Radim

26 ms·

211.
by Radim 6y ago
I'll put it bluntly: fixating on an average ("average IFR") of a wildly heavy-tailed distribution (e.g. exponential for IFR-vs-age) is idiotic. Technically yes, such average exists – the population is finite. But taking a
212.
by Radim 6y ago
I compared Covid's estimated IFR of < 0.23% to flu's < 0.07% [1]. But that's exactly the discussion I called out as not terribly relevant : The variance within Covid (due to age, location, comorbidities…) is much gre
213.
by Radim 6y ago
And don't forget scientists have continued to gather and evaluate data since. We now have a much clearer (albeit still imperfect) picture of the disease and its behaviour. For example, COVID's current "death rate" estima
214.
by Radim 6y ago
> you might notice by reading my comment I already have more than enough experience making said sausage. I noticed nothing of the sort. No need to lash out – you probably meant to describe your experience but then forgot. I'd be cur
215.
by Radim 6y ago
You forgot the father of scientist-nutcase-weirdos, Isaac Newton himself! Perhaps there's a different lesson here: at some level of intelligence, you pierce through the patterns of stability your fellow men had weaved for their surviva
216.
by Radim 6y ago
As someone who came from academia, now runs a commercial company, and having participated in several large consortia projects (including Horizon 2020), I find your perspective… charming. I hope you never get to experience how the sausage is
217.
by Radim 6y ago
Why walk inside though? All my complex architectures and systems were also conceived while pacing, so I hear you, but I never considered walking in circles in a room . Now I'm curious – is that better than walking outside? Why not wal
218.
by Radim 6y ago
Would never work without a powerful AI engine in the middle.
219.
by Radim 6y ago
I continued reading, what a rabbit hole. Apparently the cause of MS Estonia's sinking, as documented in that Atlantic article, is bogus. New evidence points to a submarine collision and tampering of evidence in the initial 1994 investi
220.
by Radim 6y ago
Chilling quote: > Thiger heard a clear comment from only one passenger, a man nearby, who joked, "Ha! Now we have sailed against an iceberg!" and took another gulp of beer. The singing continued unabated. Surely, this couldn&
221.
by Radim 6y ago
OK. What are you goal posts though? :) Here's my (hopefully charitable) reading of your argument: 1. Alternative theories of history are disingenious (sic) snake oil, because they're based on weak circumstantial evidence. They use
222.
by Radim 6y ago
Your post is seriously meta: To demonstrate how snake-oil people leap to conclusions and paint with a broad brush based on flimsy evidence… …you leap to conclusions and paint with a broad brush based on anecdata. > When I saw they had t
223.
by Radim 6y ago
> Regexps are not great in general …not great for sensitive data detection in general, of course. Hopefully clear from the context (ha!).
224.
by Radim 6y ago
Regexps are not great in general – whether something is sensitive or not is highly contextual. Regexps tend to be too narrow and produce a lot of noise = false positives , so not terribly actionable. And false negatives (missed info) are
225.
by Radim 6y ago
> There is a lot of machine learning snake oil out there, even in the most prestigious journals. I'm also deep in that industry, and can confirm this is the case. Even for research coming out of the biggest names: FB, Google, etc. T
226.
by Radim 6y ago
> They disappear when the surplus disappears, which is not good for the common people either. I contest that. The disappearance of this "surplus" (and if you follow history, you know where such surplus typically went – not the
227.
by Radim 6y ago
That's literally what the article says! I suggest reading it, it's fairly short ;) How similar (~similarly historically biased) is the story of bronze-age-collapse Dark Ages though? That one is based primarily on the even flimsie
228.
by Radim 6y ago
One reference is not like the other. Take your pick: > "However, according to medical historians, the decline in deaths from influenza in San Francisco can be partly attributed to the mandatory mask-wearing policies." From a
229.
Gensim 4.0 beta: faster and leaner fastText, word2vec, doc2vec(github.com)
3 points by Radim 6y ago|0 comments
230.
by Radim 6y ago
Last I checked with Tidelift, they target primarily web OSS that's truly ubiquitous – think vue.js, Angular, Python requests and similar. Their $$$ reward is based on installation numbers across Tidelift's subscriber base, not val
231.
by Radim 6y ago
> in democratic countries when you take actions that result in severe economic consequences, people will just vote in a govt that reverses it If you're the type of person who believes they're saving humanity, you naturally do
232.
by Radim 6y ago
One interesting link buried in the middle of that thread: the drama of Kenneth Reitz "fund raisers": https://vorpus.org/blog/why-im-not-collaborating-with-kennet...
233.
Proposal to make CPython 5x faster(mail.python.org)
2 points by Radim 6y ago|2 comments
234.
by Radim 6y ago
> discussions in the Python community in the last weeks of ways to speed up CPython This piqued my interest. I found this, by Mark Shannon: https://mail.python.org/archives/list/python-dev@python.org/... I
235.
by Radim 6y ago
Your lack of punctuation and odd syntax makes me wonder, but if I understood your post correctly, you claim that building with AWS is somehow safer / more robust / more future proof than building with OVH? A technical judgement? I
236.
by Radim 6y ago
1) Are you doing something innovative? By definition there's no clear value to baseline against. Clients don't know what's even possible, much less how much $$$ it might bring them. So the value in "value-based" i
237.
Radix Sort Revisited (2000)(codercorner.com)
3 points by Radim 6y ago|0 comments
238.
by Radim 6y ago
My favourite: "You cannot sort floats faster than O(N logN)! It's been PROVEN!" True story – I've heard this claim even from a CS PhD (!!). This is what happens when you forget your assumptions and muddle your terminol
239.
by Radim 6y ago
The results were predictably "AI" = complete rubbish. Even though I'm a white male. Our brave new world of "computer said so"… Great presentation though, kudos!
240.
by Radim 6y ago
This. I remember an anecdote by Ota Ulč, a Czech writer who escaped communism to emigrate into the USA. When asked to select his race, Ota didn't realize "Caucasian" applied to him – after all he's Czech, not from Caucas

More ›