be to lower the cost of installing broadband so that the ISP was financially able to provide free or low-cost service to low income families with school age children.

To fund the CID, property owners in the area would agree to impose an annual special assessment (collected along with real property taxes each year) from each property owner in the District. As part of the P3 agreement, the CID would agree that amounts collected each year would reimburse it for part of its cost of funding the broadband project that was not paid for out of the BEAD grant, and that the collected amounts would be applied to fund a monthly broadband service subscription subsidy for lower income families.

Tax-exempt Private Activity Bond Financing

Tax-exempt private activity bond financing may provide another option for Public Organizations that want to reduce an ISPs infrastructure cost, so that a low cost broadband subscription option can be offered to lower income residents. Tax-exempt private activity bonds are a special type of local government debt. Unlike traditional bonds where the local government is the named borrower and responsible for repaying the debt, private activity bonds are usually repaid only from money provided by a private entity (in this case an ISP).

These tax-exempt bonds can only be used in a specific geographic area where 50% or more of the structures are “unserved” (broadband service less than 25 Mbps download and 3 Mbps upload). When completed, the constructed project must provide at least 90% of the previously unserved locations with connection speeds of at least 100 Mbps download and 20 Mbps upload. If these and other requirements are met, interest paid to investors (bond purchasers) is exempt from most federal and all Missouri income tax. Because investors do not pay income tax on the bonds they own, they are willing to accept a lower interest rate than a similar taxable bond or loan. In this way, the ISP benefits because it pays substantially less interest each year on the debt used to finance the project.

 

Example of Tax Exempt Private Activity Bond to Support Broadband Affordability

 

Assume an ISP wished to bring broadband service to all structures in a county, and that the cost of constructing the network was $50 million. The ISP receives a BEAD grant for $35 million, but is responsible for raising the additional $15 million. The County’s residents want the service, but know that a significant number of residents simply cannot afford to pay the monthly subscription cost for broadband service. For a variety of reasons, the county cannot provide financial assistance to the ISP through a new tax or assessment.

Instead, the County might issue $15 million of private activity bonds that the ISP would use to fund the project. The ISP (not the County or any Public Organization) will be solely responsible for repaying the bonds.  The bonds have an interest rate that is 1.5% lower than a taxable borrowing. This means the ISP “saves” $225,000 of interest cost each year the bonds remain outstanding.

As a condition for issuing the bonds, the County requires the ISP to offer lower income subscribers a special low cost broadband plan, which it is able to fund because of the interest expense saved each year. 

Where There’s a Will….

In a perfect world, funding for broadband affordability would not have run out just as funding for broadband access and adoption was becoming available. Of course no one would say that the ACP was a perfect solution. Many would argue that the ACP and Lifeline programs need to be combined and a permanent funding model needs to be put in its place, and there is considerable merit to that position. But allowing ACP to expire without reworking the Lifeline program or providing some other workable alternative is difficult to justify. The unfortunate result for many communities is that they must work creatively with their ISPs to make sure that the broadband infrastructure built is affordable to all members of the community. The alternatives listed here all seem to be feasible, and there are many others (described in a recent “White Paper”) that communities can consider. However, all require a community effort, and some creative thinking. The question for many communities – is whether they are willing to make this effort, or will they instead risk “throwing away their shot” at digital connectivity.

8 Feb 2024

Broadband for All — Let’s Not “Throw Away Our Shot!”

by | | 0

Part 1 – Why the Next Few Months Are Critical

“I am not throwin’ away my shot!” This line, from the musical Hamilton keeps playing in my head as I think about our “once in a generation” opportunity to finally provide everyone in our state high speed internet – broadband.

This is the first of several blogs designed to alert and call on Public Organizations to help make universal access to broadband a reality.  This is particularly true now, as $1.7 billion of federal government funding is about to become available through the Broadband Equity Access and Deployment Act (BEAD) program.

“Public Organizations” include local government, such as county commissions, city councils and boards of aldermen, and school district boards; as well as their partners, chambers of commerce; regional economic development planning commissions and regional councils of government; and civic nonprofits, such as Rotary and Lions Clubs.

While most of the money for broadband infrastructure will be directed to private for-profit internet service providers (ISPs), Local Public Organizations can help ISPs create economically sustainable infrastructure and help communities use this new asset to improve the health, education and economic opportunity, for all residents.

Why Now? Haven’t We Closed the Digital Divide?

As the COVID pandemic set in during 2020, skepticism over whether there was a need for broadband quickly gave way to an understanding that, like electricity and running water, a stable high-speed internet connection provided an essential “utility” for businesses and families. This led to a dramatic increase in promised state and federal  government funding for broadband infrastructure. While some of that government funding has been awarded to ISPs much, much more is only now becoming available.

