karmacondon

There is one clear benefit to global trade: It makes things cheaper. The problem is that it only makes things a little bit cheaper, and it's difficult for most people to see the effect in their daily lives.

In broad strokes, free trade is essentially a trade off between some US workers losing their jobs and reducing the cost to make everything that people buy at Walmart. I can understand that most people on hn don't shop at Walmart, and wouldn't even notice if prices for consumer goods went up by 5-10%. But the rise in household costs due to ending free trade would have a major impact on the poorest and most vulnerable citizens. A single mother with multiple children and one income simply can't afford to pay X% more for everything that her family buys.

And for what? So that a factory worker in Michigan or Ohio can have their union job back? That just seems like fighting progress. Those jobs are gone, mostly due to advances in technology and global connectivity. It makes sense to build things where it's most efficient to do so and to replace humans with robots wherever possible. Why increase prices on everything and hurt national and global economic growth just so that some people can keep the jobs that they're used to having?

And unfortunately, "re-training" programs tend to work out better on paper than they do in reality. The fact is that no one wants to throwaway a lifetime of work in a given industry in order to start all over again in another. It's not just learning new skills, it's years of meeting the right people, learning new cultures and customs, etc. So few people want to do that, it's just not human nature to start over from scratch.

HN is starting to sound like a broken record, but a Basic Income is probably the best solution. I believe that some politicians are also proposing a kind of "workers insurance", to help people who lose their jobs due to changes in technology or culture. Either of these would be a better solution than forcing price increases onto the millions of people who have jobs but still barely make enough to get by.

It seems like IBM is using the Watson name to re-brand itself in an attempt to turn their business around. So that "using Watson" is now synonymous with "hiring IBM consultants".

Ten years ago this headline would have read, "Pentagon hires IBM to help optimize procurement". But the Watson brand has become popular with the media, in large part due to the exposure from the Jeopardy exhibition (which is seeming like a better and better investment in terms of marketing returns). I'm not sure if the re-branding will be enough to really change IBM's fortunes, but does seem to be helping them in the short term. As companies get better at using open source tools and expert knowledge to leverage their own data, it's going to get harder for tech consulting companies of all types.

I understand what you're saying and I agree with you. Human learning is based almost entirely around feedback and trial and error. But by comparison, machine learning requires a lot of implicitly labelled data, and that's what I think Yann LeCunn was talking about.

For example, to train a classifier to identify birds you need a large number of pictures of birds, maybe millions of varied examples. And then you'd need an equivalent number of images of things that aren't birds, or things that look similar to birds. Humans are able to make that same classification with a very small number of examples, maybe even n = 2. If someone saw a bird for the first time and then another one shortly after, they would be able to put the two together immediately and make a lot of inferences on top of that. Machine learning algorithms aren't even close to that yet.

Supervised learning sucks because it requires vast quantities of labelled and prepared data, which is expensive. Top ML researchers must feel like they're sitting in a Formula One racer but can't afford any gas for it. AlphaGo shows that a computer can do almost anything, but only with a significant investment of resources for each specific task. Unsupervised algorithms tend to be less effective than supervised methods right now, but once that changes it will open up a new world.

What do you think RTB is? There are websites that will sell demographic information from user profiles, but that's relatively rare. The overwhelming majority of the billions of RTB impressions are based only on location and website audience demographics. Those are things you get from cookies and ip addresses, information that comes entirely from access logs. No, companies don't sell their logs directly, but the vast majority of targeted impressions are sold solely on what is contained in a standard server log.

You seem to be knowledgeable about online advertising and forthcoming with sources. Do you have any evidence that anything near 50% of online ad data sold is based on the highest grade profile data? Because I'm certain that is not the case. I think you're confusing the business of selling profile data on individuals, which is indeed very old, with modern ad targeting. It's hard to persistently match that up with an ip address and cookie, you see. Which is why major ad buyers prefer to buy cookies that tag large demographic buckets like "young males in the midwest with an interest in cars" instead of the names and profiles of individual people. It's common sense more cost efficient, and access log data is much more prevalent and reliable than profile data.

I don't think that any of this works in the way that you think it does. Yes, re-targeting is creepy, cookies are rampant and the average page loads way too many external files. But companies do all of these things for a reason, whether you understand the details or not. If you don't like how they do business, you can just stop going to their websites.

It is their data. What they sell is essentially their http access logs, the ip address and time of every page request. While that information may pertain to you, it's not yours. They own it and can do what they want with it, including selling it to the highest bidder. That's true for any website that keeps access logs.

If the idea that some websites sell their access logs really disturbs you, don't request pages from websites that do that. Just like someone who is afraid of heights can't go to high places, people averse to ad tracking won't be able to go to most websites. Most people don't want to give up their favorite websites, so ad tracking persists.

