# ZERO

All symbolic content, words, images,
videos, code, will likely double by the
end of the year.

In fact, "doubling" is not even the
right concept, it is like saying π's
digits are going to double. The stream
of symbols is going to be infinite. It
already is.

Who can index this? Who can search this?

What does it even mean to `search` it?

Why would you search tokens?
  
  The web is dead.

At least as the "information
superhighway", that probably never was,
but people hoped it would be.

What is coming is a lens, a language
model lens that allows you to browse the
stream of infinite tokens. Be it
ChatGPT, Anthropic, Google or
Perplexity, is not important.

I see tokens on hackernews, on reddit,
on x, on linkedin.

Its not even the end of the year yet.
    
I honestly don't give a fuck, the sooner
the modern web dies the better.

At the moment it costs like 5k$ to
Setup local wikipedia with relatively
big gemma and just disconnect from this
fucking swamp.

So, I suggest you look for some cheap
3060s and nvmes and you also say:

  fuck this shit.

---

You can get the english only wikipedia
zim from kiwix:

https://download.kiwix.org/zim/wikipedia/

Use llamacpp with nomic text embed to
get good embeddings and facebook's faiss
to build an index and search them. It
will surprise you how small it is.

If you dont how to do those things just
ask claude to write you the programs.

It is almost good enough to do it in one
prompt.