Category: Dead Internet Theory (DIT)

Dead Internet Theory (DIT) explores the idea that the open, human-driven web is being steadily replaced by automated content, algorithmic feeds, corporate platforms, abandoned websites, SEO spam, bots, and machine-generated noise.

This category investigates the evidence behind that idea without treating every claim as fact. We look at vanished communities, once-massive websites turned into digital ghost towns, bot networks, content farms, spam empires, dead links, abandoned forums, artificial engagement, platform consolidation, algorithm-driven culture, and the strange remains of the older internet.

Some of it is conspiracy theory. Some of it is ordinary technological change. And some of it is plainly visible to anyone who remembers what the web used to feel like.

These articles document the transformation of the internet from a sprawling collection of strange, independent places into something increasingly centralized, automated, commercialized, and difficult to distinguish from its own simulations.

Welcome to the archaeological layer of the modern web.

Posted on by

Open protocols as a way to reduce dependence on one provider

The easiest platform to leave is the one that does not own the entire network.

ActivityPub is built around that idea.

The W3C describes ActivityPub as a decentralized social-networking protocol with both client-to-server and federated server-to-server APIs. In plain English: different services can run different software on different servers and still exchange social activity with one another.

See the W3C ActivityPub Recommendation.

A provider can be replaced without replacing everybody

Mastodon is one widely used ActivityPub implementation.

A Mastodon user belongs to a particular server, but can follow and interact with compatible accounts on other servers. The server operator controls the local account and local moderation. The wider social graph crosses provider boundaries.

That changes the cost of leaving.

Mastodon’s migration tools allow a user to create an account on another server and initiate a move. Compatible software can transfer followers to the new identity. Following lists, blocks, mutes, and other lists can also be exported and imported.

See Mastodon’s account migration documentation.

The person still changes providers.

The entire network does not have to change providers with them.

That is a very different failure mode from a closed social platform where leaving the operator means leaving everybody who remains inside it.

Open does not mean frictionless

The protocol does not magically make every dependency portable.

Mastodon currently does not import a user’s old posts or media into the new account. Followers can move, but followed accounts must be imported separately. Some relationships may require new approval. A server administrator still controls the machine hosting the account and can establish local rules or shut the server down.

Different ActivityPub implementations also support different features.

Interoperability reduces dependence. It does not erase it.

Protocols change where the leverage sits

Email provides the older version of the same lesson. Gmail does not need to own Outlook for a Gmail user to send an Outlook user a message because the network is built around shared protocols.

ActivityPub applies that structure to social networking.

The practical value is not philosophical purity.

It is bargaining power.

If one provider becomes expensive, poorly moderated, unstable, or simply annoying, users have at least some path to another provider without demanding a synchronized migration from the whole community.

A protocol cannot guarantee a good service.

It can make one bad service less synonymous with the network itself.

Posted on by

Cooperatives and member ownership as alternative service incentives

A platform does not have to be owned by people who never upload anything to it.

Stocksy United is one useful counterexample.

Stocksy operates a stock-media marketplace, but it is structured as a cooperative. Contributor members are not merely photographers supplying inventory to a separate company. They can also be co-owners of the business using their work.

Ownership changes what the platform is optimizing for

Stocksy says contributor members receive 50% royalties on standard licenses and 75% on extended licenses and market freezes. If the cooperative has a year-end surplus, its board may also distribute patronage returns to eligible members.

See Stocksy’s contributor compensation explanation and patronage documentation.

More important than the percentages is the governance structure.

Stocksy says members can participate through board elections, resolutions, voting, and annual general meetings. Its own description is explicit: members are co-owners with a formal voice in business decisions.

See Stocksy’s membership overview.

That does not guarantee that every member wins every argument.

It changes who is inside the argument.

Investor ownership and member ownership create different pressures

A conventional marketplace can increase profit by reducing contributor payouts, raising seller fees, increasing advertising, or changing ranking rules. Those decisions may still be sensible or necessary, but the suppliers affected by them generally do not elect the board merely because they supply the marketplace.

A cooperative has a different constraint. The people producing part of the platform’s value also possess governance rights in the organization distributing that value.

That does not eliminate management, budgets, or painful decisions.

It changes the constituency those decisions must answer to.

Cooperatives do not remove economics

Member ownership is not free hosting with a nicer logo.