In late 2021, Congress passed the bipartization “Infrastructure Investment and Jobs Act” (IIJA). That Act authorized over $60 billion of federal funding for broadband access, adoption, and affordability programs. $42.5 billions of that amount is dedicated to broadband access through the Broadband Equity Access and Deployment Act (BEAD) grant program. These laws (and others) have been summarized in several blogs already [here], [here] and [here] When combined with continued support provided by the Federal Communications Commission’s (FCC) RDOF Program and the USDA’s Reconnect Program, the federal government’s promised investment in broadband over the next decade  totals more than $100 billion!

So, “problem solved!” you might think, and certainly this federal funding could go a long way toward providing reliable broadband access to all unserved and underserved locations in Missouri and across the United States. Yet, there is still much more to be done. For example, the nearly $200 million awarded by the State of Missouri during the fiscal year ending June 2023, will provide access to only 8% of Missouri locations that need service.

The largest single federal program enacted in 2021 — BEAD — promises Missouri $1.7 billion of grants to provide reliable and affordable broadband service to every unserved location, and potentially every underserved location in Missouri. However after more than two years, only a small fraction of BEAD funding has been distributed to states.

Although slow, progress, is being made. At the end of last year Missouri’s Office of Broadband Development (OBD) submitted its “Initial Proposal” to the National Telecommunication and Information Agency (NTIA) for approval. The Initial Proposal includes OBD’s processes for identifying projects and distributing BEAD funds. OBD hopes to receive approval of the Initial Proposal by February 18th. Assuming this happens, OBD expects to begin accepting requests for grants this summer and possibly award the first grants this fall.

The timeline depends on NTIA’s approval of the Initial Proposal, and certainly, that timing could “slip.” It is also possible that NTIA may modify some of the details of the Missouri proposal, but the fact remains that after years of waiting, the next few weeks and months likely will determine what locations will receive a share of the BEAD money, and what broadband infrastructure technology will be built over the next five years.

In other words, for communities across the state that lack adequate broadband access, now is the time to become engaged in efforts to finally bridge the digital divide.

What to Expect

The state’s Initial Proposal calls for action to commence immediately following NTIA approval. This will involve two preliminary steps – (1) finalize the locations eligible for BEAD funding and (2) prequalify ISPs that will be eligible to participate in the BEAD proposal and funding process. Following the first two steps, OBD will begin accepting specific proposals for the first of at least two funding rounds. Proposals for round one will ideally be accepted between June 1 – July 31, 2024, with the first awards made by October 2024.

The first round of funding will be designed to favor proposals that connect locations using the fastest and most robust broadband infrastructure — fiber optic cable to the premises (FTTP). The second round of funding will target locations not funded in the first round because they are harder to reach and less economically viable. Other forms of reliable infrastructure, such as cable and wireless connections, likely will be used more for this round of funding. Both funding rounds will be numerically scored, and in each case the amount of BEAD grant funds requested will be a very important, but certainly not the only, factor in determining whether a proposal will be funded.

Identify BEAD Eligible Locations

When NTIA approves the Initial Proposal, OBD will publish its proposed set of eligible BEAD funding locations. Generally funding is available for unserved and underserved locations, determined by the download and upload data speeds, with minimum requirements for latency (the time it takes for a signal to be transmitted and received). This eliminates satellite internet providers. In addition, locations served only by DSL are considered “underserved” regardless of the stated connection speed. Finally, locations already awarded funding under most federal and state programs or otherwise subject to a “binding commitment” to provide broadband service within the next 12 months will not be eligible for BEAD funding.

Last month OBD released its preliminary map of unserved and underserved locations in Missouri. This map, based on the FCC’s FABRIC, is the most comprehensive effort yet to accurately identify locations (homes, businesses and institutions) that should have access to broadband service, locations where ISPs claim to be providing service, and finally the advertised speed of the connection at the location. The preliminary map will be modified by OBD, based on criteria set out in its Initial Proposal (as previously described).

ISPs and Public Organizations will then have at least 45 days to “challenge” the status of a particular location or group of locations based on one of several criteria. These challenges can result in locations being found ineligible for BEAD funding or be reclassified as unserved or underserved. While it is hoped that most of the inaccuracies that plagued earlier FCC maps have been corrected, there likely still will be errors, and this makes participation in the process very important to the overall success of the BEAD program.

ISP Prequalification

Experience has taught OBD the importance of carefully vetting ISPs through a prequalification process before allowing them to submit proposals for BEAD funding. The FCC’s Rural Digital Opportunity Fund (RDOF) Program illustrated the folly of failing to adequately vet participants prior to the competitive bidding process. Many months after preliminary grant recipients were publicly announced, the FCC disqualified two of the largest “winners” of the competitive bidding process because they lacked the technology and/or the financial wherewithal to actually complete the promised projects.

