leoc

12 ms·

181.
by leoc 9mo ago
I'm not making some general claim about LLMs being profitable. To be clear, are you claiming with high confidence that LLM training costs will show no meaningful reduction, for some fixed quality, over roughly the next 10 years?
182.
by leoc 9mo ago
He wasn't entirely unworldly though. He didn't like BASIC as a language, but he gave the Apple I and II a BASIC capable of running the programs from Ahl's BASIC Computer Games because that's what the market was demandi
183.
by leoc 9mo ago
Who claimed it was?
184.
by leoc 9mo ago
How good is the book he successfully sold to HarperCollins, The Uni-Verse ? Either he's pretty good, or he was quite lucky, or he had some inside track.
185.
The DBLP computer science bibliography needs financial support(blog.dblp.org)
2 points by leoc 9mo ago|0 comments
186.
by leoc 9mo ago
Taking the example of the cryptocurrency boom (as a whole) as the guide, the problem is the interaction of two realities: big money on the table; and the self-fulfilling-prophecy (not to say Ponzi) dynamic of needing people to keep clapping
187.
by leoc 9mo ago
But why should one expect no future improvement in training costs (all else being equal) from Moore's Law, never mind any future use of eg. more efficient algorithms or more specialised hardware?
188.
by leoc 9mo ago
Even if something like Gas Town isn't remotely affordable today it could potentially be a useful glimpse at what can be done in principle and what might be affordable in, say, 10 years. There's a long history of this in computing,
189.
by leoc 9mo ago
https://www.youtube.com/watch?v=15A0F5aOoPM
190.
by leoc 9mo ago
Windows 10/11’s clipboard stack lets you pin selections into the clipboard, so — and a variety of other characters live in mine. And on iOS you just hold down -, of course.
191.
by leoc 9mo ago
While it's narrowly true that CPU instruction sets generally don't have a null-pointer concept, I'm not sure how important that is: the null pointer seems to have been (I don't know enough to be sure) a well-established
192.
by leoc 9mo ago
In another way Trump is actually rather like Adams himself: his one great talent is as an entertainer and self-publicist, but he feels that he deserves success in business and leadership, so that he can be hailed as a great builder and deci
193.
by leoc 9mo ago
Probably the advantages of its location outweigh the extra costs for the IA. Having your datacentre sited on land and in a building you own, behind a non-shared front door, has legal advantages similar to the ones which drive organisations
194.
by leoc 9mo ago
Even if Apple isn't very good at reciprocating faithful service from its suppliers, there's also the matter of how it treats suppliers who cause it problems instead.
195.
by leoc 9mo ago
sigh Why am I never where the excitement is?
196.
by leoc 9mo ago
Don’t ever forget what Bezos chooses to do or not to do here, whenever his name comes up in any context in future.
197.
by leoc 9mo ago
Having clear insight in some areas and big blind spots (or worse) in others isn't just typical, it's basically all but universal (if we leave aside people who have no particular insight into anything).
198.
by leoc 9mo ago
(Harris was our head of government (ie. PM or equivalent) rather than head of state (which would be the president).)
199.
by leoc 9mo ago
For the past 50+ years there really has been a somewhat significant and quite influential body of people who genuinely want to preserve the planet’s ecosystem even at the expense of the people living on it.
200.
by leoc 9mo ago
Right, though it's also reasonably clear that part of the story there is that he finds a high-stakes AI race personally interesting and exciting on a technical and a business level. Conversely it's also fairly clear that he finds
201.
by leoc 9mo ago
Brin wasn’t bothered to wield any of that power to try to arrest the decline in Google’s search quality. He wasn’t bothered to direct the Chrome team to support MathML, or to bring back Google Reader, or do anything about a hundred small in
202.
by leoc 9mo ago
Like most of the things Spolsky says in that article it’s pretty dubious. Following it to its logical conclusion, presumably on-call debugging work be even easier if the software had been handwritten in assembler.
203.
by leoc 9mo ago
Quite certainly not on v4 Unix, which apparently had few other sites and would not have been in the running for use as a serious university timesharing system. However Gates and Allen had already been among the secondary-school students who
204.
by leoc 9mo ago
At the risk of perhaps stating the obvious, there appears to be a whiff of aggression from this article. The "fighting fire with fire" language, the "haha, we love old FakeFoster, going to have to see if we change that"
205.
by leoc 9mo ago
I'm not an expert, but a quick look at https://en.wikipedia.org/wiki/Buffer_overflow#History suggests that some people, at least, had figured it out by 1972 https://apps.dtic.mil/sti/citations
206.
by leoc 9mo ago
(As others have said, it's probably better to choose the doi.org URL over the dl.acm.org one in general. Three cheers for HOPL!)
207.
by leoc 9mo ago
The HOPL (1) book (ISBN 9780127450407 ) is at https://doi.org/10.1145/800025 (direct link to ACM: https://dl.acm.org/doi/book/10.1145/800025 ) .
208.
by leoc 9mo ago
There’s just a lot more space when you can move in three dimensions, so I don’t think the congestion limitations of non-flying cars are likely to be replicated. IIUC (I’m no expert) that’s one of the most attractive features of flying VTOL
209.
by leoc 10mo ago
For good measure you might as well see what the young Knuth looked like, too: https://www.youtube.com/watch?v=dhh8Ao4yweQ .
210.
by leoc 10mo ago
Thanks for that. They seem to have been embarrassed into putting the stanford.edu page back up for another few months. I think my first encounter with Stanford's website archival policies was when I found that they'd shot an old

More ›