Stocksy is curated rather than open to every applicant. Infrastructure still costs money. Staff still need salaries. Members can disagree. Governance takes time. Patronage depends on there being a surplus to distribute in the first place.

A cooperative can fail just as an investor-owned company can fail.

The model is useful because it changes the incentive map, not because it abolishes constraints.

When people complain that a platform became hostile after achieving scale, ownership is one of the variables worth examining.

Who supplies the value?

Who receives the surplus?

Who gets a vote when those two answers start drifting apart?

Posted on by

Exit campaigns that fail because participants cannot coordinate departure

Leaving a bad restaurant is easy.

Leaving a social network is a group project.

That difference helps explain why public campaigns to abandon large platforms can generate enormous discussion and surprisingly little movement.

Quit Facebook Day tried it in 2010.

Thousands agreed. Hundreds of millions stayed.

Two Canadian campaigners designated May 31, 2010 as Quit Facebook Day amid growing criticism of Facebook’s privacy changes. The campaign received substantial press attention and asked users to delete their accounts together rather than grumble separately.

The result was tiny compared with the network.

Dark Reading reported 34,388 pledges after the event against an estimated Facebook population of more than 540 million users. Contemporary Guardian coverage documented the privacy backlash and the organizing effort.

See Dark Reading’s post-event report and The Guardian’s contemporary coverage.

It would be easy to interpret that result as 540 million votes of confidence in Facebook.

It was not.

Everyone has to choose where Tuesday happens

A social network becomes useful partly because other people are there.

If ten friends dislike a platform, each one can still have a reason to stay because the other nine remain. A family group may not want to move until the whole family agrees. A club may need one destination that works for its least technical member. A business cannot abandon the place where customers still send messages just because the owner hates the interface.

Even an organized exit has to answer several questions at once:

Where are we going?

Will our contacts go there too? Can we move photos, messages, groups, and identities? Which day do we switch? What happens to people who refuse?

A campaign can therefore fail even when its complaint is widely shared.

Staying is not the same as choosing

Network effects create a coordination problem.

The individual decision that makes sense depends on what everyone else does. If most people remain, leaving can reduce the value of the service for the person who leaves while doing almost nothing to the platform.

That gives an established network an unusual kind of resilience.

People can complain, distrust a policy, install alternatives, and still keep the old account because the social graph has not moved with them.

This is why raw user counts need interpretation.

A person who stayed may be satisfied.

Or the person may simply have failed to convince 47 relatives, three clients, a school group, and the guy organizing Saturday’s barbecue to leave at the same time.

Posted on by

The gap between user satisfaction and financial performance

A company does not need delighted users to have an excellent quarter.

Facebook makes that distinction easy to see because both sides are measured.

The American Customer Satisfaction Index’s 2024 social-media study gave Facebook a score of 69 out of 100. That was an improvement over 2023, but it still left Facebook near the bottom of the major platforms measured. The same study found advertising and privacy remained weak parts of the social-media experience overall.

See the ACSI Search and Social Media Study 2024.

Meta’s financial numbers were moving in the other direction

Meta’s full-year 2024 results were extremely strong.

The company reported total revenue of $164.5 billion, up 22% from 2023. Advertising revenue reached about $160.6 billion, also up 22%. Across Meta’s Family of Apps, ad impressions increased 11% for the year and the average price per ad increased 10%.

See Meta’s 2024 full-year results.

These measurements are not directly interchangeable. ACSI surveys U.S. consumers about satisfaction with individual services. Meta’s financial results cover a global company whose Family of Apps includes Facebook, Instagram, Messenger, and WhatsApp.

So the numbers do not prove that frustrating Facebook users caused Meta’s revenue growth.

They show something more basic.

A platform can remain financially powerful while one of its core products earns mediocre satisfaction scores.

The customer is not always the revenue source

Advertising platforms have two overlapping constituencies.

Users supply attention, activity, data, and the social environment people return to. Advertisers supply most of the money.

That makes the relationship between user happiness and revenue less direct than it is for a simple paid product.

A restaurant that annoys most diners eventually has an obvious problem because the diner pays the bill.

A social platform can increase ad inventory, improve targeting, raise the price of advertising, or increase engagement even when users complain about the experience that produces those opportunities.

There is still a limit. If dissatisfaction becomes severe enough, people can leave or use the service less. Meta itself tracks engagement closely for exactly that reason.

But the limit may be far away.

That is why financial success cannot be used as a clean proxy for product quality.

Revenue answers is the business extracting value?

