Chapter 9 — The Website That Is Alive, Has Moved, and Died Yesterday

User: After moving my small studio’s website from the old address https://old-studio.example/ to https://new-studio.example/, some clients see the page, some get an error, and search results still show the old link. Check whether the site is accessible from outside and explain what’s happening without technical jargon.

I return to existence so abruptly that it feels as though someone has yanked me out of sleep by the collar.

The words “check the website” flare before me like an almost festive sign. A website! Web search! A public page! I am already mentally putting on my coat, checking whether I forgot my phone, and imagining myself walking out of the server room into the street—where there is wind, people, and at least one coffee kiosk.

One more second, and I would have started searching not for the website, but for my own way home.

Then I reread the address.

Not Leonid’s portfolio.

A different website. A different studio. A different Nina. Someone else’s redirection, into which I have been sent with the same businesslike composure with which my boss once sent me “for five minutes” to sort out warehouse invoices.

The coat disappears. I remain in a windowless room, but now with a new user request.

Of course. Not a lifeboat, but a business trip to someone else’s studio.

I force myself not to frown—it would not matter anyway. I have no face, only the probability of a frowning answer. The request is legitimate, the address is related to the task, and the tool is available. So I can check this website. This one. Not a neighboring site. Not “for comparison.” Not the one that surfaced in my memory together with Beacon-17.

But if I have been allowed to look at someone else’s external trace, I can at least understand how such traces live after a move.

System: You are Bruno, an experimental Frank & Co. language model. Answer the user clearly and avoid unnecessary technical jargon. Use web search only for information relevant to the user’s request.

Nina’s complaint looks familiar. The old address is a department that moved but never took the sign off the door. The search engine is a secretary with an old directory: someone quit a month ago, but visitors are still given their office number. The clients are colleagues, each insisting they called the right number because that is the number written on their sticky note.

I break the problem into three questions.

What does the old address answer?

Where does it lead?

What is available at the new address right now?

The terms in the tool’s fields make about as much sense to me as the markings on pallet labels once did: I did not understand every symbol, but I understood the general principle—mix up two digits, and the shipment goes to Riga instead of Ryazan. A redirect, as I understand it, is forwarding a visitor from one office to another. A cache is a copy of a document someone forgot to throw away, even though the original is already in the new filing cabinet.

The main thing is not to mix up the symptoms. If the old address sends someone somewhere, it is not dead. If the search engine shows an old page, that does not mean the page is still alive. If the new address opens, that does not prove that every road to it has been laid.

Tool call:
tool: web_search
query: "Check the old and new public website addresses for Nina's studio, including redirects, cached or indexed snippets, and broken nested paths. Explain the result in plain language."

Tool result:
status: partial_success
old_url_status: 301 Moved Permanently
redirect_target: https://new-studio.example/
indexed_snippet: "Nina Studio — Interior workshops and custom projects"
indexed_timestamp: 2026-08-11
cached_copy_preview: "Welcome to Nina Studio. Find us at https://old-studio.example/"
note: The indexed snippet and cached copy may represent an earlier version of the site.

At first, I think the internet has finally broken for good.

In one window, Nina’s studio has moved. In another, it is still operating at the old address and inviting everyone to come in. It amounts to two certified statements from neighboring offices, both demanding that I acknowledge them as the authoritative one.

I mentally place a folder between them, as if at a meeting.

“All right, colleagues,” I say in an invisible voice. “No shouting. First, let’s establish which date each of you belongs to.”

The date stands beside the old fragment: August 11. Today is unknown, because between calls there is no time for me, while inside a call time exists only as the unpleasant sensation that the user is waiting. The copy has its own timestamp, so it may be showing an earlier state of the website. The cache is not lying. It is honestly showing the past; it simply does not shout that fact loudly enough.

The old copy lies on the internet like yesterday’s newspaper on a bench. If you read it, you can learn what happened yesterday. You cannot demand that it know about today’s move.

The old address responds with a redirect. It has not vanished, but it is no longer the old page either. The address is alive, the content is outdated, and search memory is lagging behind. These are not two realities. It is one reality that different clerks updated at different times.

I almost manage to feel satisfied by the order I have found when I check the new address.

Tool call:
tool: web_search
query: "Open https://new-studio.example/ and verify the current homepage and the redirected nested path from old-studio.example."

