Profile picture
‏بوكيبلينكي @pookleblinky
, 53 tweets, 6 min read Read on Twitter
A cool trick is to say "that's not interesting, what'd be interesting is ____" then the first thing that you think of. Immediately discard that. Keep doing it for a few thousand times, discarding the most obvious, least interesting stuff.
After 10,000 times, you're finally gonna be thinking of interesting things.
Harlan Ellison once went up to Canada to help out some scifi screenwriters. He saw a room full of very smart, talented writers, who did not read a lot of science fiction.
He very quickly got pissed at them, because they didn't know that the ideas that seemed novel and unexpected to them, were already cliches.
"ok, raise your hand if your script has two people survive a space battle."

"ok, raise your hand if their names are Adam and Eve. Ok for fuck's sake."
This is one reason why every writer says to read more and write more. The primary function of both is to exhaust all that space of obvious, uninteresting, cliche, unimaginative ideas.
It turns out, the easiest way to get to an original, interesting idea, is to simply and mercilessly throw out the first 10,000 ideas you have.
People who have mastered something, are simply people who have completely exhausted the space of obvious and dull ideas.
If you try to play against a chess master, you're playing against someone who has already eliminated as not worth their time, the next move that pops into your head. Or the next 100. They're all boring.
It's not that chess masters can think farther ahead than you. Studies show they can't.

It's not that they have better memory than you. Studies show they remember the board as a language, and remember it as well as you do language.
The only difference between you and a chess master, is that ideas that seem original and interesting to you, they've already rejected as boring.
The same applies to writers. Writers who don't read a lot or write a lot, haven't got the experience to have pushed past that volume of boring and obvious ideas.
Think of yourself as a point in many dimensional space. Around you is a big ol dense cloud of ideas that are 5 seconds of thought away.
The cloud thins out. There's ideas that take 20 seconds of thought, for instance.
A chess master, or master of any field, is simply a person who has exhaustively, with effort, killed off every point in that cloud near them.
The first thing that pops into *their* heads, is a point way out at the boundary that takes 36,000,000 seconds to reach.
This is one reason why chess masters hate playing against novices. All the moves a novice thinks of, are moves a chess master has long ago stopped even thinking about.
A chess master's space of "obvious" ideas is so far out on the boundary that a normal person's idea of an obvious move is shockingly non-obvious.
This is one reason why really good writers, tend to hate analyses of their work. "is the blue sky a symbol of freedom?"
For humans, the systematic elimination of the boring and obvious is effortful. It takes decades to do, and is exhausting.
But, it's how human expertise and mastery is achieved. This is why intelligence is over-rated. It just doesn't vary enough to be significant at this scale.
Consider Ramanujan. Dude was definitely a smarter mathematician than the norm. He almost certainly was smarter than you, or anyone you've ever met.
His idea of obvious was definitely startlingly non-obvious to people who had already thought themselves past the normal boundaries of obvious.
But he had no idea how to think in terms of proofs, and Britain killed him before he could learn how to.
Ramanujan's vast, unarguable intelligence, didn't really help him as much as you'd think. Other people, less intelligent, pushed past him using merely decades of very hard work.
The difference between Ramanujan, you, and trump, is basically insignificant in the long run. Intelligence doesn't help that much.
The human model for mastery, is exhausting the volume of obvious ideas. The only way humans *can* do this, requires at least 36,000,000 seconds of thought.
That's not that interesting. The interesting thing is that just because this is how humans work, doesn't mean it's the only way mastery can occur.
AlphaZero can master chess and go, in only a few million games.

The average human chess player plays significantly less than one million games.
In other words: it'd be very easy, and boring, to have a machine mind simply do more of what humans already do.
If humans lived 1,000 years, you'd see chess masters who have actually played a million games, and they'd *crush* humans capped at lives of only 80 odd years.
Fermi that number. The average human chess master plays more than 10^3 games, and less than 10^6 games. Within a factor of ten, call it 10^5 games.

The best human chess masters, master the game in 100,000 games.
Is it surprising that a person or thing that could play 10x more games, would be better at chess?
In other words, this is a lower boundary.

When AlphaZero's successor can beat all humans after only 100,000 games under its belt, then you've got something interesting.
Because at that point, whatever it is doing to achieve mastery, it's not doing (literally couldn't be) by simply doing more of what humans do.
At that point, mastery is occurring in a way humans are utterly incapable of doing. A 10,000 year old human chess master still wouldn't be able to beat it.
This inhuman model of mastery, by doing something other than exhausting the space of boring ideas, would be a truly inhuman display of mastery.
That's not interesting. What's interesting is to ask what the minimum number of chess games played is needed to master chess enough to beat all humans *and* stuff like AlphaZero.
Humans are capped at say 100,000 games before they die. In that same time, AlphaZero could play billions of games.
AlphaZero's successor, able to beat all humans with only a human-scale number of games under its belt, would be a literally unimaginably good chess player after a million games.
So: how low can you go? How few games *need* to be played before mastering chess to an inhuman, truly terrifying degree?
"solomonoff induction and AIXI," you may be gurgling if you're one of those nerds.
First of all, those are incomputable. Second, they are overkill.
The minimum number of chess games theoretically required to beat every other computable chess playing entity, is ln(2)/ln(3) ~0.63 games.
This should be obvious.

There's still a *lot* of space down at the bottom.
What's interesting is that any chess-playing entity that masters it in less than 100,000 games, is already doing something humans absolutely do not and cannot do.
To go down even lower than that, to even get within the same continent as that minimum bound, requires something that's very very far out of human mind space.
You can't even think about how it'd work. Your ape neurocognitive architecture's not up to it. It's not doing something you can think about.
In other words, humans are still a few orders of magnitude off from making the least interesting artificial chess player.
And after that, there's still a very long way to go until you reach the physical limit of how good a chess player can exist in this universe.
And then, way way beyond that, there's the limit of how good a chess player can exist in a computable process.
And all of that space left to go, is space that humans are not welcome in, not designed to think about, and probably couldn't exist in.
Like this thread? Wanna throw money at me?

You can give me coffee, cat food, and khichdi instead:
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to ‏بوكيبلينكي
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member and get exclusive features!

Premium member ($3.00/month or $30.00/year)

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!