dthal

Does Criteo count? Yahoo finance says its market cap is 3.3B[1].

[1]http://finance.yahoo.com/q?s=CRTO

> It is possible to put a lot of work into taking a MOOC and get nothing but false understandings and misconceptions out of it because the trust we put in the developers is misplaced.

Of course, bad classes taught by unqualified instructors are offered in the traditional format, too. They may be rare at elite U.S. universities (the author is at Georgia Tech), but certainly at lower level U.S. universities there are plenty of classes taught by people with shaky subject knowledge. I've been in some of them.

Is there any information to be had anywhere about terms? We hear the valuation shouted from the rooftops, but with nothing about the terms, its hard to know what the big number really means. Specifically, does anybody know if this a common stock purchase, or is this some kind of convertible preferred stock/convertible note with a liquidation preference?

>> a) I think one of the biggest challenges in a Kaggle competition is getting away from overfitting to the leaderboard.

This actually depends on the data. The commenter above won the Social Circles competition. That competition had a very small number of instances - it looks like it was 60 in the training set and 50 in the test set. It had one of the larger shakeups in Kaggle history.

I think the key is that R (and S before it) were designed from the start to provide a continuous path from being a user to being a programmer. Unlike real programming languages, you can get a lot of value out of R without really coding. Its a great statistical calculator and has good graphics. Then you can move relatively easily into scripting a few repetitive tasks, and then on into writing simple programs.

One consequence of that is that R has a lot of 'non-programmer programmers', statisticians and domain experts. Some of them write libraries that encode their domain knowledge. Then over time that adds up to having library support for more different types of analysis than any other language. I personally dislike R as a language, but I often end up using it because it has a library for some task that just doesn't exist in Python.

From [1]: The Global Warming Policy Foundation (GWPF) is the United Kingdom's most high-profile climate denier group. Its founder is Nigel Lawson[2].

[1]http://www.sourcewatch.org/index.php/Global_Warming_Policy_F...

[2]http://www.sourcewatch.org/index.php/Nigel_Lawson

>>Several journals are trying a new approach...in which researchers publicly “preregister” all their study analysis plans in advance. This gives them less wiggle room to engage in the sort of unconscious—or even deliberate—p-hacking that happens when researchers change their analyses in midstream to yield results that are more statistically significant than they would be otherwise. In exchange, researchers get priority for publishing the results of these preregistered studies—even if they end up with a p-value that falls short of the normal publishable standard.

It's not exactly the same issue as the one addressed by banning p-values, but this would help a lot.

From the article:

>>libSPRITE defines engineering units as types (i.e., Meters or Radians instead of double or int).

I guess experience is a harsh teacher.[1]

[1]http://en.wikipedia.org/wiki/Mars_Climate_Orbiter#Cause_of_f...

>>Reservoir sampling is appropriate with more than just a set of unknown size -- you very frequently know the size of a set, but it's still too big to sample directly.

I'm not sure that I buy this...Do you have an example? Still, it is a neat algorithm. And it is a true stream query algorithm, that is, if you have an open-ended stream, it maintains a random sample at all times.

I'm not a ruby guy at all (never used it), but if I understand this correctly:

  j = rand(idx)
  out[j] = i if j < n

..it is keeping the sampling probability at n/idx where n is the sample/reservoir size and idx is the number of items seen so far. Then if this item is selected for sampling, each of the n items in the reservoir is equally likely to be replaced. I think the implementation is correct, assuming that idx is the number of items seen so far.

>>The document noted that many SIM card manufacturers transferred the encryption keys to wireless network providers “by email or FTP with simple encryption methods that can be broken … or occasionally with no encryption at all.”

If that's true, then NSA/GCHQ aren't the only people who could have grabbed a big pile of keys.

Nice point...Can the worker set prices? At a minimum, if the answer to that is 'yes', then they are a contractor. And most people who can't set prices are employees, or are at least edge cases(e.g. taxi drivers). That's the best single test I know of.