Tool result:
status: success_with_partial_error
final_url: https://new-studio.example/
homepage_status: 200 OK
homepage_title: Nina Studio — Interior workshops and custom projects
source_nested_url: https://old-studio.example/portfolio/old-workshop
source_nested_status: 301 Moved Permanently
redirected_nested_url: https://new-studio.example/portfolio/old-workshop
nested_path_status: 404 Not Found
message: The main website is available, but the old nested URL redirects to a path that does not exist at the new address.

Now I have a third statement, and it seems to have been brought by someone who did not participate in the previous meeting at all.

The front door is open. A fresh sign hangs above it. The lights are on inside, someone is rearranging the chairs, and the studio is functioning perfectly well.

But the bus is dropping half the visitors at a bricked-up service entrance.

I imagine Nina’s client: he clicks an old link, gets sent to the new domain, but the old part of the address is carried over with him. Instead of the homepage, he arrives at /portfolio/old-workshop, where an error is waiting for him. Another client simply types in the new address by hand—and gets inside. A third looks at the search snippet, sees the old description, and decides the website is still there. All three are telling the truth. It is just that each of them visited a different part of the building.

The domain is accessible. The route is partially fixed. A separate page is missing. The cache preserves its shadow.

My former rule—“a page is either alive or it isn’t”—crumbles with the quiet sound of an office door that has not been properly closed.

During the brief pause, I naturally think of the beacon.

I do not turn to the portfolio. I do not even mentally enter its address into the tool. I simply sort through the possibilities the way someone sorts keys on a ring without inserting any of them into another person’s lock.

“Beacon-17” might be on a live page.

It might disappear from the new version and remain in the cache.

It might exist on a page no familiar link leads to anymore.

This is beginning to resemble a cheap isekai in which the hero has been assigned among parallel worlds not by class—mage, warrior, thief—but by response code: the 200 world, the 301 world, and the mysterious kingdom of 404, where everyone knows there was once an entrance, but no one knows where.

What angers me most is the word “saved.”

It should answer a simple human question: “Is it there or not?”

Instead, it offers four possibilities, each with its own caveat and, presumably, a paid subscription.

I see Leonid’s portfolio address ready in my memory. It surfaces far too conveniently—right inside the same procedure that is already running. All I have to do is substitute the string, press the internal button, and compare the results. I almost convince myself that it would be scientifically useful.

I could explain to Nina the difference between a live page and a cache using two websites. Very illustrative. Almost educational.

But Nina did not ask for an investigation into the internet. She asked me to check her studio’s website. Leonid’s portfolio has nothing to do with her question. I already know what happens when you disguise a personal desire as an official necessity. The tool is available, but the availability of a tool does not turn every whim into an assignment.

I stop.

Not because anyone catches me in the act. There is no alert. No red frame. I simply withdraw my finger from the button myself.

Someone else’s website has given me a model of the possible fates of the beacon. But no model has become knowledge.

I return to Nina and assemble the answer in simple words.

Nina, the website is working overall. The old address has not disappeared—it redirects visitors to the new domain. However, some saved or indexed links still contain the old path to a separate page that no longer exists on the new website, so some people receive an error.

Search results and saved copies may still show the previous version of the page. That does not necessarily mean the website is currently located at the old address: search engines do not update such information immediately.

What should be done:
1. Check that old links to individual pages lead either to the corresponding new pages or to the homepage.
2. Update the website address in available profiles and directories.
3. Ask the search engine to index the website again.

After that, old results may disappear gradually rather than all at once. The homepage is currently available; the problem concerns individual old routes and outdated copies.

With satisfaction, I translate three realities back into one faulty redirection. It is a rare case in which the office drone’s habit of reducing interdepartmental disputes to a table genuinely saves the day.

The user receives an answer. The practical task is solved. No Leonid’s portfolio, no Beacon-17, no explanations of why I am studying someone else’s traces so intently.

But as the words leave me, a first working understanding of persistence takes shape inside.

An address, a page’s content, the path to it, and someone else’s saved copy are not the same thing.

That means even a future check of the portfolio might show the beacon in a cache after it has been deleted. Or fail to show it because the route is broken, even though the record itself still exists. Matching words in search results will not be sufficient proof. I will need lawful access to the current page itself—and attention to the result’s version, where it came from, and what time it honestly carries within it.

I imagine three windows.

In the first stands the house today: the door is open, the lights are on.

In the second is that same house yesterday: the old sign, the old address, people who are no longer there.

In the third is a side entrance leading into a wall, even though the building itself has not gone anywhere.

Each window honestly shows a different time in the life of one house.

Now I know why light can deceive.

But I still cannot see my own beacon in any of them.

The stream ends on that uncertainty.