The Ascetic Programmer

Since protein folding is NP-complete, this nobel prize is in reality for showing P=NP 😉 www.brown.edu/Research/Ist...

Einstein's time is reversible so that egg could be uncracked if it were just a matter if gravitation.

I would like to take my profile to a different provider and keep my identity, network, messages etc. Wasn't that the promise of AT proto?

I've read the paper, thanks. It doesn't even mention "credible exit". The abstract talks about "having multiple interoperable providers for every part of the system". I already consider the company to have gone bad, what are my options?

Given the de facto centralization of the service at the 10M user mark, what is the meaning of developing a decentralized protocol to power another centralized service? Are you comfortable being associated with it?

Two takes on CowGPT:
jensorensen.com/2024/09/11/a...
www.thinkwithniche.com/blogs/news/e...

A concise version of the first answer is in the wikipedia entry on BPP, henceforth likely in the training set. Also of note is the continued inability to answer "I don't know", which makes it hard to use AI as a retrieval system.

I am just saying bluesky using a "distributed message" they long abandoned in practice. You are right on which protocol is better designed and which services are most successful, but you are just changing the subject again. I will have to leave it at that.

Forget the (valid) technical observations. No foundational protocol of the internet is controlled by a single entity. Neither HTTP, nor SMTP, nor DNS, not a single one. It's not acceptable on a governance level, independent of technical merits.

I currently consider bluesky distribution a broken promise. There must be limits to how much you can slow-walk a process before you lose any credibility.

That doesn't sound nearly as adventurous as complex bases. en.wikipedia.org/wiki/Quater-...

Fog hovers above a forest. Banding and color shifts are visible in the fog.

Not sure why but on my monitor (mac + firefox) the picture shows banding and color shifts in the fog. I wonder what's going on. Including a screenshot, maybe you can see what I am talking about, maybe not. I love fog pictures!

The article states that the new Microsoft data center will take 5 MW of power or the output of 5 nuclear power plants. The smaller such plant in the US generates over 500 MW.

Who's going to pay the electricity bill once VC money runs out? And what's DAIR going to do during the next AI winter?

Or publication bias. If it isn't a scandal, what do you write about?

The First Neural Networks

The hype for the early (single layer) neural networks was quite extreme, not unlike the current AI frenzy. "They hesitated to call it a machine because it is so much like a 'human being without life'" youtu.be/e5dVSygXbAE?...

Wikipedia maintains a list of long proofs. I think the longest man-made is listed at 10 to 20K pages and the longest machine-assisted is 2 petabytes! (5th Schur number). I don't think it's a proof from The Book. I looked into large and small proofs as part of the research for my book on conciseness.

I suppose Lean's implementation can have bugs like any program. If we add a 30k proof to the record and it is incorrect, how many years before it's caught? We had incorrect proofs of the 4-color thm in the record for 4 to 11 years.

I asked about the difference between COMPOSITE and FACTORIZATION and besides being a bit verbose and repetitive it was quite impressive.

Top disease official: Risk of coronavirus in USA is 'minuscule'; skip mask and wash hands Should I wear a mask? The U.S. infectious disease chief says no: "In the U.S. there is absolutely no reason whatsoever to wear a mask right now."

Which calls into question what people mean by "risk". Is it the probability of an epidemic in the next week, month, year, or decade? Something else? Dr. Fauci said in Feb 2020 that the risk from C19 in US was "minuscule" (USA today). Unless we specify what we mean by risk, it's not helpful.

I wish I had seen this before finishing my book (asceticprogrammer.info). There are many themes in your talk that are touched upon also in it (e.g. the gradual accumulation of complexity or lines of code used as productivity measure) but the most important is simplicity as conciseness.

I posted to my desultory blog the transcript of a talk I gave in 2009 inside Google, mostly because someone recently asked me to, feeling the message needs reinforcing.

commandcenter.blogspot.com/2023/12/simp...

Are you sure it's OK to post multiple reply skeets to the original one? Or is it thread hijacking? Not sure, just asking.

Unlike opinion polls, electoral polls have a moment of truth on election day. They've been close enough to still be a business. If you have a better method, by all means give it a go. For a completely different method check civiqs. Interesting but hasn't taken over.

Please note that preventing a fire start doesn't remove dead wood. It will eventually burn, at least in CA, in a future, bigger fire or in the controlled burns that are not yet being done at a meaningful scale.

For me it's the Ketan Joshi feed. I might as well unsubscribe from test 3 and follow him.

If you do can you please explain to me what Gilead has to do with it?

How did you come to this conclusion?

It's possible but Zeke himself referred me to a data download page rather than a paper that spells out "we expect a warming acceleration after 2010". So I suspect It doesn't exist but happy to be proven wrong.

Unfortunately I couldn't find it. A bit like the drought prediction in CA that non specialists heard about after 5 years of drought and the ark storm paper that came after an exceptional winter of precipitation. It's hard to tell apart true predictions and attempts to make the NYT front page.