To avoid repeating this result, OBD’s Initial Proposal requires ISPs to prequalify before they can participate in the BEAD funding program. Assuming the Initial Proposal is approved as expected, the Prequalification Process for the first round of BEAD funding is expected to begin on April 17 and close by May 31.

The BEAD Award Process

After establishing eligible funding locations and prequalifying ISPs, OBD will proceed to “round one” of the BEAD awards. During the month of April, OBD will amend its maps to create areas that will be eligible for BEAD funding. These areas will be selected by OBD based on factors such as the location of nearby qualified ISPs and the feasibility of connecting unserved and underserved locations to existing networks.

Approved ISPs will then be able to submit proposals for BEAD funding for any one or combination of these areas, but they must agree to serve every location in the area or areas they have selected. Further, in this first funding round only, proposals that offer FTTP for all locations, will be awarded BEAD funds even if other technologies could provide service at a lower cost, so long as the amount of BEAD funds requested does not exceed a maximum BEAD outlay amount established by OBD.

In the case of multiple proposals for a given area, awards will be based on a numeric scoring matrix, that takes many factors into account. However, three of those factors seem likely to be most relevant. First, the amount of BEAD funds requested (the BEAD Outlay) likely will be very important in determining which ISP receives a BEAD grant. The scoring matrix will favor proposals that request a lower BEAD Outlay. Second, while the BEAD Outlay is important, proposals that offer FTTP rather than other alternate technologies and proposals with shorter deployment times will receive more points. Finally, a substantial number of points will be awarded for proposals that have public support, and this is especially true for proposals where Public Organizations are providing some level of financial support.

What is the Strategy for Public Organizations?

This is a multipart blog, and future articles will focus on specific ways Public Organizations can help make the BEAD program a success in their communities. However, it’s clear Public Organizations can take steps now to set the stage. OBD’s preliminary map enables every community to see each home, business and structure that needs a broadband connection and to identify the ISPs that claim to be able to provide service to that location – and the maximum level of service they can provide.

Local Public Organizations are uniquely able to verify that data because they are a part of the local community. This is especially important because individuals cannot challenge the FCC data or OBD’s map directly. This means that Public Organizations will need to play a critical role in developing challenges to the map when that is necessary. Equally important, local Public Organizations can serve a vital role in assessing, and if necessary, rebutting challenges to the map that are raised by ISPs in the area.

In addition, the data in the preliminary map gives Public Organizations a list of the providers serving each location – both in the community and in surrounding areas. In all likelihood, these ISPs will be making the business decision regarding whether they will participate in BEAD funding, and if so, the specific locations where they will extend service, and the type of technology (fiber, cable or fixed wireless) they will use.

These decisions will not be made in a vacuum. Among other considerations, ISPs will assess the ease of obtaining permits, right of way, and easements to construct the improvements, and the level of customer subscriptions they can expect in the new service area. As will become evident in later blogs, Public Organizations can do much to help and encourage ISPs as the grapple with these issues. However, the first step (and perhaps the most important one), is for Public Organizations in the community to reach out to each local ISP to let them know they are interested in participating and assisting them in the process of developing a BEAD proposal.

*** BEAD offers communities across the state an historic opportunity to finally become connected to the internet in ways that will improve the health, education and economic opportunity for everyone. For our state, this truly is “our shot” .. and we must not throw it away.

16 Jan 2024

The Affordable Connectivity Program May “Go Away” Just When it is Needed Most

by | posted in: | 0

By Marc McCarty

They say timing is everything, and that certainly likely will be the case as we move forward this year to implement major components of the 2021 Federal Infrastructure legislation (the Infrastructure Act). You may recall that the Infrastructure Act appropriated $65 billion with the objective of providing every residence, business and institution in the United States a high-speed internet connection – broadband, and the skills to use it. The Act has three goals: building out the infrastructure needed to connect unserved and underserved locations (broadband “access”), getting individuals the skills-based training and resources they need to use a broadband connection effectively (“adoption”), and finally, making broadband affordable for households that lack the financial resources to subscribe for the broadband connection they need (“affordability”).

The rationale for this “three-prong” approach is logical. It makes little sense to build out a broadband network in unconnected communities if most of the targeted individuals are afraid to go online or lack the skills needed to use the internet applications that would help them the most. These skills, include using the internet to start a business, pay bills and bank online, connect with a health care provider, or get an advanced education certification or degree. Equally obvious, having a fiber optic broadband connection at your home is of no value if you can’t afford to subscribe for service, or if you can’t afford a basic device to connect to the internet efficiently.

