Miss the days when I could mansplain Jevon's Paradox
Gunson
gunson@primal.net
npub1pn9x...2xn0
Low status fiat heretic. Often wrong. 2 + 2 = 4
This was peak general knowledge accumulation


Got a subscription for ElevenReader since it was by far the best TTS service and could parse all my docs and links.
But actually never used it. I much prefer listening to something if it's natively audio, and it's nice to have a backlog of things to read (and most of these things have images too).
The exception is a well narrated audio book, so perhaps the reason I still didn't use the 11R app is because the AI gap is still too big, even if it's still quite impressive - my brain just disengages.
Have the AI labs considered?:
Instead of giving their models more knowledge and testing their performance ...
... what if they rather let them do whatever they want and just give them participation certificates for every output?
Buy a HWW with lightning, use a fresh email and fake name.
3 days later: FedEx has your parcel and needs you to pay custom duties with a credit card.
FedEx and DHL probably have one of the most comprehensive Bitcoiner databases.
Think clearly, speak freely.
Me: Hmmm I wonder if the Coinbase self custody wallet could be a reasonable suggestion to a noob who wants the comfort of a big brand.
*Checks wallet dot Coinbase dot com*
Wallet: Fresh drop! Lucky Cat Zombie NFT!
๐ฌ
I often try to think about business ideas, and then struggle to land on anything that others aren't already doing.
But then I watch a TV show on Channel 4 and there are ads for 7 different online gambling apps.
Uniqueness is clearly not necessary for being an entrepreneur!
Twelve unread articles and newsletters in my browser, but keep checking my email for new ones ๐
Many don't know that
E[(Y โ fฬ(X))ยฒ] = Var(ฮต)
and it shows.
(The difference between your prediction and the true data generating function in real life is some irreducible entropy you can never know, no matter how good your model or AI is)
I heard there's a Saylor podcast with the DOAC guy (literally such a popular podcast that my mom sometimes recommends it to me).
Listening to Saylor makes my ears bleed - it's like Iago from Aladdin if it was a high temperature AI continuously self-prompting about digital energy.
But for some reason I feel like I need to listen in order to check how bad it was, and to pre-empt normie questions about it.
Chaios monkey
Sometimes I imagine how easy and comforting life would be if I just answered the Palantir recruiter on LinkedIn, put all my savings into an index fund, and started caring about sportsball. But then I realise how much I'd hate myself, and make myself buy more Bitcoin instead.
Current (forever?) human advantage vs. AI:
- We want things (self-prompted)
- We can see the big picture (huge context window)
- Can spot mistakes (robust world model)
- Can be hurt by failure (nuanced reward framework)
I'm completely enthralled, but absolutely exhausted by AI. On net it's added way more stress to keep up with it personally and professionally vs. the nice conveniences (quicker searches mainly).
"It's all AI's fault"
"Our AI's didnt find any vulnerabilities"
Is the Coldcard event even a hack? Is it criminal? They didn't break into anything or trick anyone. They just found a number. Obviously shit to deprive people of their money, but if they didn't someone else would have.
Not endorsing it at all, but I'm cautious that even if the people could be identified there could be a case against them.
Seems we only think we know the total value of CC funds stolen because of a few very large (in terms of number of inputs) transactions?
There's no reason the exploiter wouldn't also be doing some 1 to 1 utxo transactions as a way to ensure it's more difficult to link their stolen funds. So probably we are massively underestimating?
There is always vendor competence risk for any HWW, so I wonder if the lesson is actually "always bring your own entropy" and not "extra super duper don't trust a vendor with an obnoxious CEO".
So, just from this point of principle I'd argue Coinkite deserve credit for enabling dice roll input which many used specifically to protect against device entropy risk.
While they haven't had any issues (well done!) maybe the real lesson is to be pushing other providers to ensure they're mitigating the RNG risk of their own devices - according to Opus 5 these vendors don't have an in-built way to add your own entropy:
- โ Foundation Passport
- โ Bitbox02
- โ Trezor
- โ Ledger
The only other major HWW that has a dice roll input path seems to be โ
SeedSigner.
"This is now the slowest air travel will ever be!"

