Jake

NFTs are just tokens with some metadata attached. No different than your bsky DID. There *could* be a market for your DID. But if there isn’t, it’s not because of the tech behind your DID, it’s because no one wants to own your DID.

If you’re the bsky CEO… there’s probably a market for your DID.

And what if there is no double coincidence of wants?

The reason why “following the herd” (market liquidity) works is because you don’t have to hope that both parties in the exchange have the exact amounts of some asset that each counterparty wants.

en.wikipedia.org/wiki/Coincid...

Yes exactly. An NFT price might have gone from $5000 to a year long stable price of $100 on the chart… but if there’s no one trading it, no actual depth of bids at $100… you’re not getting $100… if the next bid is $1 then you’re getting $1 for it regardless of what the stabilized price looks like

Earlier you said “maximally immune to market sentiment” which makes me assume the market doesn’t actually care about that.

Either the market cares and thus trades it or it has no value.

If I’m wrong then please give an example of an asset that doesn’t trade that I can sell with zero price impact

But who would be the buyer?

I have gold, you have a token you just generated which deletes itself after it moves… why would I give you my gold for that token?

But to buy something… eggs, wood, metal, atoms… there has to be someone on the other side of the trade.

If you have a “long term store of value” and there’s no trading activity… then there is no buyer to take the store of value.

Think about it. Imagine you own your favorite store of value. But it has no market activity, no bid/ask, zero vol. if you want out, who is the buyer?

You’ll find zero assets with market depth that are stable or lacking in any sort of vol.

Something that is priced flat forever likely has no bidders if someone tries to offer out of it.

Incorrect. Imagine for a second you have some store of value called X (replace with your favorite store of value.) Imagine the market sentiment for X is anemic. There’s no market activity, no liquidity, no speculation. You decide to exit the store of value. What price will you get for it?

A currency that disappears after a single use is effectively the same as a currency printed to infinity after a single use. It’s worthless because it loses all attributes of a good money. It becomes about as valuable as the Zimbabwe trillion dollar bill.

That is to say no one cares what the central authority says the rate of exchange is for a currency. The market trading it’s top of book liquidity will determine its worth.

If a central authority says “this silver coin is worth $1” but the market will buy it for $100, then it’s worth $100.

A currencies value depends on the market rate for that currency.

If a central authority says “one unit of this currency is worth a dozen eggs” but the egg farmers will only give you half a dozen eggs for it, then the currency is worth half a dozen eggs.

The energy crisis hits a bubble for ants

I actually don’t think people lean into the idea of LLMs being mythical supernatural enough.

Wait so if you energize a trillion floating point numbers with gradient descent and backpropagation you get a neural net & circuits to do X and no one has full interpretability of its inner layers?

🧙‍♂️

The posit that "its just a bunch of computing" is unironically close to the actual understanding we have, which is to say we don't really have much. "Sometimes LLMs do math with something like trigonometry" cool, but like how did it derive the circuits to do that?

Your claim was we can explain how claude works with very precise detail, but every researcher has found various limitation in frontier interpretability of the internal structures of LLMs:
- arxiv.org/html/2407.02...
- www.lesswrong.com/posts/X26ksz...
- aclanthology.org/2024.eacl-tu...

If you're actually able to explain how neural nets program themselves through gradient descent and backpropagation with a validated, layer-by-layer mechanistic model, you'll be making a significant scientific contribution and up for several awards.

unfortunately "bunch of computing" does do not actually paradigmatically explain the internal mechanistic layers of even small neural networks never mind how billions to trillions of DL neurons produce token predictions

We actually can't explain in very precise detail how claude works.

We can't even in explain what happens in a single layer of an open source model.

Try it. Explain what happens, with very precise detail, in layer 17 of qwen-2.5-7B or llama 3, llama 4, etc

“Not solving complex problems”

Water-Thirsty Golf Courses Need to Go Green There are about 16,000 golf courses in the United States, and they all need huge amounts of water. The sport must take notice of limited resources and develop courses that are more in harmony with the...

That’s crazy. If these numbers are right that means this one data center will use 5.4x more water per day than the avg American golf course
www.npr.org/2008/06/11/9...

This was op who blocked me

Oh no I made a cognitive dissonance

There is no punk rock trains. To be into trains is to wed yourself to the worst 19th century robber barons in the world. And I think this makes train people crazy. You just can’t be into it and not be a shill for a rail tycoon. You can’t be into trains and be cool. And it makes them all bitter weird

Failure of imagination is working 9am-7pm doing soulless work for somebody else when you can have your intellectual capabilities leveraged through computers.

This is like a 6th grader sitting in a college level epistemology class

> posted to a decentralized social network built on self-authenticating merkle search trees that uses nonfungible tokens (DIDs) for identity

yeah whatever happened to the blockchain

Might be able to catch me at hereticon