With this in mind, an earlier Blog noted that the Infrastructure Act had allocated at least $46 billion of grants and low interest loans for broadband access, up to least $4.75 billion for broadband adoption programs, and $14.2 billion for broadband affordability (the Affordable Connectivity Program or “ACP”).

The ACP was designed to permanently replace the Emergency Broadband Benefit (“ECB”) a similar temporary program enacted during the COVID pandemic. Like the ECB, ACP is targeted to help families that lack financial resources pay for the internet service they need to use the internet effectively. While there are several ways to qualify for ACP, generally families earning less than twice the annual poverty income ($60,000 for a family of 4) are eligible for the ACP.

The ACP provides these households a $30 per month credit that can be applied to monthly cost of internet service and a one-time $100 credit toward the cost of a desktop, laptop or tablet computer to connect to the internet. The program began funding in 2022 and as of last week, over 22.5 million households were receiving benefits. One advantage of the ACP is that the “credit” can be applied by families toward any level of broadband service offered, so even if a household was able to pay for some internet connectivity before, the ACP enabled them to upgrade to a higher more expensive level of service, so they can take advantage of applications such as telemedicine and online learning that require a faster and more stable internet connection.

Many Missouri families now use this benefit. In several rural Missouri counties more than one in five households that are connected to the internet are receiving ACP. The truth is ACP has been so successful, that it is about to run out of money. The FCC administers this program, and it has already instructed internet service providers (“ISPs”) to send out the first of three written notices beginning January 25, to customers warning them the benefit will expire (likely sometime in May).

The ACP has wide support among participants, internet providers and the general public. Last week bipartisan legislation was introduced to extend funding for the ACP through the end of 2024. Certainly given the political environment, that may be the best we can do at this time, and even in this case passage of this legislation likely will not occur unless constituents make their wishes known.

Ironically, this comes at the very time when almost all the Infrastructure Act money set aside for broadband access (BEAD) and broadband adoption (DEA) remains unspent! The delay in funding BEAD and DEA occurred for many reasons, some of which I’ve written about already. However, Missouri’s Office of Broadband Development (OBD) now awaits approval of its “Initial Proposal” to distribute the first 20% of the $1.7 Billion dollars of BEAD funding along with a smaller DEA “State Capacity Grant” to fund broadband adoption. Those approvals (granted by the National Telecommunications and Information Agency –NTIA) are expected as early as late spring, followed by competitive grants that could begin funding new projects by year-end.

For a variety of reasons, the State’s BEAD proposal primarily contemplates using existing private and public ISPs to extend broadband service to the 400,000 locations in the state with no – or with inadequate – broadband access. Those ISPs provide access to the internet (fund capital expansion, maintain, and operate) primarily through subscriber revenues paid by businesses and residents. Locations needing broadband access lack it in most cases because the ISPs that potentially could serve them cannot make an adequate profit to justify the investment. The objective of the BEAD program is to use just enough public money to induce ISPs to expand service to these unserved and underserved locations.

For example, if the average cost of extending service to a group of locations was $5,000 per location, but an ISP could only be profitable if installation costs were no more than $1,500 per location, an efficient BEAD grant program would provide the ISP a grant of $3,500 per location, conditioned on the ISP going forward to provide broadband access to all of these unserved locations. But that example assumes that families and businesses in those locations actually will subscribe for the service (at levels as high as – or higher— than other areas where service is currently available). In other words – ISPs don’t need or necessarily want more locations with access to the internet – instead they want more locations with internet subscribers.

That is one practical reason why both the DEA, and the ACP exist. From a purely economic standpoint, both of these programs are designed to work alongside BEAD to increase demand for broadband service (the “take rate” as it’s known in the industry). The DEA gives folks the skills needed to appreciate and safely use internet based applications and technologies; the ACP makes that service affordable, so families to pay for the service they need.

Taken together one can think of these broadband programs BEAD, the DEA and the ACP as something like a three legged stool. The three legs of that stool provide funding for broadband access, adoption and affordability. Remove one of those legs, affordability in this case, and the stool – a $65 billion stool – may well topple over.

Of course there are many good reasons for providing financial assistance to families that can’t afford the broadband service they need besides creating a broadband network that is financially viable. Connecting most businesses and individuals to broadband has led to extraordinary gains in productivity and quality of life, but as the COVID pandemic illustrated, those gains have been uneven, with millions lacking adequate internet service and unable to use these new internet-based applications. We all pay when we leave the most economically vulnerable families disconnected to the internet applications we take for granted, in higher costs for healthcare, basic government services, substandard education, and fewer economic opportunities. A program like the ACP can help remove financial barriers to internet connectivity, and certainly it can be justified for that reason alone. However, allowing a program like the ACP to go away after all the work done over the past two years, and just as we are ready to spend over $45 billion on broadband expansion, is most certainly a mistake.