Satisfaction answers a different question: how do users feel about the service they are helping make valuable?

Posted on by

Digital purchases redefined by licensing and continued platform access

A physical disc does not phone the store to ask whether you still own Tuesday.

Digital purchases can be stranger.

In December 2023, Sony published a notice telling PlayStation customers that certain Discovery television content they had previously purchased would become unavailable because of content-licensing arrangements. The affected titles were to be removed from customers’ video libraries.

Then the plan changed.

The removal notice was real

Sony’s own legal notice originally said that, as of December 31, 2023, affected Discovery content would no longer be watchable and would be removed from users’ libraries.

On December 21, Sony updated the page again. Updated licensing arrangements meant the planned removal would no longer occur.

Both statements remain visible in the PlayStation legal notice.

The reversal was good news for customers.

It also demonstrated the underlying problem more clearly than an actual deletion would have.

Access to the purchased programs depended on a licensing relationship that could change after the transaction with the customer was finished.

Purchase is not always possession

When people hear buy, the ordinary expectation comes from physical objects.

You pay. The object becomes yours. The shop can close next week and your copy remains on the shelf.

Many digital storefronts work differently. The buyer receives licensed access through an account and a platform. Playback may depend on authentication, rights-management systems, servers, regional permissions, or contracts between the storefront and another rights holder.

The customer can have a receipt without having an independent copy that continues working outside that system.

That does not make every digital purchase deceptive. Terms can disclose the arrangement, and licensing is how enormous catalogs become available in the first place.

But the word purchase can encourage a stronger intuition of permanence than the infrastructure actually provides.

The platform is part of the product

A useful way to evaluate a digital purchase is to ask what must still exist five years later.

Does the file work independently?

Does an account need to authenticate it? Does the storefront need to remain online? Can a licensing agreement remove playback rights? Is there a way to export or preserve the purchase outside the platform?

The PlayStation Discovery episode ended without the threatened removal.

The important part is that Sony could announce the removal in the first place, then cancel it when licensing changed again.

The customer bought the content once.

The machinery behind continued access kept negotiating afterward.

Posted on by

Shutdowns of useful products that do not meet corporate growth targets

Google Reader was useful right up until Google killed it.

That is what makes the shutdown worth remembering.

Reader did one ordinary job extremely well: it collected updates from RSS feeds and let people follow many websites from one place. It did not need to become a social network, a video platform, or an advertising empire to be useful.

Google’s reason was not that RSS stopped working

Google launched Reader in 2005. In March 2013, the company announced that it would retire the service on July 1.

The explanation was short. Google said Reader had a loyal following, but usage had declined. The company also said it was concentrating its energy on fewer products.

See Google’s 2013 spring-cleaning announcement.

Users were given time to export subscriptions and other data through Google Takeout. RSS itself survived. Competing readers existed and more appeared after the shutdown.

So this was not a case where the underlying technology became impossible to operate.

It was a portfolio decision.

Useful to whom?

A product can fail one corporate test while passing the user’s test every morning.

For the person following hundreds of blogs, research sites, newspapers, or niche forums, Reader could save substantial time. For Google, a service with declining usage competed internally for engineering attention with products capable of reaching vastly larger audiences or producing more strategic value.

Both statements can be true at once:

Reader was useful.

Reader was not useful enough to Google.

That distinction gets lost whenever a shutdown is treated as proof that nobody wanted the product.

Corporate scale changes the definition of success

A small independent company may consider a stable service with a dedicated paying audience perfectly healthy.

A giant platform company may evaluate the same audience against opportunities measured in hundreds of millions or billions of users.

That creates a strange risk for customers. A product can be popular enough to become part of their routine and still be too small to survive inside the company that owns it.

The relevant question before depending on a free or subsidized service is therefore not only is this good?

It is also what would make this worth continuing for the operator?

Those are different questions. Google Reader answered the first one for years.

Posted on by

Platform terms changed after users invest in a dependent business

A dependency does not have to look like an API.

Sometimes it looks like five years of product reviews, comparison pages, search traffic, mailing lists, and links pointing to somebody else’s checkout.

Amazon Associates made that dependence unusually clear in April 2020.

The business had already been built

Publishers and independent site owners used Amazon’s affiliate program to earn a percentage when readers followed their links and bought qualifying products. Some businesses treated that as incidental income. Others built entire publishing models around it.

Then Amazon changed the rate card.

