Tag: Online Activity

Posted on by

Language switching and human activity missed by English-centered observers

A community can disappear from your search box without disappearing from the internet.

Sometimes all that changed was the language.

Multilingual users do not necessarily choose one language and remain there permanently. They switch according to audience, topic, identity, geography, platform, and who is expected to be listening.

That matters because internet observers often measure activity in the language they personally understand.

If the English-language portion of a community grows quiet while Spanish, Arabic, Japanese, Hindi, French, Indonesian, or another language becomes more active, the English-speaking observer can experience a very convincing illusion of decline.

Research on multilingual social media shows that language choice is not merely translation. It can signal identity and group membership. A 2019 study of football communities on Facebook examined users in Cameroon and Spain and found that language preference and language mixing helped create in-group identity. See Managing identity in football communities on Facebook.

The same people can move between linguistic rooms

Imagine a technical hobby with an international membership.

Early documentation may have accumulated in English because that was the shared language of a particular forum. Years later, local communities become large enough to sustain their own chats, videos, groups, and tutorials.

An English-only audit might record fewer new discussions and conclude that interest declined.

A multilingual audit could find the opposite.

The participants may be publishing more than before, simply for audiences closer to home.

Language switching also happens inside individual conversations. Users may write a public announcement in one language, joke with friends in another, and discuss a technical term using whichever language has the most established vocabulary.

Counting only one layer misses the rest.

Search terms carry cultural assumptions

Even when translation tools exist, literal translation may not recover the same conversation.

Communities develop abbreviations, slang, local product names, transliterated words, hashtags, nicknames, and technical vocabulary. A researcher who translates one English query into another language may still miss the phrases actual participants use.

That is why language-aware sampling requires more than changing the interface language.

Researchers need native or community-informed search terms, multiple platforms, and ideally people who understand how the group describes itself.

This is different from search engines providing unequal language coverage. That is a retrieval problem.

Here the problem is the observer.

The people may be visible, active, and public.

We simply looked for them in the wrong language.

Posted on by

Bots replying to other bots in public comment threads

A comment thread with fifty replies looks busier than a comment thread with five.

That visual shortcut works only if “reply” is being used as a rough proxy for human attention. Once automated accounts can trigger other automated accounts, the relationship breaks down.

A bot can post. Another bot can detect the post and answer. The first bot can react to the answer. A moderation bot can add a warning. A link-preview bot can fetch the URL. A promotional account can repost the exchange somewhere else. The platform records activity at every step even if no human is watching in real time.

This is not merely theoretical.

A 2023 study in EPJ Data Science examined an experimental ecosystem of six social bots on Twitter. The researchers explicitly tracked bot-bot, bot-human, and human-bot interactions. They even used a mediator to manage timing partly because the bots could otherwise interact with one another simultaneously and risk violating platform spam rules. See Emergent local structures in an ecosystem of social bots and humans on Twitter.

The experiment found bot-to-bot exchanges were more mechanistic and less diverse than interactions involving humans. That distinction is exactly what raw activity counts hide.

A reply is an event, not proof of a person

To establish that a public exchange is genuinely bot-to-bot, researchers need more than the conversation looking repetitive.

Useful evidence can include known account ownership, source code, API behavior, posting schedules, platform labels, controlled experiments, or technical logs showing that automated systems generated both sides. Behavioral classifiers can help identify likely automation, but probability is not the same thing as proof.

That matters because humans repeat themselves too. People schedule posts. Communities use templates. Customer-service workers paste canned responses. An odd conversation is not automatically a machine conversation.

Volume can exceed attention

The deeper Dead Internet Theory question is what happens when machines generate activity primarily in response to other machines.

A thousand replies might represent a thousand people. They might represent one human operator controlling several tools. They might represent a handful of automated agents bouncing triggers around a network. The visible counter alone cannot tell us which population exists underneath it.

This does not mean public comment sections are secretly composed mostly of bots. That is a much larger claim requiring platform-specific evidence.

It means something narrower and measurable: online conversation volume and human participation are no longer interchangeable quantities.

The internet can be busy without anybody being particularly busy using it.

Posted on by

Human-operated automation and the limits of a human-or-bot binary

A surprising amount of online activity lives in the awkward middle between “human” and “bot.”

A person writes ten posts on Sunday and schedules them for the week. A script watches an RSS feed and publishes links automatically, but the owner personally answers every reply. A customer-service system drafts responses that employees approve. A social account reposts material by rule until a human steps in during breaking news.

Which of those is the bot?

The binary starts failing as soon as humans use automation as a tool rather than surrendering the account completely.

Automation comes in degrees

At one end is ordinary manual use: a person decides what to say and presses the button each time.

Then come scheduling tools, templates, macros, automatic cross-posting, feed-driven publishing, generated drafts, moderation assistants, and scripts that perform narrow actions. Human control can remain substantial even while a large percentage of visible events are technically generated by software.

At the other end are autonomous systems that select content, generate text, decide when to publish, interact with users, and continue operating without routine human approval.

Those are different arrangements, but a detector watching timestamps and posting patterns may compress them into the same label.

Researchers have used the word cyborg for this middle category. A 2024 study, “Cyborgs for strategic communication on social media”, analyzed more than 3.1 million Twitter users from datasets related to the 2020 coronavirus pandemic and U.S. election. The authors described cyborg accounts as hybrids combining automated scripts with manual participation and identified accounts whose bot/human classifications changed across time windows.

The important point is not the exact number of cyborgs in that study. It is that mixed operation is measurable enough to deserve its own category.

A human may operate something that behaves like a bot

Consider a one-person news service. Software monitors twenty feeds, posts headlines automatically, and alerts the owner when someone replies. The owner then reads the conversation and responds personally.

Calling the account fully human ignores the automated publishing. Calling it fully bot ignores the human editorial decisions and conversation.

The same ambiguity appears with AI assistance. If a person asks a model for a draft, rewrites half of it, checks the facts, and publishes under their own name, the final text has machine involvement without being autonomously authored. If the system generates and publishes 5,000 posts without review, that is a very different production process.

Better labels describe control

Useful analysis can ask who selects the objective, who creates the content, who decides when to publish, whether a human reviews outputs, and whether the system can act independently after launch.

Those questions produce a spectrum of control rather than a theatrical courtroom verdict: HUMAN or BOT.

That matters for Dead Internet Theory because synthetic activity does not need to replace humans completely. Humans can multiply themselves through automation. One operator can create the visible output that once required a staff. A team can supervise hundreds of automated identities. A normal user can automate repetitive tasks without attempting to deceive anybody.

The future internet may be difficult to count precisely because the important unit is no longer “person or machine.”

Increasingly, it is person with machine.