10 Jul 2023

Critical Thinking and the Internet: Developing Skills to Counter Online Disinformation & Confirmation Bias Algorithms

by | posted in: | 0

It’s impossible to overstate how much the internet has changed our lives over the past three decades. Internet-based technologies and products have unleashed exponential economic growth and efficiencies. It is no accident that the five largest companies in the S&P 500 are Apple, Microsoft, Amazon, NVDIA and Alphabet (Google). Certainly the internet has been, and will continue to be, a driver of economic growth and internet-based innovations that promise to continue to improve the health, education and economic opportunities for all of us. However, as has been true with all new technological innovations, there is a dark side to the internet; challenges that if ignored, will substantially reduce the benefits expected to be realized from universal access to broadband and its applications. These challenges need to be understood and addressed, particularly as we work to connect the remainder of homes and businesses in the United States to broadband by 2030.

One of these challenges was described earlier this year in a three-part blog on cybersecurity. More recently, I explored some of the potential questions and challenges associated with generative artificial intelligence. This blog discusses yet another challenge:  How do we spot internet disinformation and counter internet-based algorithms that tend to confirm our preexisting biases, and blind us to opposing viewpoints?

For representative democracies like the United States, particularly now when most of us rely on the internet to get our news and form our opinions, the ability to analyze and test the accuracy of sources of  information on the web (to engage in critical thinking) has and will continue to be a vital skill to effectively use the internet. While there is no “magic bullet” solution, ignoring this issue risks more than just continued economic progress, it could threaten the very institutions that sparked the creation of the internet itself.    

Of course the goal of this blog (and the others that preceded it) is not to discourage the development of internet infrastructure and internet-based applications. But the degree to which our goals of better health, education and economic opportunity will be realized, depends in large part on how well we adapt to use these new internet-based applications and technologies effectively. Developing these skills is an important part of the University’s Broadband Initiative, and programs such as the pilot Digital Ambassadors project are expected to be an important part of that effort.

Defining the Challenge

At the outset, it’s worthwhile to spend time defining the challenge. The term “disinformation” is closely related to its companion – misinformation. Misinformation is simply inaccurate or false information. Disinformation on the other hand is misinformation put to a purpose. It’s the use of misinformation in a way specifically designed to deceive or hide the facts.

The motives that lead a person (or more recently an artificial intelligence algorithm) to place disinformation on the web, most often are not perceived to be morally wrong by the person responsible for posting. In fact, in many cases the opposite is true; disinformation is used to serve what is perceived as a higher purpose or objective. In other words, the means (an intentionally false or misleading story or headline) are justified by the belief that it supports a view or position that is in the best interest of society.

Unlike disinformation, “confirmation bias,” requires no intentional act. Instead it describes our unconscious tendency to seek out and treat information as true if it supports our biases and predispositions. We all use confirmation bias to make decisions in our daily lives, and it is not necessarily a bad thing. For example, most of us have a “confirmation bias” that would make us hesitant to climb into an enclosure to get a better view of a grizzly bear at the zoo. It might well be that the particular bear was trained and well-behaved, but we know from books or film that these animals often can be dangerous, and we run the risk of being “lunch” if we get too close. Research indicates that internet-based algorithms make extensive use of our tendency for confirmation bias in ways that we may not fully understand or appreciate, usually with the goal of keeping us engaged and online.

Disinformation – How Does the Internet Differ from Earlier Forms of Mass Communication?

History contains many examples of individuals that used mass media to disseminate disinformation. In addition to charlatans, disinformation has been used by some that we hold in high regard. In fact, none other than Benjamin Franklin apparently is guilty. In order to stir up revolutionary fever, he apparently made up a story accusing King George III of promoting attacks on colonists — offering a cash bounty for each colonist scalp that was collected! Disinformation in mass media also has always been difficult to stop. This is particularly true in societies like ours that value freedom of expression. Early attempts by our government to rein in false or “fake” news, even if motivated by a noble purpose have been unpopular and ineffective. 

