From the official announcement found in the NYT article (full of details we mostly already know) there comes an update with the team's response:

So, while they deserve the year ban, the apology is nice. It's a shame we can never know what results a fair competition could have yielded ... and an even bigger shame that the media misreported Baidu as overpowering Google. I suppose the damage is done and the ILSVRC has made the right choice.

Perhaps I'm misunderstanding the classification problem but why isn't this run like most other classification problems (like Netflix and many other data challenges) where you get ~80% for training and the remaining 20% are held back for the final testing and scoring? Is the tagged data set too small to do this? Seems like wikimedia would contain a wealth of ripe public domain images for this purpose ...

Again, you're conflating two things that are significant enough that I don't see a simple one-to-one comparison here.

The clear difference here is that the trolls in the article are a nebulous entity whereas the media trolls are not. I know to laugh at Glenn Beck and Katie Couric. I know who they are. I recognize their blubbering stupid talking heads. They're a trainwreck of lies and half truths. On the other hand, you can't stop google from returning search results that confirm what you're looking for. When it's a "trending hastag" on Twitter, you can't figure out if it's legit or not. How do I know that podonski432 on Twitter is the same individual on Youtube named ashirefort posting videos of an explosion is the same person retweeting podonski432 and adding ashirefort's video to their tweet?

Mass media doesn't employ subterfuge and I sure as hell can stop reading the New York Post & Washington Times & CNSNews & Huffington Post and all that other drivel. I can't, however, identify easily that this account on Twitter is just the new troll account that tricked me last time.

You do know that it's news if the New York Times is caught lying or spreading known falsities, right? I watched Jon Stewart hold a "reporters" feet to the WMD fire on one of his recent episodes. There's no self-policing mechanism like that among trolls.

Well, astroturfing is no new tactic but ... I think what this article deals with is scale. 400 clearly skilled (bilingual at the least) individuals running multiple catfish personalities online day in and day out ... the whole thing on a budget of $400k a month? That level and size is probably unparalleled by ... say, Digg's conservative idiots.

You have one entity orchestrating the 12 hours a day work of 400 individuals on topics that are pro-Russian and tangentially pro-Russian. They are sophisticated enough to "hit play" at a certain time to unfold a natural disaster or assassination or anything to destabilize/confuse a region and they do so over many accounts on multiple social media platforms. They create video, screenshots, websites, etc. And they use proxies and sufficiently sophisticated means to appear to be disjoint at first glance.

They appear to have run an exercise on a rubber plant explosion in Louisiana for no other discernible purpose than to test out their new super powers or demonstrate their abilities to their customers/leaders.

Frankly put, I'm unaware of "American organized political trolling" that rivals this. This is paid. This is tightly controlled. This is prepared. This is unified. American organized political trolling is just a run-of-the-mill monkey shitfight with the occasional Koch Bros/Soros website (usually easily sourceable) thrown in.

Now if you can point me to a faked ISIS attack on American soil right before an election that was done by some political group stateside, I'd be interested to hear about it.

No, I can't. I started out and I sucked. I got better eventually through experience. In order for it to be truly bimodal, people have to start in either camp A or camp B and end in the same camp they started in. Because if you transition from one to another over time, any point in time will capture a group of people in between the modes. Now, you can argue that people don't spend much time in between those modes but you haven't presented any evidence for that. What's more likely is you have geocities coders on one tail and John Carmack/Linus Torvolds on the other tail. And in between are people like the presenter and I. And since I'm not instantaneously going from bad to good, the reality of the situation is most likely some degree of a normal curve filled with people trying to get better at programming or even just getting better though spending lots of time doing it and learning a little along the way.

For all your attacks on the presenter, your argument of a bi-modal distribution sounds more flawed to me. I would love to see your study and hear your argument.

To be fair, you can spend a great deal of time talking about something and make progress on the issue without solving it.

For example the current metrics are abysmal so it's worth explaining why they're abysmal. I just was able to delete several thousand lines of JavaScript from one of my projects after a data model change (through code reuse and generalization) -- yet I increased functionality. My manager was confused and thought it was a bad thing to get rid of code like that ... it was absolute dopamine bliss to me while he felt like our production was being put in reverse. KLOC is a terrible metric. But yet we still need to waste a lot of breath explaining why it's a terrible metric.

Another reason to waste a lot of time talking about a problem without reaching an answer is to elaborate on what the known unknowns are and speculate about the unknown unknowns. Indeed, the point of this article seemed to be to advertise the existence of unknown unknowns to "recruiters, venture capitalists, and others who are actually determining who gets brought into the community."

Perhaps ... but that would imply that one does not transition over time from one hump to the next or if they do, it's like flipping a light switch. When I read this I assumed that he was talking only about people who know how to program and not "the average person mixed in with programmers."

