jebus989

As a European (ish), ditto.

Ha, no you crunch the numbers, you get the insights. Google does so whether you opt-in or not.

Something I'd like to do (someone else asked for an e.g.) is look back at programming language searches, as they'll nearly always have the lang in the query. I'd be able to see when I started learning e.g. python and had to google everything, when I picked up other languages, how those searches decreased (or increased) as I learned more. Lots of interesting stuff.

How about just loosening the API rate limits, or making a better token request process with resource allocation e.g. I'd like 1500 requests per 15 min window (as opposed to 15, for some things) for 72 hours. I guess this could be limited to those with a academic email address if they insist.

I know we're meant to be freaked out by this but I think it would actually be pretty cool to have data for tens of thousands of search results stretching back a decade or so. With a bit of analysis you could get some cool insights into past interests, their dynamics over time etc.

It's a logistic regression model, a basic statistical technique which wouldn't have even come under "machine learning" a few years ago. Later they use some kind of LASSO regression to penalise the inclusion of redundant features.

"Combing through the correlations", it seems, literally means calculating the (Pearson) correlation between two variables (success vs. an input feature) and adding some interpretation, as they do on p6 of the arXiv paper. For test/training data, it looks like they just used a 30/70% split rather than k-fold cross-validation and holdout, but I'm sure it makes no difference either way and in this case (as often) is trivial to design and implement. Presumably their AUROC could be increased just by dropping in an SVM or a Random Forest in place of the logistic regression.

From what I've skimmed of the paper you're over-estimating the complexity of the study.

Or a soviet version of Passport to Pimlico.

All the usual HN nitpicking and obscure edge case discussion should probably be done in the light of ehamberg's own comment: "This is a horrible idea in many ways and was a quick hack for fun." [0]

[0] https://github.com/ehamberg/9m

I don't follow your logic. A forced sale of bitcoins to assuage creditors is different to speculators losing confidence in the currency and trying to exit their positions. The latter causes market volatility and price crashes. Temporarily increased liquidity is not inherently disastrous.

Agree, they're not a finite resource that should be miserly handed down every few years by ICANN. Register whateveryouwant.whoopdedoo for all it matters.

"Organic" as in organic chemistry is to do with compounds containing carbon.

"Organic" dictionary definition: "relating to or derived from living matter", i.e. [0]

> They add that the carbon they use is new as well—it's an organic compound grown from cotton fibers

[0] http://phys.org/news/2014-05-power-japan-dual-carbon-battery...

Agree, and Wikipedia does this e.g. for Wikipedia books [0]. If you render any book as a PDF it will append a ~10pt list of all article contributors.

[0] https://en.wikipedia.org/wiki/Wikipedia:Books

In the US 18% engineering students at undergraduate level were female as of 2012 [0].

In fact, apparently in Iran almost 70% of science graduates are female [1], making the courses less diverse in terms of gender than the US, which I believe is closer to 50/50 for "science" collectively (can't find data).

[0] http://chronicle.com/article/In-Terms-of-Gender/135304/

[1] http://contexts.org/articles/spring-2011/what-gender-is-scie...

I can't believe ~=< 30% health gives the same catch rate as 1hp, so many wasted save/loads trying to get the other guy down to a slither of health.

I know a guy, mid-twenties, who if wanting to search for something will type "google" in the chrome address bar, enter, click the top hit, then search from the recognisable google homepage.

It drives me insane.

It's acceptable to automate twitter activity per the rules here: https://support.twitter.com/articles/76915

I'd guess on Indian review-aggregator websites written in Hindi the trend is reversed ;)

Excuse the plugging of my own wares, but by coincidence I did some analysis the other day of the disparity between User/Critic ratings of movies on RottenTomatoes: http://benjaminlmoore.wordpress.com/2014/05/05/what-are-the-...

Example: "Spy Kids" got a 93% fresh rating from critics, but 45% from audience ratings.

Code attached should you want to investigate for yourself.

Strange comment, you realise that's not an inherent truth of language? Unique words per song is trivial to calculate

That's a great video thanks. A relevant part for this thread is where he says:

> We're really lucky that the first and second-generation users haven't broken their silence but at one point they will ... [they'll] say "well actually it's been pretty tough going, actually I had to wait another three weeks for this part thing and my extruder jammed up and ... I'm spending more time on the machine than making stuff." [0]

Seems to be what's coming up here; personally I had no idea they were this far from consumer-ready.

[0] http://youtu.be/OZPtKiK9idY?t=13m2s

"Data scientist" just sounds more impressive than analyst and less boring than statistician.

When a scientist says "an explanation for X could be Y" it shouldn't be interpreted as "here's a guess, I want it to be true and am undermining science", rather it's a sign they are being extraordinarily careful but they're reporting that they have observed some evidence which appears to agree with a given hypothesis.

Also I suggest reviewing high school biology, glycolysis isn't simple the breakdown of glucose into simpler molecules but involves many seemingly complex stages of isomerisation and reactions which are shepherded under specific conditions within various enzyme's active sites.

Consider reading the paper for more information: http://msb.embopress.org/content/10/4/725

That's a silly caricature of how the place works, and not something I've experienced in the 8 years I've had an account there.

I don't understand your expression but I was just quoting a news source without any interpretation implied. I can imagine how such a scenario would hamper the prosecution moreso than someone just not keen on pressing charges.

Apparently she went further than declining to press charges, the defendant's lawyer claimed that the victim "'provided photographic proof of an unblemished complexion to SFPD' and filed an affidavit requesting that the investigation be terminated" [0].

[0] http://www.sfexaminer.com/sanfrancisco/attorney-for-internet...

Someone == a PR firm: http://www.eileenkoch.com/ (warning: noisy) but as the current revision reflects, Wikipedia articles are built with respect to due weight and neutrality. While it's not ok to hide unflattering info, it's also not ok to bloat a short encyclopaedic article with court case details with the intention of highlighting how bad a guy the subject is.

I'm relieved we've gone back to light-heartedly reporting DPRK's eccentricities rather than worrying about the potential consequences of this powerful and deluded leader's take on international relations.

Lots of fields primarily use LaTeX to typeset papers, e.g. physics and computational biology. Typesetting and hosting a PDF never justifies a $1500 publishing fee given that reviewers, editors and of course authors are all working for free. Hence even the good-guy (non-profit) publisher PLOS took about 20% excess revenue in 2012, primarily by charging $1300 odd dollars for publication in PLOS ONE.

Why do we even need the dated journal model? Push everything to a central arXiv, biorXiv like repo (funded as they are by small donations from institutions). Couple this with open post-publication peer review and endorsements. Millions of tax dollars saved around the world, research open to everyone.

The counter argument is the rubber stamp that accompanies a prestigious journal (and filtering the crazies) but if the repo was accompanied with a suitable array of article-level metrics and thorough categorisation the good science can continue to filter to the top, much like how HN and reddit work.

This really isn't worth worrying about. I know it seems like a Pascal's wager scenario but "youth of stem cells" isn't a problem. We know how to lengthen telomeres (hTERT) and generating human iPSCs is a routine procedure.

Can't say I've tried it but the above-mentioned docopt has a Go implementation: https://github.com/docopt/docopt.go

Ha yeah I came to terms with this after doing a double-take some time ago. But we still pronounce it code academy right?

OT: Reminds me of when I learnt I'd been reading deprecated as depreciated for years, until seeing: http://english.stackexchange.com/questions/45295/why-is-ther...