So it’s fair to point out that  disinformation in public media (whether it’s a printed pamphlet, a newspaper, television or radio) is nothing new. However, there are several unique aspects that make disinformation on the internet more challenging to identify and counter. The same characteristics that make the internet such an effective tool for learning and disseminating information, also have made it much more effective in spreading disinformation. One reason the internet spreads disinformation so effectively is that multimedia (videos and audio) can be used along with text  to get ideas across. It is not surprising then, that researchers have found that most disinformation on websites today consists of images and videos.   A second reason is that the internet permits information and disinformation to be shared far more easily and quickly than earlier technologies. In the past few could afford a printing press, and more recently television and radio stations could make use of audio and visual images, only after obtaining a license from the Federal Communications Commission.  Today, anyone with an online connection can create and share text and full resolution video content with millions in just a few hours, and do so anonymously.  

The risks posed by disinformation on the web seem destined to grow. For example, software has been developed that permits most anyone to create near perfect video imitations of public figures that can say anything the programmer desires. The age-old adage “seeing is believing” seems destined to become a quaint anachronism.

Internet Algorithms & Confirmation Bias – “There’s No Such Thing as a Free Lunch”

Most of us don’t reflect on why there is so much information available to us on the web “free of charge.” Of course, in some cases government, nonprofits and public-spirited individuals have provided content as a public service, and there are many subscription fee-based websites, but that does not explain the millions of commercial websites that provide news, entertainment, and personal connections free of charge.

These websites have a profit motive, and most exist to sell advertising.  Many of these ads are structured as a “cost per click” arrangement – meaning that advertisers pay the website owner a set amount each time someone clicks on a hyperlink that directs the web browser to the advertiser’s content. Again, intellectually, many of us realize this is happening, but we may not fully appreciate just how significant this revenue has become. In 2022, ads of this type were estimated to generate $95.2 billion. To put this in perspective, that’s over $275 of cost per click revenue for every man, woman, and child in the United States.    

With this much money at stake, it is not surprising that a primary goal of many commercial websites – particularly social media websites, is to keep us online and engaged with the website’s content for as long as possible.  Like print media advertisers that came before, today’s web designers know they can do this most effectively by keeping us emotionally engaged. Again, the ultimate goal of these efforts most often is to increase ad revenue. After all, the longer you are online and looking at website content, the more ads you will see, and potentially the greater the chance that you will click on at least one, and earn revenue for the website’s owners. There’s a name for these targeted efforts to keep us on a website, it’s call “clickbait,” and anyone who suddenly realized they are late for a meeting because they have spent the past 30 minutes looking at “50 cute kitten” videos knows clickbait can be very effective.

Applying Critical Thinking Skills to the Internet

We tend to lose sight of how quickly the internet has become a part of our lives. Thirty years ago blogs like this one did not exist. Google was founded in 1998. The first Facebook page was created by Mark Zuckerberg less than 20 years ago, and Jack Dorsey posted the first “tweet” on Twitter in 2006. Given the speed of these developments, it’s not all that surprising that critical thinking skills training may not adequately address web-based disinformation or the impact of web-based search algorithms on our confirmation bias.

Certainly there is nothing new in the idea that critical thinking skills are essential to the sound functioning of democratic institutions. Skilled critical thinkers are able to:

  • Raise vital questions and problems, formulating them clearly and precisely.
  • Gather and assess relevant information, using abstract ideas to interpret it effectively.
  • Come to well-reasoned conclusions and solutions, testing them against relevant criteria and standards.
  • Think open-mindedly within alternative systems of thought, recognizing and assessing, as needs be, their assumptions, implications, and practical consequences.
  • Communicate effectively with others in figuring out solutions to complex problems.

Even if critical thinking is not taught as a stand-alone subject in elementary and secondary schools, most children do receive instruction in basic critical thinking skills as part of their education, although the amount of training varies and surprisingly, may decline once the student enters middle school and high school. Most colleges and universities, including several of the University of Missouri System campus libraries and schools, provide critical thinking skills training designed to help students when they do internet research.

Yet a comprehensive 2019 study found that students are not well equipped in ferreting out disinformation on the internet. This also is not all that surprising. There may be other reasons for this, but one suggested by the study and other related research is that critical thinking skills taught to identify disinformation on the web may not be particularly effective today.  Those methods included assigning credibility to information contained on “.org” websites and discounting those that had a “.com” designation; relying on the website’s statement located on its “about” webpage to understand its mission; assigning more value to content on websites that have professional looking “error free” layouts; and giving credence to web-based materials that includes footnotes or hyperlinks referencing journals that are not generally well-known, but that have professional-sounding names. While all of these may seem reasonable or have direct corollaries to fact checking traditional printed text material, the study discounted their value in discovering disinformation on the web.

For example the fact that a website has a “.org” label, does not mean that it is sponsored by a nonbiased nonprofit organization; instead, it simply is an alternate catch-all designation available for any website  that does not wish to be classified as having a commercial (.com), government (.gov), or educational institution (.edu) sponsor. Nor is the website’s “about” page or its professional design particularly helpful in ferreting out disinformation. After all, the text of the “about” page was written by the same folks who wrote the content that is being checked, and modern website design programs enable most anyone to prepare a very professional looking website.

