Connect with us

Expert Opinion

Trevor Wagener: State Regulation of Content Moderation Would Create Enormous Legal Costs for Platforms

Published

on

The author of this Expert Opinion is Trevor Wagener, director of research and economics at CCIA

Dozens of bills regulating internet content moderation have been proposed in at least 30 state legislatures, and one in Utah currently sits on the governor’s desk awaiting his veto or signature. If enacted, many of the bills would impose prescriptive state regulations governing internet content moderation practices, with requirements differing significantly from state to state.

Many of the bills would also create private rights of action or state enforcement action powers. This could lead to enormous legal costs for platforms. If several of these bills are enacted into law, the resulting patchwork of regulatory requirements and legal risks could create daunting compliance challenges, astronomical costs, and unpredictable legal liability, which only the largest of platforms might be able to manage.

At present, each American internet platform can choose to moderate user-generated content using a variety of approaches depending on the nature of the platform and the desired user experience. A common approach is to utilize automated content moderation to perform triage on specific categories of prohibited content that can be categorized easily. For some, databases already exist. Human moderators are then used to judge a small fraction of hard-to-categorize content.

Automation is the primary tool in content moderation because of efficiency

Using automation as the primary tool in content moderation is common because of cost-efficiency—for example, Microsoft and Alibaba both offer third-party automated content moderation services for about $0.40 per 1,000 images or posts at scale. This implies an automated content review cost of about $0.0004 per image or post. When a human moderator must review content, by contrast, it takes an average of 150 seconds for them to review a post on even the most efficient platforms.

This work requires skill and judgment: Moderators often must quickly make complex decisions with limited context. Even assuming services could retain personnel to do this challenging work at $15/hour, the cost would be at least $0.625 per reviewed post, or 1,500 times the cost of automated moderation.

Some state bills, including Florida’s H.B. 7013 and Utah’s S.B. 228 require platforms to send a written notice to the user and/or the state government every time a post is moderated. This could require a human moderator for most or all user-generated content in those states, increasing moderation costs by a factor of 1,500. In cases where moderation would affect users from multiple states, the process could require two human moderators’ time, duplicating costs.

Florida’s moderation bill also mandates that users be able to request detailed information about the moderation of their posts, while Utah’s mandates that users be allowed to appeal moderation decisions, have appeals reviewed by a human moderator, and even have that moderator explain their decision to the user.

State bills will exponentially increase the cost of content moderation

Depending on the post, the requests for information, the appeal process, and the specific requirements of each bill, that could take anywhere from 5 minutes to half an hour of a human moderator’s time for each instance. At $15/hour, that would cost between $1.25 to $7.50 on top of the $0.625 for initial review. For each request, the process could cost 3,000 to 18,000 times as much as the present baseline.

Not every moderated post would be appealed. However, with current appeal rates of one in 40 to one in 10, the costs would add up. With content moderation rates ranging from one per 20 users annually on some platforms to almost twice per user annually on others, we can establish a range of incremental compliance costs for a hypothetical startup with 20 million users.

These compliance costs multiply with the number of distinct state content moderation bills enacted. If users from multiple states engage with a post, that could require duplicative reviews of the same content based on distinctive state requirements. If only two out of fifty states enact such bills, duplicated reviews might be required for 4 percent of moderated posts; if thirty out of fifty states enact such bills, duplicated reviews might be required for 60 percent of moderated posts.

In the low case, where one in 20 users has a post moderated, one in 40 moderations are appealed, each appeal takes about 5 minutes, and only 4 percent of moderations involve duplicated review of content from users from multiple covered states, that adds up to about $0.68 million in aggregate annually, or $0.034 per user annually.

In the high case, where two posts per user are moderated, one in ten moderations are appealed, each appeal takes about half an hour, and 60 percent of moderations involve duplicated review of content from users from multiple covered states, that adds up to $70 million in aggregate annually, or $3.50 per user annually.

Even in the best case, a ‘low-cost’ compliance total could be ruinous

For a startup, even the low-case total compliance cost of about $0.68 million annually could be ruinous. The high-case total cost of $3.50 per user annually would be an enormous financial hit for platforms of all sizes, since leading platforms only have global average revenue per user of about $10 annually. Therefore, a patchwork of state content moderation bills could unwittingly create a significant barrier to entry, discouraging tech startups from operating in certain states. This would reduce competition, inhibit job creation and curtail the range of online services available in those states.