Many companies that produce or aggregate content do so with the expectation that their efforts will be rewarded with money from ads. When you use an ad blocker, you reap the benefits of their work while knowingly depriving them of what they expected to earn in return. It would be better if you just didn't read their content, effectively voting against their behavior with your feet. Blocking the ads is having your cake and eating theirs too. It's rather benign, there are much worse things, but it isn't really right either.

Where this analogy breaks down is that the people sent to track you are invisible and can't be seen without the aid of special technology. So what you end up telling people is that they are being followed everywhere by invisible ghosts, who's only desire is to change what ads appear in their newspaper. And it appears that the reaction of most people is about what you'd expect.

If a store has policy of "If you come into our store, we'll have employees follow you home" and you don't like that policy, then don't go to that store. That simple. It doesn't make sense to go into the store and have your goons beat up their employees. That might mean that you can't go to the stores you want to go to, but that's how it goes. It seems as clear online as it does in the physical world.

(tldr without the analogy: The overwhelming majority of people don't care about being tracked online because there are no obvious ill effects. The problem with ad blockers is that it makes more sense to just avoid sites that show ads, but most people don't want to do this because it would exclude their favorite sites.)

I would be all for banning submissions from zerohedge.com outright. It's just noise that makes it harder for me to skim headlines for actual content.

Relevant web comic: http://www.smbc-comics.com/?id=3540

You compared the current state of deep learning to the early internet. Can you clarify that in any way?

I'm having a hard time seeing how deep learning could improve exponentially from where it is now.

Do you know of a previous case that's similar to what you're describing here? Because I'm having a hard time understand this. Unless Apple entered into an agreement with Uber, I don't see how Uber would have an expectation that their relationship with the App Store would continue in its same form into perpetuity. And I'm almost positive there is some condition in the apple developer agreement that says "We reserve the right to terminate without notice or reason" or something similar.

What you're saying is reasonable, and I could see how a judge might agree with the arguments you've made. But it also seems like Apple could say, "We don't like the CEO of Uber's haircut", which is as valid as any other reason to terminate a non-contractual relationship. If Uber wanted protection for the billions they invested, they should have negotiated a proper agreement. Maybe I just don't understand how the law works.

I have a question for the lawyers on HN. Why doesn't the judge in the case hold Apple in contempt of court until they modify the OS, in the same way they would hold a reporter in contempt until they revealed a source?

In both cases, you have a third party who is being compelled to assist an investigation against their will. My only guess is that there is a precedent set for reporters, but not for computer programmers.

What's the killer app for AR? speadsheets:computing, email:internet, ____::AR?

That's brilliant, you should try it. I'd kind of be surprised if someone hasn't done it already.

It's interesting that all of the numbers in your post seem to be significant, but really aren't.

The US "owes" China a trillion USD because the Chinese, investing on a national scale, need a safe investment that can support trillions and trillions of yuans. And in that regard right now, the US is the only game in town.

China's GDP growth is high, but that's because they're starting from a GDP that's much lower than that of the US. As with any logarithmic curve, growth starts quickly and then becomes harder over time. China has a ways to go.

The number of cars sold and the number of children enrolled in school are indicators that China does indeed have a much larger population than the US. Almost all of the BRIC nations have large populations, but are still in the process of turning that into wild economic growth. We'll see.

Other than the missing context for the numbers you provide, I agree with your general sentiment. There is a bit of schadenfreude among the US media whenever China struggles economicallly. This is probably because China is the US's biggest economic and military rival now that the Soviet Union is gone. It's natural to cheer for your team and boo the other team. I wouldn't take too much offense.

I was wondering exactly how the NSA trained models to detect terrorists, and was surprised by the level of detail in this article. So the NSA performs classification using a Random Forest and about 80 input features? Huh. That actually sounds a little too similar to a Kaggle contest for my liking, but I wasn't expecting that much quasi-technical information anyway.

It seems a little silly to write this whole article based on a few powerpoint slides from years ago. Even as an amateur practitioner, I can see several obvious things that could be changed about the presented methods. I'm sure what the NSA is doing now in this regard is much further along than what was portrayed in those documents. Whether or not they should be doing it is a discussion for others to have. But it is interesting to learn a little more about the details of truly high stakes machine learning.

My favorite quote about Watson is "If Watson is so smart, how come it can't figure out how to stop three years of declining sales numbers at IBM?"

More than anything, this ad makes me think of how long of a way we have to go before this type of AI will be useful to humans, much less able to run a country. Why doesn't President Obama consult Watson before making decisions? Because every decision would require a massive data collection and processing project, training and tuning of delicate models and rigorous testing. And the net result would be what? Processing of factual information that any human could get by reading a brief prepared by an aide?