But you just said there's no way to measure this ... how could he have collected data? What data set could have satiated us? The answer is quite obvious and such collection would have been a larger fool's errand than the original article's content.

Sure, if you want to kill someone, you're right. I think the argument against drones is that if you push a button and someone dies on the other side of the Earth and you didn't have to go to war to do that ... well, fast forward two years and you're just sitting there hitting that button all day long. "The quarter solution" or whatever you want to call it is still resulting in deaths and, as we can see here, we're not 100% sure whose deaths that button is causing. Even if we study the targets really really hard.

No, no we don't. You might say "Al Queda hit us now we must hunt them to the ends of the Earth" but it doesn't mean that diplomacy and sovereignty just get flushed down the toilet. Those country borders will still persist despite all your shiny new self-appointed world police officer badges. Let me see if I can explain this to you: If David Koresh had set off bombs in a Beijing subway and then drones lit up Waco like the fourth of July and most of the deaths were Branch Davidians, how would you personally feel about that? Likewise, if Al Queda is our problem and we do that, we start to get more problems. Now, that said, it's completely true that Pakistan's leadership has privately condoned these strikes while publicly lambasting the US but that's a whole different problem.

Yep, think of the children -- that's why we should use drone strikes, right? Look, war means death. Death doesn't discriminate and neither does war. If you're hung up on it being okay to take a life the second that male turns 18, you're pretty much morally helpless anyway. War is bad. Drone strikes are bad. There's enough bad in there for them both to be bad. This isn't some false dichotomy where it's one or the other. It's only one or the other if you're hellbent on killing people.

News flash: you can argue against drone strikes and also be opposed to war at the same time. It does not logically follow that since you're against drone strikes, you're pro war and pro killing children. That's the most unsound and absurd flow of logic I've seen in quite some time.

So you're saying everyone knew she was lying about her charity donations as well? Or was it only the charities that knew that? From the article:

So you're saying the 300,000 downloads are by people that knew they were downloading the app architected by a liar? And they were paying $3.79 to Apple and this liar for a recipe app that contain recipes that someone lied about helping her cure cancer? And you're saying that everyone at Apple that featured her app on the Apple Watch knew they were showing a snake oil app on their brand new shiny device? And that the people at Penguin did all their fact checking on any additional information this cookbook might contain about Belle Gibson's alleged cancer survival? And that everybody involved in these events know society's been parading around a fucking liar and rewarding her with cash money while she basically capitalizes on a horrendous disease that afflicts millions of people worldwide ... that she never had?

No, this is not the same as "water is wet" and it needs to be shown that holistic medicine is temporarily propped up on a bed of anecdotal lies ... anybody who accepts it as the sole cure for their ailment is putting their health in the hands of such charlatans and quacks.

Why don't the automakers just seek refuge under the DMCA from all those evil automobile hackers? Clearly, figuring out how your car works is a direct attack on the very hard work and property of those automakers.

Time to pass a bill state by state. I'm the sure the invisible hand of the free market will line all the right politicians' pockets to rush those through. Hopefully someday we won't be able to own our cars and we can go back to the Ma Bell days when every phone was rented.

It's like how a real terrorist would not joke about a bomb at an airport. But someone who does is detained or arrested, and time is spent by TSA that could be better spent looking for real terrorists.

I studied and tutored experimental design and this use of inferential statistics. I even came up with a formula for 1/5 the calculator keystrokes when learning to calculate the p-value manually. Take the standard deviation and mean for each group, then calculate the standard deviation of these means (how different the groups are) divided by the mean of these standard deviations (how wide the groups of data are) and multiply by the square root of n (sample size for each group). But that's off the point. We had 5 papers in our class for psychology majors (I almost graduated in that instead of engineering) that discussed why controlled experiments (using the p-value) should not be published. In each case my knee-jerk reaction was that they didn't like math or didn't understand math and just wanted to 'suppose' answers. But each article attacked the math abuse, by proficient academics at universities who did this sort of research. I came around too. The math is established for random environments but the scientists control every bit of the environment, not to get better results but to detect thing so tiny that they really don't matter. The math lets them misuse the word 'significant' as though there is a strong connection between cause and effect. Yet every environmental restriction (same living arrangements, same diets, same genetic strain of rats, etc) invalidates the result. It's called intrinsic validity (finding it in the experiment) vs. extrinsic validity (applying in real life). You can also find things that are weaker (by the square root of n) by using larger groups. A study can be set up in a way so as to likely find 'something' tiny and get the research prestige, but another study can be set up with different controls that turn out an opposite result. And none apply to real life like reading the results of an entire population living normal lives. You have to study and think quite a while, as I did (even walking the streets around Berkeley to find books on the subject up to 40 years prior) to see that the words "99 percentage significance level" means not a strong effect but more likely one that is so tiny, maybe a part in a million, that you'd never see it in real life.