In addition, some state content moderation bills, such as Oklahomas S.B. 383, Iowas S.F. 580, and North Dakotas H.B. 1144, create legal risks for online platforms. Purported failures to follow state-specific content moderation rules could subject platforms to both government enforcement actions or private rights of action. Even in clear-cut cases with all facts in the platform’s favor, litigation could cost $80,000 through a Motion to Dismiss or $150,000 through a Motion for Summary Judgment, in addition to possible statutory and/or punitive damages. This would be on top of compliance costs.

It is difficult to forecast legal costs under a patchwork of distinct state regulations, as they scale with the number of affected users and the number of states with such bills enacted. Some of the bills would grant rights of action to the user posting content, while others would grant rights of action to users prevented from seeing content. Some would grant private rights of action to both. Many create enforcement powers for state governments in addition to private rights of action. In principle, the moderation of a single post with comments from residents of multiple states could result in a number of enforcement actions and private lawsuits greater than the total number of affected users times the number of states they reside in.

Spiraling legal costs associated with even a single post by a user in a covered state

The moderation of a single hypothetical post by a user in one covered state with comments by users in 29 other covered states would result in 30 affected users in 30 jurisdictions. This could mean more than 900 lawsuits, each plausibly costing the platform $80,000 for a Motion to Dismiss. If each were litigated, that would cost the platform at least $72 million for a single content moderation decision even if one assumes the success of every Motion to Dismiss. These legal liabilities would be capable of bankrupting a startup or midsize platform. If even one in a thousand moderated posts were fully litigated, platforms would have to account for this enormous tail risk as a recurring phenomenon, as many platforms moderate thousands of posts per day.

Startups with limited initial capital are reluctant to operate in jurisdictions with higher legal risk. They may not only avoid locating offices and jobs in high-risk states but also bar state residents from using their platforms altogether. Platforms that continue operating or offering services in such states may dramatically curtail content moderation to manage risk, resulting in a dramatic increase in financial scams, lewd content, and all-around abusive and unpleasant content. This would likely reduce use of such platforms, decreasing both consumer welfare and platform revenue.

In addition to deleterious impacts on platforms, users, and competition, a patchwork of state content moderation bills would burden taxpayers with administrative costs, enforcement costs, and possible litigation damages.

Administrative costs for most state content moderation bills would likely be in excess of Utah’s estimate of $90,000 annually for each such state. In addition, enforcement costs would likely surpass Iowa’s estimate of $700,000 annually for each such state. Given constitutional concerns around such bills and potential legal challenges to governments that force private businesses to host content to which they object, each such state could easily find itself paying damages and fees.

Trevor Wagener is the director of research and economics at the Computer & Communications Industry Association.  Wagener previously served as deputy chief economist of the U.S. Department of State. This piece is exclusive to Broadband Breakfast.

Broadband Breakfast accepts commentary from informed observers of the broadband scene. Please send pieces to commentary@breakfast.media. The views expressed in Expert Opinion pieces do not necessarily reflect the views of Broadband Breakfast and Breakfast Media LLC.

Expert Opinion

Dmitry Sumin: What to Do About Flash Calls, the New SMS Replacement

Why are flash calls on the rise and how do operators handle them to maximize revenue?

Published

on

The author of this Expert Opinion is Dmitry Sumin, AB Handshake Corporation Head of Products

Chances are you’ve received several flash calls this week when registering for a new app or verifying a transaction. Flash calls are almost instantly dropped calls that deliver one-time passcodes to users, verifying their phone numbers and actions. Many prominent apps and companies, such as Viber, Telegram, WhatsApp, and TikTok, use flash calls as a cheaper, faster, and more user-friendly alternative to application-to-person SMS.

With the flash call volume expected to increase 25-fold from 2022 to 2026, from five to 130 billion, it’s no wonder they’re a hot topic in the telecom industry.

But what’s the problem, you may ask?

The problem is that there is currently no way for operators to bill zero-duration calls. This means operators don’t make any termination revenue from flash calls, which overload networks. What’s more, operators lose SMS termination revenues as businesses switch to flash calls. SMS business messaging accounts for up to five percent of total operator-billed revenue in 2021, so you can see the scale of potential revenue losses for operators. 

In this article, I’ll discuss why flash calls are on the rise, why it’s difficult to detect and monetize them, and what operators can do about this.

Why are flash calls overtaking SMS passcodes?

Previously, application-to-person SMS was a popular way to deliver one-time passwords. But enterprises and communication service providers are increasingly switching to flash calls because they have several disruptive advantages over SMS.

