Not because dial-up was great. It wasn’t. Waiting half an hour for a file and discovering at 97% that somebody’s connection died was character-building in the same way food poisoning is character-building.
I miss the internet because it used to feel like it was made out of people.
Personal websites. Weird forums. FTP servers. Somebody’s directory full of files they had been collecting since 1996. Pages maintained by one guy who knew absolutely everything about one obscure subject and apparently nothing about web design.
It was messy.
That was part of what made it useful.
You could follow links for an hour and end up somewhere you didn’t know existed. Not because an algorithm decided you had a 73% probability of engaging with it. Because somebody linked to somebody who linked to somebody else.
There were actual trails.
Now most of the internet feels like five enormous buildings connected by parking lots.
Search. Social media. Shopping. Video. AI.
Everything else still exists, but you have to look harder.
And increasingly, when you do search for something, you aren’t necessarily finding the thing. You’re finding somebody’s summary of somebody else’s summary of a page that may have originally contained the thing.
Sometimes written by a human.
Sometimes not.
There is a thing called the “dead internet theory.” The extreme version says most of the internet is already bots talking to bots, fake engagement, generated material, automated accounts, and manufactured activity.
I don’t buy the whole conspiracy package.
But I understand why the idea has legs.
Spend enough time searching now and you start recognizing the texture.
The same answers rewritten twenty different ways. Sites that exist mostly because somebody discovered a keyword. Articles that take 1,800 words to tell you something an old forum post would have explained in four sentences.
Then AI arrived and handed everybody an industrial machine for producing even more of it.
Yes, I see the irony of writing this in 2026.
The problem isn’t AI by itself. Tools are tools.
The problem is that the web already had too much material written for machines before machines started writing the material.
SEO pages written for Google.
Social posts written for engagement systems.
Product pages written for conversion systems.
Articles written because an analytics dashboard said somebody might click them.
Eventually you start missing the guy with the terrible HTML page who just wanted to explain how to rebuild a carburetor.
He wasn’t optimizing anything.
He knew how to rebuild the carburetor.
That was enough.
A lot of that older internet is gone now.
A surprising amount isn’t.
That’s where CacheRat came from.
I’ve spent a ridiculous amount of my life looking for things online.
Old software. Manuals. documents. Images. Abandoned sites. Databases. Archived discussions. Mirrors. Files sitting under filenames no sane person would think to search for.
Eventually you learn something important:
“Missing” and “gone” aren’t the same thing.
A website dies.
The files get copied somewhere else.
The copy disappears from search.
Somebody mirrors the mirror.
The directory gets renamed.
The documentation vanishes but the ZIP survives.
Twenty years later everybody assumes the thing is gone because Google doesn’t show it on page one.
Sometimes it is gone.
Sometimes it’s sitting three directories deep on a server in Finland.
CacheRat exists for the second kind.
The idea is pretty simple.
I look for useful or interesting material that’s buried, scattered, badly indexed, awkward to use, or otherwise annoying to find.
If the rights allow it, I clean it up where necessary, organize it, document what it is, and put it here.
Sometimes that means an old public-domain document.
Sometimes a pile of files that make more sense as one collection.
Sometimes something scraped or indexed from public information.
Sometimes just a strange artifact worth preserving.
I’m not pretending everything here is secret.
It isn’t.
I’m also not pretending you couldn’t find some of it yourself.
You probably could.
That’s sort of the point.
You could spend Saturday afternoon chasing dead links, checking mirrors, comparing copies, renaming files, removing duplicates, and figuring out which one isn’t broken.
Or I can do it.
You’re buying your Saturday back.
There is another reason I wanted this site to exist.
I want somewhere on the internet that still feels like somebody is home.
Not a content farm.
Not a faceless storefront populated from a product feed.
Not twelve paragraphs of “valuable insights” before you reach the download button.
A human looked at the thing.
A human decided it was worth keeping.
A human wrote the description.
If something is broken, I’ll say it’s broken.
If the scans suck, I’ll say the scans suck.
If the original free source is still available, I have no interest in pretending I discovered Atlantis.
The value has to survive the truth.
That’s the standard.
I don’t know exactly what CacheRat will turn into yet.
That’s deliberate.
The shelves will probably get weird.
Old documents. Technical material. databases. Images. Archives. Strange collections. Things I haven’t thought of yet.
Some will sell.
Some will sit there gathering digital dust.
Some will probably make me wonder why the hell I bothered.
Fine.
That’s how junk stores work.
You build the shelves first.
Then you start bringing things home.