myopicgoat

Dh

I think that may be only partially fair: some people appreciate his writings because they question things which, at the time, (and possibly still now) would have been considered “true” or unquestionable (particularly stuff on moral). One may not like it but I think can still benefit from asking similar questions (while ignoring the answers which are, anyway, often somewhat cryptic). I personally enjoyed Thus spake Zarathustra for that reason, not so much the cake one.

You may be referring to p-values, and the arbitrary 0.05 threshold but in that case, the connection is risky at best and wrong in general (common misinterpretation of p-value as probability of alternative being false)

Thsee questions pop up every time Julia comes on HN and have been answered time and again for example by Chris Rackauckas [1], at the end of the day there will always be pros and cons and people unwilling to change from languages they’re proficient at. Personally I’ve been using Julia for scientific computing for 4 years and I’m hooked.

[1] https://thewinnower.com/papers/9323-i-like-julia-because-it-...

In Oxbridge, colleges compete for public exposure because ultimately that’s what leads to donation & to get top undergrads applying who, eventually, will donate £££ to the college once they land a good job. So there’s quite a lot of PR relating to sports & academic achievements. What’s a bit ironic in this case is that no research is really done at colleges, and grads tend not to be central to college life, it’s really the authors & the research lab who deserve the creds!

This is a tricky line of thinking. Once surveillance is there you can always hide behind the “now we’re safer”, the issue is that you can always survey more and always claim that going back would be less safe and so on and so forth. A good example for this is airport security. As far as I’m aware it’s quite unclear how effective all the security apparatus is yet it piled up over the years since 9/11, full body scans are now the norm etc. The question left to ask is: when is it too much? How much intrusion are we ready to bear for how much marginal safety gains?

Last I saw him (last year) he was still a prof at Oxford, pushing for a startup called « invrea » (that does try to use PP [1]). I think UBC was before that.

[1] http://invrea.com/index.php

This is a question of perspective, you can in fact just take it as a regression over the continuous 0-1 interval with loss |y-o(Xw)| where o is the sigmoid and could report that loss (in fact a package such as Sklearn will usually return the parameter values that minimise that loss, possibly with a penalty). If you want to use it as a classifier then you threshold the predictions.

I agree with this though opening up to external scrutiny (up to a reasonable extent similar to - say - CERN) would help dispel a feeling of “yet another faster-than-light type experiment”. I also don’t think that the “combativeness” with which criticisms of the experiment are addressed (cf article) is in any way useful to the scientific endeavour.

No, because that would take full rows of the feature matrix (thereby corresponding to the full information of one individual). The idea here is to “generate” rows corresponding to plausible artificial individuals. That way you can give a third party artificial data to build an ML model without compromising (too much) the privacy of the real individual in the initial data.

Right. No offense but where do you live in China? Beijing or Shanghai is not exactly the same as Xinjiang. Also foreigners (still) enjoy a de facto “upper class” status in China so it’s likely you may be unaware (or choose to be) to much of what happens (e.g. situation of migrants in big cities).

MAC address randomisation has been proven to be essentially useless: https://arxiv.org/abs/1703.02874v1

Seems a bit overreacting dissing the whole of journalism for the odd pun seen in an article. Eg: John Oliver’s late night show is, objectively, quite good journalism (even if he denies it) though it’s not very “serious”.

From 4 years ago... a lot has changed since. Consider this recent post for example: https://tk3369.wordpress.com/2018/02/04/an-updated-analysis-...

The question discussed is interesting and important but I felt the structure of this article is somewhat unfortunate: it explains very well and objectively some of the statistical issues and solutions with censored data but then when discussing how it’s applied to the particular case of police homicide does so with a rather obvious bias towards assuming that the situation is very bad and comes up with a number, manually inflated, of 1500/y. It may be, but it’s unhelpful to mix stats and opinion and not separate the two clearly IMO. It would have been nice to present the raw data, then the statistical techniques and then finally an opinion based on the results.

That being said some important points are highlighted in terms of a lack of accountability of the police in the US.

This is very neat! It made me think of hackmd.io which offers similar features though maybe less minimalist/distraction free but with collaboration. Thanks for sharing!

I agree with the sentiment but your last paragraph needs some (serious) backing "incredibly easy to stop 80% of anthropogenic emissions", really?

As much as Chomsky may have been incorrect, I think it is foolish to call him "depressingly ignorant" without justifying your statement. His international recognition both in his field and in his commentaries did not appear out of nowhere.

Nothing is sufficient by itself if you want to figure out what's going on in the world. Reading from different sides even biased sides, Chinese, Russian, Qatari or Western 'propaganda', is the best way to train a critical mind and not be fooled into believing simple 'truths'.

This is hilarious, well worth a read.