First and foremost, flash calls are considerably cheaper than SMS, sometimes costing up to eight times less. Cost of delivery is, of course, a prime concern for apps and enterprises.

Second, flash calls ensure smooth user interaction, which boosts user satisfaction and retention. On Androids, mobile apps automatically extract flash call passcodes. This makes the two-factor authentication process fast and frictionless. In comparison, SMS passcodes require users to read the SMS and sometimes insert the code manually.

Third, on average flash calls reach users within 15 seconds, while SMS sometimes take 20 seconds or longer. The delivery speed of flash calls also improves the user experience.

The problem: Flash calls erode operators’ SMS revenues

While offering notable advantages for apps, flash call service providers, and end users, flash calls create numerous challenges for operators and transit carriers.

As we discussed before, flash calls erode operators’ SMS revenues because much of the new flash call traffic will be shifted away from current SMS business messaging. The issue is only going to become more pressing as the volume of flash calls grows.

So from the operator’s standpoint, flash calls reduce revenue, disrupt relations with interconnect partners, and overload networks. However, there is still no industry consensus on how to handle flash calls: block them like spam and fraudulent traffic or find a monetization model for this verification channel, like for application-to-person SMS.

Accurate detection of flash calls is a challenge

The first crucial step that gives operators the upper hand is accurately detecting flash calls.

This is difficult because operators have no way of discerning legitimate verification flash calls from fraud schemes that rely on drop calls, such as wangiri. The wangiri fraud scheme uses instantly dropped calls to trick users into calling back premium rate numbers. In addition, flash calls need to be distinguished from genuine missed calls placed by customers.

The problem is that even advanced AI-powered fraud management systems struggle to accurately differentiate between various zero-duration calls. The task requires AI engines to be trained on large volumes of relevant traffic coupled with analysis of hundreds of specific call parameters.

Dedicated anti-fraud solutions are the answer

There are only a few solutions on the market that are capable of accurately distinguishing flash calls from other zero-duration calls. Dedicated fraud management vendors have made progress on this difficult task.

The highest accuracy of flash call detection now available on the market is 99.92 percent. Such tools allow operators to precisely determine the ranges from which flash calls are sent. As a result, operators can make an informed decision on how to treat flash calls to maximize revenue and can proactively negotiate with flash call providers.

Flash call detection creates new opportunities

Our team estimates that flash calls make up to four percent of Tier one operators’ international voice traffic. Without accurate detection and a billing strategy, this portion of traffic overloads operators’ networks and offers no revenue. However, with proper detection flash calls offer a new business opportunity.

Now is a crucial time for operators to start implementing flash call detection into their system and capitalize on the trend.

There are a few anti-fraud solutions on the market that give operators all the necessary information to negotiate a billing agreement with a flash call provider. Once an agreement has been reached, all flash calls coming from this provider will be monetized, much like SMS.

All flash calls not covered by agreements can be blocked automatically. This will help to restore SMS revenues. Once a flash call has been blocked, subscribers will most likely receive an SMS passcode sent as a fallback.

Moreover, modern solutions don’t affect any legitimate traffic because they only block selected ranges. This also helps to prevent revenue loss.

Essentially, the choice of how to handle flash calls comes down to each operator. However, without a powerful anti-fraud solution capable of accurately detecting flash calls in real time, it’s nearly impossible to monetize flash calls effectively and develop a billing strategy.

Dmitry Sumin is the Head of Products at the AB Handshake Corporation. He has more than 15 years of experience in international roaming, interconnect and fraud management. Since graduating from Moscow State University, he has worked for both vendors and network operators in the MVNO and telecommunications market. This piece is exclusive to Broadband Breakfast.

Broadband Breakfast accepts commentary from informed observers of the broadband scene. Please send pieces to commentary@breakfast.media. The views reflected in Expert Opinion pieces do not necessarily reflect the views of Broadband Breakfast and Breakfast Media LLC.

Continue Reading

Expert Opinion

Bjorn Capens: Strong Appetite for Rural Broadband Calls for Next Generation Fiber Technology

The first operator to bring fiber to a community creates a significant barrier to entry for competitors.

Published

on

The author of this Expert Opinion is Björn Capens, Nokia Fixed Networks European Vice President

In July, the Biden-Harris administration announced another $401 million in funding for high-speed Internet access in rural America. This was just the latest in a string of government initiatives aimed at helping close the US digital divide.

These initiatives have been essential for encouraging traditional broadband providers, communities and utility companies to deploy fiber to rural communities, with governments cognizant of the vital role broadband connectivity has in sustaining communities and improving socio-economic opportunities for citizens. 