Distinguishing disinformation simply based on how the content appears likely will be even more difficult in the future because of the development and widespread availability of new programs using sophisticated graphic design and artificial intelligence. This was illustrated just this past month, when several attorneys were sanctioned and fined $5,000 for filing a legal brief, authored by an generative artificial intelligence program. The problem wasn’t that the attorneys used an AI program to write the brief; instead they were sanctioned because the AI program had “made up” the names and legal citations for several cases to support the brief’s legal position! The attorneys made the mistake of assuming that the information was accurate, because the legal citations appeared to be in the proper format. In other words, they relied on the superficial appearance of the information, rather than taking the time to check that it came from a legitimate source.

A Sandford University study recommends applying a different mindset to web-based publications, one that takes into account how easy it now is for anyone to impersonate legitimate resources and to post false information. The recommended approach is not to look at the website or its materials to validate the information, but instead to access external unrelated sources to evaluate the website’s sponsoring organization and materials contained on that website. To do this effectively and efficiently, web-based search engines (such as Google) and multiple fact checking websites can be used. While this approach may not be foolproof, it does capitalize on the ease of finding independent third-party resources to evaluate both the efficacy of the website sponsor and the accuracy of information it contains.

Social & News Media Sites: Applying Critical Thinking Skills to Overcome Conformation Bias Algorithms.

No amount of fact and source checking can fully counter the internet’s ability, through search engine algorithms, to feed us a nearly unending supply of whatever information we ask for. These algorithms have been very beneficial. Most of us use them every day to do a variety of tasks, such as evaluating a product we are considering purchasing, repairing or obtaining instructions on how to use an appliance, selecting a hotel or vacation resort, or even finding source material for a blog. However, these algorithms are most useful if we do not lose sight of the fact that most often they are optimized to raise revenue for the sponsor. We also need to consider that the same algorithm that feeds us an endless supply of cute kitten videos may also be used to keep us engaged on websites featuring news and social issues.

Again, no conspiratorial motive seems to be at work here; it’s just application of a time-honored principle of advertising to this new form of mass-media: if you want to get someone’s interest, use flashy emotion-based content, and if you want to keep them interested, show them more and more of it, making each subsequent “click” just a little flashier and more emotional. Today this happens with little or know human input at all; it’s a product of algorithms that could care less whether the topic you are viewing is cute kittens or gun control.

A 2022 study published by researchers at the University of California-Davis illustrates how this technology operates when the topic involved is a political or social issue. The goal of the study was to see how YouTube’s recommended content would change over time, when viewers initially selected a political topic.  The study, described in an August 2022 article, assessed the effect of following YouTube’s “recommended” videos. The idea behind the study was to determine what happened if users viewing a political video, followed the YouTube recommendation for the next video, and the video after that one.

Researchers in the study created fictitious YouTube accounts (sock puppets) that were programmed to access and “view” a video initially tagged by the researchers as having either a slightly conservative or as slightly liberal/progressive viewpoint. After viewing the video, each sock puppet automatically accessed the next recommended video, and viewed it. The sock puppets repeated this process over a number of days, accessing many videos.    

One result that is not all that surprising, was that sock puppets that initially viewed a conservative or a liberal/progressive bias tended to only be recommended videos that matched their original view. That makes sense; that is how confirmation bias operates. If one initially prefers and connects to content that had liberal/progressive bias, they are more likely to like and view more videos that support or “confirm” that bias, as opposed to videos that promote an alternative “conservative” viewpoint.  Of course, the same principle holds true for those that prefer content with a more conservative bias. Muhammad Haroon, the leader of the study noted: “Unless you willingly choose to break out of that loop, all the recommendations on that system will be zeroing on that one particular niche interest that they’ve identified.”

What was more disturbing, was that the YouTube algorithm tended not only to limit views to conservative or liberal/progressive content (depending on which bias was initially selected); the content selected tended to become increasingly more radical the longer the program ran. Again, this doesn’t imply any nefarious intent on the part of the YouTube algorithm programmers; it just seems to be a logical extension of a program designed to give the viewer more and more of content that supports their original bias. Again, the purpose might be solely to maximize the time spent on the website and the revenues generated from advertising, but obviously for the viewer, the algorithm shuts out competing voices and apparently over time tends to emphasize more extreme positions.