We're worried about AI representing an existential threat or creating wide spread unemployment, but right now the best and brightest computer in the world can't provide any more practical political value than Monica Lewinsky. This is a good ad campaign, timely and provocative. But it also highlights how the path ahead is as long and arduous as a trip to Mordor. glhf, IBM!

You could research the funding of startups. Or more broadly, how should capital be allocated?

Admittedly I have no idea how something like that could be quantified, and the study of economics has been trying to figure it out for decades. But if YC Research is looking to make maximally impactful investments, it seems like studying investing itself would make sense. There's a high risk of not producing anything useful, but also a huge upside if you get just a little better at figuring out who to fund. YC Research is in a position to bring together people from the worlds of business, machine learning and economics who might be able to make a dent in the problem.

Practically, it seems like navigating the financial investment process is the core skill of YCombinator, so you'd already be pretty far down the learning curve. And you don't have to convince people to adopt your ideas right away, you can try out whatever ideas your research produces simply by making investments. Better capital allocation of all forms could have a really big impact on society at large, if the right systems can be developed.

Completely subjective, but I'm very drawn to the timeline. I would be interested in a site that was organized entirely around that, with the videos and pictures displayed chronologically and by importance.

This seems like someone saying "I don't agree with the direction that BitCoin is heading in" and then taking his ball and going home. A bit dramatic yes, but not uncommon in open source projects. It's like someone saying "Linux is dead because it doesn't confirm to XYZ principle", which seems to occur every other Tuesday. Forks happen.

It seems like he makes some cogent criticisms, for which the best solution seems to be to fork the project, or begin developing for any of the other *coins that were so popular a couple of years ago. But BitCoin has the dominant brand and the people who control the mining and the source code can do whatever they want. Something tells me that bitcoin itself isn't dead because there's been too much invested in it by too many people, whatever the problems. But it's dead in the eyes of the author and maybe for several others, and that's all that matters in the context of his blog.

This is expected. As the parent comment said, it's common for people to be "too lazy" to unsubscribe from meetup groups. It cost nothing to be a member (though I understand that group owners are charged a fee based on the number of members), and I don't think that most people like clicking "unsubscribe" from anything. It's some kind of Thinking Fast and Slow mind trick involving familiarity and the fear of missing out, or something.

For a lot of people, meetup group emails are like a newsletter or an rss feed. They might skim it occasionally, but rarely go through the trouble of actually showing up. So, it's not you, it's them. It's probably best to see it as "I have a meetup group with 10 people in it", which is an accomplishment on its own.

The internet is difficult to describe to someone who has no experience with it. It's like "There's this flibity divit called Google that lets you search for hurgfbr and find veruhryhr written by people you like." If someone has no way to conceptually anchor terms like "search engine", "web pages" or "blogs" then it will be difficult for them to understand what's being said, much less know what they're missing out on.

With bare minimum exposure to the ideas via Free Basics, people will be able to imagine what else is out there. It becomes "Google is like when you type words into the search box on facebook, except you get more results and they're more relevant." I would imagine that once exposed to even a limited version of the internet, people will quickly demand unrestricted access from their local government. But they have to know what they don't have before they can ask for it.

If this is all an Evil Plan by Mark Zuckerberg to set up an internet monopoly in India, then it's a very stupid plan. Facebook is a communication platform. People will use it to communicate with people who have normal internet, and will no doubt hear about all the things that they don't have access to. It's only a matter of time before "Free Basics" becomes "Low Cost Government Internet Access For All".

For many people from my generation, AOL was the internet. It was all we knew. But eventually we figured out that there was more out there, and the people of India are capable of doing that as well. There are hackers in every culture, at every social strata. If you give them an inch, they'll turn it into a mile or more. Free Basics isn't perfect, but for many people it's better than nothing.

Hey mattmaroon, I just noticed your comment here. I was thinking of you specifically when reading this article. I could have sworn you were talking about daily fantasy 7 or 8 years ago, back before it was "a thing".

I'm curious, did the market play out like you thought it would? Is this about where you expect daily fantasy will end up, or are there still big changes ahead?

I've never had any of these experiences when riding in cabs. The cars were always ok, the drivers were polite and well spoken, they took direct routes and never demanded cash.

I use Uber for one simple reason: It's 30% cheaper than a cab. It isn't the quality of the cars, the personality of the drivers or the nature of the service. All of those things vary from case to case for every transportation service. Uber's only compelling feature is their lower price.

From what I understand, the average Uber driver makes very little money, < $9/hr after Uber's fees and vehicle maintenance costs. And the low customer prices are effectively being subsidized by venture capital money. How long can this last? Could this be another Groupon situation, where the Ponzi scheme doesn't fall apart until after Uber's IPO?