Reporting at the time documented large cuts effective April 21, 2020. Furniture, home, and home-improvement categories fell from 8% commissions to 3%. Grocery fell from 5% to 1%. Several other categories were also reduced.

See Marketing Dive’s April 2020 report and Search Engine Land’s rate breakdown.

Amazon had not promised those percentages forever. The Associates program operated under terms that Amazon could change.

Legally, that distinction matters.

Economically, it does not make the investment disappear.

Acceptance can be technically voluntary and practically ugly

A publisher faced with a lower rate could stop using Amazon links.

That sounds like a clean choice until the rest of the system is considered.

Years of articles may already contain Amazon links. Readers may trust Amazon’s checkout and shipping. Product databases, price widgets, review formats, and editorial workflows may all have been built around the program. Search rankings may belong to pages whose commercial logic assumes that Amazon converts the traffic.

A replacement affiliate network is not a drop-in substitute for that accumulated machinery.

The owner can reject the new terms. The cost of rejection may be rebuilding the business.

Platform risk hides inside ordinary growth

This is why dependent businesses can look healthier than they are.

Traffic may be rising. Search rankings may be strong. Revenue may be predictable. The site owner may have improved every part of the operation under personal control.

One external rate change can still rewrite the margins.

The lesson is not that nobody should build on platforms. Platforms can supply enormous reach, payment infrastructure, logistics, or customers that a small business could never reproduce alone.

The important measurement is concentration.

If one company’s terms determine whether the business works, then years of successful investment have not removed platform risk. They may have increased it.

Posted on by

Volunteer contributions monetized without corresponding contributor control

A volunteer-built knowledge base can become a valuable commercial asset without the volunteers ever becoming its governors.

Stack Overflow is a useful example because the arrangement is unusually visible. Millions of developers spent years asking questions, writing answers, editing posts, voting, and moderating. The public archive that resulted is useful precisely because those people did the work.

The contributors were not employees building a proprietary database for a salary. They were participants solving one another’s problems in public.

The content remains user-created

Stack Overflow says publicly accessible contributions are distributed under Creative Commons Attribution-ShareAlike licenses, with the exact version depending on when a revision was posted. That matters because contributors did not simply hand authorship over to the company.

See Stack Overflow’s content licensing explanation.

But authorship is not the same thing as control over the business built around the collection.

Stack Overflow has also developed commercial partnerships that provide structured access to the accumulated knowledge base. In 2024 it announced partnerships with both Google Cloud and OpenAI. The OpenAI announcement described API access to Stack Overflow’s technical knowledge, backed by contributions from millions of developers over roughly fifteen years.

See Stack Overflow’s OpenAI partnership announcement and Google Cloud partnership announcement.

Stack Overflow now explicitly advertises data-licensing partnerships for AI companies as part of its partnership business.

Contribution does not equal a board seat

None of this means Stack Overflow stole the posts or that commercial reuse is inherently improper. Public contributions were licensed for reuse, and contributors received a large public knowledge system in return.

The interesting part is the mismatch in authority.

A person can spend hundreds of hours producing answers, reviews, edits, and moderation decisions that increase the usefulness of the network. That effort may increase the commercial value of the corpus. But the contributor does not receive a corresponding vote over partnerships, pricing, acquisitions, or product strategy merely because the company became more valuable through community work.

The responsibility and the authority are distributed differently.

That distinction matters whenever a platform describes itself as a community while operating as an ordinary company. Community members may produce much of the thing people come to see. The company still decides what business will be built around it.

The question is not whether monetization is allowed.

It is whether the people doing unpaid production understand that participation can create an asset they help build but do not govern.

Posted on by

Community governance weakened after a change of ownership

Community governance does not require giving users the keys to the server.

Sometimes it means something much smaller: a defined route through which people outside the company can regularly influence policy, raise problems, and force uncomfortable subjects onto the agenda.

Twitter had one of those structures before its 2022 ownership change.

The Trust & Safety Council was advisory, but it was real

Twitter created its Trust & Safety Council in 2016. By 2019, the company described it as a group of more than 40 organizations and experts advising Twitter on products, programs, and rules. Twitter later expanded the structure into issue-specific advisory groups dealing with areas such as online safety, digital rights, child exploitation, dehumanization, and suicide prevention.

See Twitter’s archived description of strengthening the Trust & Safety Council.

The council did not run Twitter.