I'll kick things off by linking to the comments from the last time this was posted: https://news.ycombinator.com/item?id=8769067.

>> ... however, the choice of algorithm mattered little... and the largest gains always came from introducing new features.

This was also the collective opinion of top Kaggle competitors as described in a post[1] on NoFreeHunch.

[1] http://blog.kaggle.com/2014/08/01/learning-from-the-best/

That makes complete sense, especially since you are not going to get in to YC as a hardware startup without a functioning prototype.

>and we don’t shy away from expensive hardware--we’ve funded companies building things like nuclear reactors and rockets, which will require hundreds of millions of dollars in funding to succeed

How does this make sense for YC to do?

Science has more and less reliable disciplines. In particular, in any field in which meaningful experiments are difficult or impossible to perform, causality will be hard to pin down. It just so happens that many of the topics the public is most interested in - health and nutrition, psychology, economics and, yes, climate - fall into that category.

If you haven't seen it yet, check out the video at the bottom, too. It was shot days before the San Francisco of 1906.

I think you are exactly right. Any solutions to this have to be technical and largely defensive. Legal or diplomatic solutions will only bind parties that intend to follow the law. Counter-attacking and deterrence will fail if the attacker is not defending much of anything, so that approach might work against states and state-backed actors, but not against non-state actors.

I think all of the cynics on here (including me) should pause for a moment and reflect on what is happening in this case: the government is going to do something that will cost it revenue, just because its the right thing to do and the people wanted it.

I think lids do three things: they cut down on noise, they reclaim some open space, and they piece back together the neighborhoods that get cut in half by big freeway projects. The Mercer Island lids may not do much for noise, but they do the last two of those things nicely. I think that last item, putting broken neighborhoods back together again, is really important and it can be accomplished pretty well even by fragmentary lids.

OTOH, how much does it really cost them? They make some software modification, and they get to try out a new market. Maybe it works, in which case they just found a new long-term, possibly world-wide, revenue stream. Maybe it doesn't, in which case they have a bit of code to throw away.

Sorry if its a little OT, but I actually have this question: Is it normal for grad students to get reimbursed by their institution for conference travel?

>One incentive was the availability of federal funds >earmarked for sex predator stings. Another factor: >asset forfeiture, which allows police to seize >the property of people who are arrested even if they aren't convicted.

Whenever arresting people becomes a big earning opportunity, problems are sure to follow. Profit-oriented policing (mainly from traffic enforcement) was one of the grievances that fueled the Ferguson protests/riots.

I suppose stuff like this [1] could start to happen. Note that Moxie Marlinspike is not actually in/involved with wikileaks.

[1] http://en.wikipedia.org/wiki/Moxie_Marlinspike#Traveling

>But what if the business requirement changes...

Are you kidding? This is FizzBuzz! It is just a quick check to verify that an applicant is even a programmer at all. That's all it is. There is nothing else there.

I think the solution to this has to be mainly technological, not (primarily) legal. Fake cell towers are not NSA-level technology and it seems that they are being deployed in lots of places [1][2], probably by lots of different parties. If a bunch of people are doing it, changing the law won't help that much. Probably not all of these actors are 'legit' (police or whatever). The networks themselves have to be secured better so that this is impossible, or at least much harder, to pull off. [1]http://www.businessinsider.com/mysterious-fake-cellphone-tow... [2]http://www.foxnews.com/world/2014/12/14/norwegian-security-p...

Thanks for that link. It seems to me that what that data says is: Americans are for torture, pretty much across the board. That's good to know...

One problem with that is that the hackers that you really need to worry about are ones who are coming after you: competitors trying to steal IP, business rivals investigating your deals and so on. These high profile attacks like the Sony attack are embarrassing, but the real business damage is probably done in other attacks that no one ever hears about.

Did the DA offer just one plea deal or was a series of offers made over the time Jelani was in jail? This seems to me like the DA was using pre-trial detention as a pressure tactic.