Something bad would have to happen to random citizens as the result of government surveillance. Something like "Private Citizen X criticized the government and embarrassing information about his life was revealed as a consequence."

There are lots of bad things that the government could do. But it just hasn't happened. They've had mass surveillance technology in place for over a decade now. The world hasn't fallen apart, Hitler hasn't risen from the dead and everything is pretty much the same as it was before.

I guess we can check back in another ten years to see if your apocalyptic visions have come to pass yet.

We both made the mistake of discussing the US government like it's a single entity. We're talking about hundreds or thousands of individuals spanning multiple generations. I'm not going to worry about government metadata collection because of something that happened during the Eisenhower administration. Each person and group of people should be evaluated based on their own behavior and merits, not the reputation of the organization that they are affiliated with.

It looks to me like the US agencies, and the Five Eyes in general, are capable people who are just doing their jobs. They aren't bothering me and I'm not bothering them. The past actions of the US government or hypothetical scenarios based on historical examples just aren't very convincing. Anything could happen. But I'm not going to concern myself with it until I see some evidence.

The problem with drawing historical parallels is that they never apply exactly. Saying "A thing happened in the past" can be instructive, especially to people who didn't even realize that the thing was possible. But what's much more practically useful is to say "I think what happened before will happen again in the current context and for these reasons". An example from the past is only useful if it can be tangibly connected to the current situation, right now in the present.

I can't think of a case where stable and mature democratic bureaucracy has ever used surveillance to influence the majority of its populace. Germany in the early 20th century was a very instable government in a bad economic situation. Soviet East Germany was communist, which isn't quite the kind of democratic that I meant. It's true that any government could turn bad, in the same way that anything is possible. But there's very little evidence for that in the current context.

So my position is this: Given that I live in the United States in 2016, I'm not worried about the government randomly deciding to screw with me by looking at my electronic communications and acting on them. It just doesn't make sense. I'm not significant relative to the scale of the US government, the government itself just doesn't work that way and all of the negative scenarios I've heard seem to be very contrived.

If you really think that it's possible that the government of a modern western nation could turn into communist East Germany, then it seems like your problem might be with governance, not privacy. If it's possible for the government to go all Walter White and just turn evil over night, then no amount of personal privacy is going to save any of us. And until it seems like that's a thing that's actually possible, I'm going to make practical decisions about my own privacy.

You're saying, "We should not trust the government because they did things that I didn't agree with in the past". This seems like an unfair standard to hold any person or group of people to. I would be unhappy if people said "I think that karmacondon has made mistakes in the past, so he shouldn't be trusted to do his job ever again."

I understand what you're saying, and I think I get where you're coming from. But like the GGP post, you're begging the question and assuming that your beliefs are so correct that anyone who disagrees with them must be insincere.

I don't think there's anything inherently wrong with the government monitoring potentially criminal groups or building databases. That's what we pay them to do. If they get out of hand then we, the people, will deal with it.

The underlying assumption to your hypothetical is that thought crimes exist. I would say that someone in that situation doesn't have a privacy problem, they have a governance problem. Either a dictator has seized power or their fellow citizens have voted to make it illegal to express certain ideas. And in either case, encryption isn't going to be much of a solution. It'll only delay the inevitable. Someone who talks about illegal ideas is taking a big risk anyway.

It seems common that the arguments for privacy trumping other values depend on bad behavior by state actors. In which case, reforming the state by whatever means necessary would probably do more good than advocating for philosophical concepts.

Still not convincing, at least not to me and many of the people that I know. I'm pretty sure my insurance company has already calculated my premiums based on my demographics and claim history. I made not have told them "I have more than five drinks almost every weekend", but they know. They were making inferences from Big Data since back when it was just "data". And practically, I don't think most insurance companies have an infrastructure set up for handling snitching, especially not the smaller ones.

I'm really still waiting to hear a convincing argument as to why I have something to hide, ideally something practical as opposed to hypothetical or philosophical.

Not everyone agrees with you that the tech sector is contributing to the building of a surveillance society or police state. There are a lot of people who have carefully considered the issue and come to the conclusion that facebook knowing what posts you liked or ad networks knowing which pages your IP address has visited is not a Bad Thing. It's clear that you don't agree and all debate is welcome, but I caution you not to trip in your rush to claim the moral high ground.

I don't think there's any need to rehash the debate here. Simply, I and many others do not believe that any western government is going to use information gathered by tech companies to preempt threats to entrenched interests and the status quo. I've seen the same arguments made here for years, and none of it is convincing.

It's admirable that you are so certain in your beliefs. If you don't like what the tech sector is doing, please by all means continue to advocate. Shout it from the mountain tops, go to work for the EFF. But don't discount people that legitimately disagree with you as being irresponsible. At least some of us have made the effort to understand your point of view. The least you could do is to try to understand ours.