jmmcd

18 ms·

181.
by jmmcd 6y ago
> Logic isn't worth much without plugins A lot of people do say that, including a lot of professionals, but in my opinion Logic with its stock plugins is already absolutely great. Some people buy a lot of plugins because they don&#x
182.
by jmmcd 6y ago
We've had these discussions many times. Of course it makes sense to use libraries. For me, what stood out in this submission was that none of the 100 lines are talking about markup. Thus what we have here is 100 lines of front-end and
183.
by jmmcd 6y ago
I think superoptimisation refers to it doing a type of program optimisation beyond what an optimising compiler would do. GP is certainly on-topic here! People are using GP to do this kind of thing too. But I think this system is not using G
184.
by jmmcd 6y ago
This is exactly why third-level teaching is so much easier.
185.
by jmmcd 7y ago
My "white whale of the internet" is a 4-track EP from mid-1990s by a band called The Mark Kramer Band (not that Mark Kramer) with track titles Dust Bowl Rain and Wild Prairie Dog . It sounded like a bluesier Jeff Buckley, I t
186.
by jmmcd 7y ago
You might not be familiar with how neural networks work. When training they do use a lot of computing power. But when running they don't. Yes, they still require some external boilerplate code to multiply the matrices, but you already
187.
by jmmcd 7y ago
Not really similar. Nutonian did straight-up genetic programming symbolic regression. This does genetic programming to discover ML algorithms.
188.
by jmmcd 7y ago
Even if they killed it, it wasn't accidental!
189.
by jmmcd 7y ago
Some more related research here https://www.sfi.ie/research-news/stories/ai/autonomous-lifeg...
190.
by jmmcd 7y ago
Wait, what does mkcd do?
191.
by jmmcd 7y ago
I think the point about an "obvious" demand for a product is important, but what is obvious changes over time. After the industrial revolution it was obvious that certain improvements could make a lot of money in industry, but tha
192.
by jmmcd 7y ago
You might enjoy reading the article.
193.
by jmmcd 7y ago
Isn't that exactly what are doing in your previous comment?
194.
by jmmcd 7y ago
> I wonder if we shouldn't group kids by size for sporting activities instead of by age. This is how it's done in rugby in New Zealand and maybe a few other countries, but not most. It's much more important in rugby than i
195.
by jmmcd 7y ago
If you don't know what the background success rate is, then why do you need to know what this commenter's success rate is? All you need to know is that it's significantly higher.
196.
by jmmcd 7y ago
Well, all models are wrong, but some are useful. I think L-systems and similar are useful for giving a basic conceptual understanding of branching and recursive processes which organisms do use, even if reality is more complex.
197.
by jmmcd 7y ago
I object to this paper title. The idea of unlearning is already well-known, given there are benchmarks and previous results mentioned in the abstract. This paper introduces a new model and gets good performance but doesn't deserve to b
198.
by jmmcd 7y ago
That's the point of the story - Stallone didn't realise it had been turned up, but something felt a little punchier to him (sorry). It's a well-known effect.
199.
by jmmcd 7y ago
Maybe they did and it left no trace noticeable to us, or their policy is not to contact life-bearing planets, and did you read the article?
200.
by jmmcd 7y ago
> why go through all that trouble if there are heaps of good psychology books out there that aren't plagued with errors as this one? All of psychology has suffered in the replication crisis, but my understanding is that Kahneman &am
201.
by jmmcd 7y ago
Fridman's podcast has to be among the highest concentrations of high-calibre guests ever assembled in this area. Week after week it's another huge name.
202.
by jmmcd 7y ago
> Borges isn't comparing a good act and an evil one. He is a bit -- at the end of the article there is a suggestion that they two acts cancel each other out, and we can agree that the burning of books is an evil one.
203.
by jmmcd 7y ago
"doesn't handle change robustly" is a good point. Doctorow doesn't just lack a good headline, he fails to really make this point. Data bias and the context-free nature of text auto-suggest are worth discussing, though.
204.
by jmmcd 7y ago
They can feel or imagine an "ought", but the point of Hume is that they can't argue for it.
205.
by jmmcd 7y ago
This was one of the stupidest parts of a fairly weak essay. Of course you can distinguish a wink from a twitch without asking, otherwise we wouldn't use winks for surreptitious communication.
206.
by jmmcd 7y ago
Also quite a bad example of tech journalism in the really weird writing - repeating a couple of phrases multiple times verbatim, like it was written a bot or is SEO optimised or something.
207.
by jmmcd 7y ago
Please please please, just keep telling Sundar and everyone, "don't be evil".
208.
by jmmcd 7y ago
I can paraphrase the vast majority of answers as "why don't you just remove the word 'just'?" The real issue is that if you don't know how to smile and be friendly and respectful in your body language, minor
209.
by jmmcd 7y ago
Not a chance, I reckon. My evidence is just the whole setup/scenario on the blog post, and the fact that Ctrl-F "smile" only gives one result.
210.
by jmmcd 7y ago
`pexpect` is used by Jupyter Notebook, and gets some maintainence from the Jupyter people. So, like the article argues, it is likely to be stable for the long term.

More ›