Yet there is still work to do, even in countries with the most advanced connectivity options. For example, fixed broadband is missing from almost 30 percent of rural American homes, according to Pew Research. It’s similar in Europe where a recent European Commission’s Digital Divide report found that roughly 18 percent of rural citizens can only get broadband speeds of a maximum 30 Mb, a speed which struggles to cope with modern digital behaviors. 

Appetite for high-speed broadband in rural areas is strong

There’s no denying the appetite for high-speed broadband in rural areas. The permanent increase in working from home and the rise of modern agricultural and Industry 4.0 applications mean that there’s an increasingly attractive business case for rural fiber deployments – as the first operator to bring fiber to a community creates a significant barrier to entry for competitors. 

The first consideration, then, for a new rural fiber deployment is which passive optical network technology to use. Gigabit PON seems like an obvious first choice, being a mature and widely deployed technology. 

However, GPON services are a standard offering for nearly every fiber broadband operator. As PON is a shared medium with usually up to 30 users each taking a slice, it’s easy to see how a few Gigabit customers can quickly max out the network, and with the ever-increasing need for speed, it’s widely held that GPON will not be sufficient by about 2025. 

XGS-PON is an already mature technology

The alternative is to use XGS-PON, a more recent, but already mature, flavor of PON with a capacity of 10 Gigabits per second. With the greater capacity, broadband operators can generate higher revenues with more premium-tier residential services as well as lucrative business services. There’s even room for additional services to run alongside business and residential broadband. For example, the same network can carry traffic from four G and five G cells, known as mobile backhaul. That’s either a new revenue opportunity or a cost saving if the operator also runs a mobile network. 

This convergence of different services onto a single PON fiber network is starting to take off, with fiber-to-the-home networks evolving into fiber for everything, where homes, businesses, industries, smart cities, mobile cells and more are all running on the same infrastructure. This makes the business case even stronger. 

Whether choosing GPON or XGS-PON, the biggest cost contributor is the same for both: deploying fiber outside the plant. Therefore, the increased cost of XGS-PON over GPON is far outweighed by the capacity increase it brings, making XGS-PON the clear choice for a brand-new fiber deployment. XGS-PON protects this investment for longer as its higher capacity makes it harder for new entrants to offer a superior service. 

It also doesn’t need to be upgraded for many years, and when it comes to the business case for fiber, it pays to take a long-term view. Fiber optic cable has a shelf-life of 75 or more years, and even as one increases the speeds running on fiber, that cable can remain the same.  

Notwithstanding these arguments, fiber still comes at a cost, and operators need to carefully manage those costs in order to maximize returns. 

Recent advances in fiber technology allow operators to take a pragmatic approach to their rollouts. In the past, each port on a PON server blade could only deliver one technology. But Multi-PON has multiple modes: only GPON, only XGS-PON or both together. It even has a forward-looking 25G PON mode. 

This allows an operator to easily boost speeds as needed with minimal effort and additional investment. GPON could be the starting point for fiber-to-the-home services, XGS-PON could be added for business services, or even a move to 25G PON for a cluster of rural power users, like factories and modern warehouses – creating a seamless, future-proof upgrade path for operators. 

The decision not to invest in fiber presents a substantial business risk

Alternatively, there’s always the option for a broadband operator to stick with basic broadband in rural areas and not invest in fiber. But that actually presents a business risk, as any competitor that decides to deploy fiber will inevitably carve out a chunk of the customer base for themselves. 

Besides, most operators are not purely profit-driven; they too recognize that prolonging the current situation in underserved communities is not great. High-speed broadband makes areas more attractive for businesses, creating more jobs and stemming population flows from rural to urban centers. 

But rural broadband not only improves lives, but it also decreases the world’s carbon emissions both directly, compared to alternative broadband technologies, and indirectly by enabling online and remote activities that would otherwise involve transportation. These social and economic benefits of fiber are highly regarded by investors and stockholders who have corporate social responsibility high on their agendas. 

With the uber-connected urban world able to adopt every new wave of bandwidth-hungry application – think virtual reality headsets and the metaverse – rural communities are actually going backwards in comparison. The way forward is fiber and XGS-PON. 

Björn Capens is Nokia Fixed Networks European Vice President. Since 2017, Capens has been leading Nokia’s fixed networks business, headquartered in Antwerp, Belgium. He has more than 20 years of experience in the fixed broadband access industry and holds a Master’s degree in Electrical Engineering, Telecommunications, from KU Leuven. This piece is exclusive to Broadband Breakfast.