Another university study published in 2023, found that the risk of being caught up in group think and misleading or false information on the web tends to be directly related to the viewer’s analytic reflection skills. The study compared the web browsing activities of individuals that scored higher in analytic reflection on a standardized Cognitive Reflection Test (CRT), to a second group that tended to rely primarily on “intuitive reasoning” to reach conclusions. The study found that individuals with higher analytic reflection skills as measured by the CRT, appeared to be better able to counteract the tendency to view only those websites that confirmed their initial bias.

The CRT used in the study was designed to measure analytic reflections skills by asking participants to answer a number of questions that included an option that, while they at first seemed intuitively correct, on reflection were obviously wrong. For example:

“If you’re running a race and you pass the person in second place, what place are you in?”

The “intuitive” answer – the one that initially seems most appealing – is “first place.” However, after a little reflection one recognizes this answer is clearly wrong. After all, if the person you were trailing in the race was in second place, passing them only means you are now in second place, and you still need to catch the person who is in the lead.

When comparing the web browsing activities of the two groups, the study found that individuals who scored higher in analytic reflection skills also tended to rely on more traditional and reliable sources of news and information. The study concluded that these participants tended “to be more discerning in their social media use: they followed fewer [Twitter] accounts, shared higher quality content from more reliable sources, and tweeted about weightier subjects (in particular, politics).”

Critical Thinking is a Skill, Not a Measure of Intelligence

One misconception about critical thinking is that it is somehow directly related to the level of education and innate intelligence. In other words individuals with higher levels of formal education will be good “critical thinkers.” Academic research does not support this view. While critical thinking can be a complex process that includes clearly defining the issue, identifying and analyzing the sources of the information, testing and seeking confirmation from alternative sources and checking for alternative viewpoints; it is a skill. Equally important, hundreds of separate studies over many decades have shown that critical thinking can be learned, regardless of an individual’s age or education level.  Learning that skill is not limited to those with a college degree, and having a college degree is certainly no guarantee that the individual is a good a critical thinker.

Applying Critical Thinking Skills to the Internet – A Path Forward

The challenges posed by internet disinformation and confirmation bias algorithms can seem insurmountable.  We live in a free and open society. Any form of censorship or controls over web content run against our core belief in freedom of speech and expression. Of course, as a practical matter the internet itself is structured in ways that make regulation of its content, either by the government or industry difficult to implement even in authoritarian societies.

Responsible internet content providers may be able to provide tools that are useful in identifying disinformation on the internet. However, internet experts polled in a 2017 Pew Research Center study were evenly split on whether misinformation on the internet can be reduced in the future, and there is little reason to believe the results would be different today. Likely real improvement will depend in part on us, the individuals that use the internet on a daily basis. If we develop and use the skills necessary to avoid falling prey to disinformation, the rationale for posting it in the first place will be reduced. However, this will happen only if critical thinking skills training are a core part of improving adoption of the internet and internet-based applications.

The internet is a relatively new medium of mass communication, and new products and innovations that enhance its capabilities to provide information come to market almost daily. It is a much more powerful means of conveying information – and disinformation – than anything that has come before. There are few legal or practical restrictions on what content can be added to the internet, or on who is able to add it. But most of us would not want it any other way. As was true in the past, critical thinking skills can be learned by anyone, at any age, and there is evidence that they can be highly effective in identifying disinformation, particularly if the skills learned are adapted to account for ways in which the web differs from other forms of mass media technology that came before.

Yet it also is evident that developing the skills necessary to separate facts from false or misleading content on the web may not be enough to reduce society’s polarization and help us find common ground to peacefully resolve our most difficult policy issues. As a society, we have always had issues of disagreement that require compromise, but we never have had to reach compromises when large segments of the population have isolated themselves in “opinion silos” created by web-based algorithms.

More of us now rely on the internet as our primary – and in some cases our sole – source for news and opinion.  Limiting consumption of news and opinions on the web to one or two social media websites seems to create the very real possibility that, like the “sock puppets” in the YouTube video research experiment, we will be fed only a steady diet of confirmation-biased information that resonates with our world view and suppresses all others.

In the past, our parents and grandparents could counter this risk by subscribing to multiple newspapers, or at least they could read and consider opposing viewpoints on the op-ed page of their newspaper. Our generation can find alternative viewpoints as well. In fact, an important advantage of the internet is that it contains many points of view and information sources. However, accessing those views today likely requires that we take affirmative steps to break out of the “comfort zone” of algorithm-generated content, so that we at least understand the views of those with whom we disagree.  

Over the next several years, government and private business will invest more than $100 billion to bring affordable reliable internet infrastructure to every home and business in the United States. Funding through the Affordable Connectivity Program is available to help make the internet affordable for everyone regardless of income. However, to take full advantage of this resource, to use it safely and effectively, we also must develop and implement programs to make skills-based critical thinking training generally available.