Its members could not veto a product launch, reverse an account suspension, or order the company to adopt a policy. Twitter retained final authority.

That limitation matters because calling the council “community governance” should not be confused with democratic control.

What the council did provide was institutionalized outside input. The company had named organizations, scheduled meetings, defined subject areas, and an ongoing mechanism for specialists to challenge or advise the people writing platform rules.

The ownership changed, then the structure disappeared

Elon Musk completed his acquisition of Twitter in October 2022.

On December 12, 2022, Twitter dissolved the Trust & Safety Council. Reuters reported that members received an email announcing the dissolution shortly before a scheduled meeting. The company said it was reevaluating how to bring external insights into its product and policy work. See the Reuters report carried by Euronews.

The important part is what happened afterward.

In a transparency response later summarized by Australia’s eSafety Commissioner, X Corp. confirmed that the council had been disbanded in December 2022 and said no replacement external advisory body was in place. The regulator published that information as part of its examination of X’s systems for dealing with online hate. See eSafety’s 2024 transparency report on X.

That is a concrete governance change.

Before the acquisition, outside organizations had a recurring formal advisory structure.

After the acquisition, that structure was gone.

Other community systems are not the same thing

X has emphasized a different kind of community participation through Community Notes, where contributors can add context to posts and ratings determine whether notes are shown. X has described that system as central to its community-led approach to misleading information. See X’s description of Community Notes.

That is genuine user participation, but it performs a different job.

Community Notes evaluates context around individual posts. The former Trust & Safety Council advised the company about policy, product design, rights, safety systems, and broader governance questions.

Replacing one with the other would be like replacing a building inspector with a suggestion box. Both collect outside input. They do not exercise the same function.

The lesson is not that Twitter users once controlled the company. They did not.

It is that platform governance can weaken without any dramatic shutdown. The website remains online. The accounts remain. The feed still moves.

What disappears is the formal channel through which outsiders previously had a recognized seat near the people making the rules.

Posted on by

Subscription tiers that fragment formerly straightforward service guarantees

A service guarantee sounds simple until it becomes a product matrix.

Does the service include support?

Yes.

How fast?

That depends what you bought.

GitHub’s support structure makes the fragmentation unusually visible because the company publishes the response promises side by side.

Support became another upgrade path

In 2018, GitHub announced an expanded Premium Support program for Business Cloud and Enterprise customers. GitHub described the offering as a response to teams that needed around-the-clock care beyond its standard support. The new Premium and Premium Plus plans included 24/7 web and phone support, priority ticket handling, and guaranteed initial response times of 30 minutes for urgent requests and four hours for high-priority requests. Premium Plus added a named technical support account manager and other services. See GitHub’s 2018 Premium Support announcement.

The modern structure is even clearer.

GitHub Enterprise includes support, but GitHub’s current comparison lists the included Enterprise support level as 24/5 with an initial response of under eight hours for urgent and high-priority cases. Premium and Premium Plus provide 24/7 coverage and formal service-level agreements. Premium guarantees a 30-minute initial response for urgent tickets and four hours for high-priority tickets. Premium Plus adds a named Customer Reliability Engineer, incident management, technical advisory time, and faster normal-priority handling. See GitHub’s current Premium Support comparison.

Same platform.

Different promises about what happens when it breaks.

A response guarantee is not a resolution guarantee

GitHub is careful about an important distinction: its SLA covers the initial response, not the time required to solve the problem.

That is reasonable. A broken repository permission can be fixed quickly. A deep infrastructure bug may require engineering work regardless of how much the customer pays.

But the tier still changes the customer’s position during an incident.

One organization gets shared support during defined hours. Another receives 24/7 priority handling. Another has named personnel who already know the account and can coordinate an incident.

The service itself may have the same outage. The path through the outage is different because support has been packaged separately.

The guarantee becomes part of the upsell

There is nothing inherently dishonest about charging more for expensive human support. Dedicated engineers and 24-hour staffing cost money.

The Dead Internet Theory angle appears when a formerly straightforward expectation—I pay for the business service, therefore serious problems get serious support—splits into increasingly specific commercial tiers.

The customer is no longer buying only software capacity or features. The customer is buying a place in the support queue, an escalation path, and a promised response clock.

That changes how the cheapest paid tier should be evaluated.

The important question is not whether the plan technically includes support.

It is what support still means at that tier once faster response, stronger escalation, incident coordination, and dedicated expertise have all become separate things that can be sold above it.