Broadband Breakfast accepts commentary from informed observers of the broadband scene. Please send pieces to commentary@breakfast.media. The views reflected in Expert Opinion pieces do not necessarily reflect the views of Broadband Breakfast and Breakfast Media LLC.

Continue Reading

Expert Opinion

Johnny Kampis: Federal Bureaucracy an Impediment to Broadband on Tribal Lands

18% of people living on Tribal lands lack broadband access, compared to 4% of residents in non-tribal areas.

Published

on

The author of this expert Opinion is Johnny Kampis, director of telecom policy for the Taxpayers Protection Alliance

A new study from the Phoenix Center finds that as the federal government pours tens of billions of dollars into shrinking the digital divide in tribal areas, much of that gap has already been eliminated.

The report, and a second from the U.S. Government Accountability Office, are more indications that regulations and economic factors that include income levels continue to hamper efforts to get broadband to all Americans.

The Infrastructure Investment and Jobs Act of 2021 allocated $45 billion toward tribal lands. This was done as part of a massive effort by the federal government to extend broadband infrastructure to unserved and underserved areas of the United States.

George Ford, chief economist at the Phoenix Center for Advanced Legal & Economic Public Policy Studies, wrote in the recent policy bulletin that while there is still plenty of work needed to be done in terms of connectivity, efforts in recent years have largely eliminated the broadband gap between tribal and non-tribal areas.

Ford examined broadband deployment around the U.S. between 2014 and 2020 using Form 477 data from the Federal Communications Commission, comparing tribal and non-tribal census tracts.

Ford points out in the bulletin that the FCC has observed several challenges for broadband deployment in tribal areas, including rugged terrain, complex permitting processes, jurisdictional issues, a higher ratio of residences to business customers, higher poverty rates, and cultural and language barriers.

Ford controlled for some of these differences in his study comparing tribal and non-tribal areas. He reports in the bulletin that the statistics suggest nearly equal treatment in high-speed internet development.

Encouraging results about availability of broadband in Tribal areas

“These results are encouraging, suggesting that broadband availability in Tribal areas is becoming closer or equal to non-Tribal areas over time, and that any broadband gap is largely the result of economic characteristics and not the disparate treatment of Tribal areas,” Ford wrote.

But he also notes that unconditioned differences show a 10-percentage point spread in availability in tribal areas, which indicates how much poverty, low population density, and red tape is harming the efforts to close the digital divide there.

“These results do not imply that broadband is ubiquitous in either Tribal or non-Tribal areas; instead, these results simply demonstrate that the difference in availability between Tribal and non-Tribal areas is shrinking and that this difference is mostly explained by a few demographic characteristics,” Ford wrote.

In a recent report, the GAO suggests that part of the problem lies with the federal bureaucracy – that “tribes have struggled to identify which federal program meets their needs and have had difficulty navigating complex application processes.”

GAO states that 18 percent of people living on tribal lands lack broadband access, compared to 4 percent of residents in non-tribal areas.

The GAO recommended that the Executive Office of the President specifically address tribal needs within a national broadband strategy and that the Department of Commerce create a framework within the American Broadband Initiative for addressing tribal issues.

“The Executive Office of the President did not agree or disagree with our recommendation but highlighted the importance of tribal engagement in developing a strategy,” the report notes.

That goes together with the GAO’s dig at the overall lack of a national broadband strategy by the Biden Administration in a June report. As the Taxpayers Protection Alliance reported, the federal auditor noted that 15 federal agencies administer more than 100 different broadband funding programs, and that despite a taxpayer investment of $44 billion from 2015 through 2020, “millions of Americans still lack broadband, and communities with limited resources may be most affected by fragmentation.”

President Biden has set a goal for universal broadband access in the U.S. by 2030. These recent reports show that the federal bureaucracy under his watch needs to do a better job of getting out of its own way.

Johnny Kampis is the director of telecom policy for the Taxpayers Protection Alliance. This piece is exclusive to Broadband Breakfast.

Broadband Breakfast accepts commentary from informed observers of the broadband scene. Please send pieces to commentary@breakfast.media. The views reflected in Expert Opinion pieces do not necessarily reflect the views of Broadband Breakfast and Breakfast Media LLC.

Continue Reading

Signup for Broadband Breakfast

Get twice-weekly Breakfast Media news alerts.
* = required field

Broadband Breakfast Research Partner

Trending