The comparison with lack of blah at a shelter is telling, don't you think, when you think about how shelters are usually a place where people are "doled" out stuff? It is not merely a patronizing suggestion (which the billions of people are already aware of and accustomed to), but it is also taking a lot of people down a path where their choices will become severely restricted in the future. How is that "A Very Good Thing"?
Also, think about what companies start doing when they start failing. And don't make implicit assumptions that these megaliths will never fail or stumble. What would you, as the CEO who is supposed to "maximize shareholder value" do? Are you going to say "Well, we could do some shady things to exploit the data we have and buy ourselves some more time, but it is a better idea to declare bankruptcy and close the company"? Imagine a company like Facebook nearing bankruptcy. You can bet on your life that they will offer the appropriate third parties greater control of their data mining prowess in return for a bailout. Is that also overall a Very Good Thing?
Also, notice how the 2008 financial crisis unfolded. None of the offending parties got anything more than a slap on the wrist, because they had the keys to an engineering construct of extraordinary complexity (the financial system). It was just "heads I win, tails you lose". The way our lives are becoming intertwined with these tech behemoths is no different, and I don't see it playing out any differently if there is a similar crisis in the tech sector. How can that be overall a Very Good Thing?
"Many other toolkits create examples that appears to be identical to the original implementation of a paper, and then offer a converted model for people to download and evaluate. This is highly irresponsible in our opinion. In CNTK we pay a lot of attention in tracking down bugs and make sure the toolkit can be used to actually train the model from scratch and achieve state-of-the-art accuracy.
One example is the Inception V3 network (https://arxiv.org/abs/1512.00567) originally developed by a few researchers at Google. Tensorflow shared the training script for Inception V3, and offered pre-trained models to download. However, no one were able to retrain the model and achieve the same accuracy with their script. The best accuracy that were achieved by a third party (Keras in this case) is about 0.6% worse that what the original paper reported. Researchers in the CNTK team worked hard and were able to train a CNTK Inception V3 model with 5.972% top 5 error, even better than the original paper reported! The training script is shared as an example (https://github.com/Microsoft/CNTK/tree/master/Examples/Image...) and you may verify it yourself."
Ha ha, that does read like sour grapes.
Is the idea that they will somehow provoke comparison with the 800lb gorilla, even at the expense of looking juvenile? What a bunch of whiny cry babies!
Is it? Because it seems to me that they are having plenty of issues with gaining the trust of advertisers too. Considering that advertising is the bread and butter of their business, this would normally be something of a mini-crisis.
A quick search for "do advertisers trust facebook" gives me
Of course, no one actually believes anything Facebook says with regards to privacy, and probably never did.
I would seriously doubt if there is anyone who actually looks at their practices who doesn't think they conduct large scale psychological experiments on the human lab rats who continue to use FB.
In other words, seems like there are enough ingredients here for an implosion. Something has to give.
Not disagreeing, but I would certainly like some references if you could provide us any. In fact, it would be genuinely troubling if you cannot find any good sources. Here is why: this notion of fingerprinting seems to be an invention of the legal wing, to be brought out as a CYA when these requests were inevitably going to be demanded.
Time for some math:
Since it is only a 0.01% chance, it means you need 10000 discrete pieces of information collected on a single individual before there is a chance of error. If a company indeed has that many pieces of information on you, you first of all need to know that for a fact.
There is a chance the company will counter that this is aggregated probability, as in, with an uneven distribution of errors. If it is indeed aggregated probability, the companies which advertise on these platforms need to demand their money back because for all you known, none of the folks they are targeting are actually correct fits for their ads. Fingerprinting puts the burden of proof on the shoulders of the company that they are indeed allowing advertisers to target the audience they want. How can they be so sure if the errors are unevenly distributed?
In any case, everyone should demand the information anyway, and let us start using this fingerprinting theory as an excellent opportunity to get deeper into the practices of these companies.
Oh, come on! Why did you have to ruin the fun? Don't you know we can have a ton of fun chuckling sarcastically to ourselves when we see folks like GP keep making these kind of shockingly delusional statements?
This was exactly my point a while back. That the default option should be to get explicit permission for each piece of data you collect AND infer, even if the inference is being done in situ as the code executes (otherwise it will become another out clause). Of course, the geeks who get all delirious by seeing a mountain of data to analyze would not want that kind of friction in the process.
I wonder what would happen if someone would spend the time to completely dissect and reverse engineer exactly how lookalike profiles are being generated. My guess is that it will expose data collection practices which will confirm our worst fears.
Moral: of, relating to, or concerned with the principles or rules of right conduct or the distinction between right and wrong
I think it is "clearly wrong" that Facebook creates shadow profiles, because it is violating the freedom of people who have not signed up for their service, in the same way that it is clearly wrong for me to take away your favorite pet for ritualistic sacrifice against your wishes, even if "everyone" in the community agrees that it has been a consistently effective method for pleasing the Gods.
Of course, you know these analogies are tenuous, and you will eventually go into very precise definitions of words (or worse, you will actually start taking my pet sacrifice analogy and dissecting it). Perhaps you could tell us about something you think is clearly unethical, and we will try and draw the connection for you.
At the moment, there are only two kinds of employees at Facebook. Those who care and are getting irritated each time these issues are raised on HN (see here [1]), and the dregs who bury their heads in the sand. I bet there is someone who works there who is reading this and realizing that either they will have to change their attitude, or soon the company will turn into another Enron. We don't still have Enron in our midst anymore, do we?
Once one company goes down, it is only a matter of time before the rest fall in a domino sequence because people will start wondering about the practices of its peers. I would like to think that these companies are a little more sensible than to imagine they are somehow infallible, its better for them to change now before it gets to the point where they are made to.
>> We really need someone to fight for our privacy and neutrality.
Well said.
I certainly hope it is not a PR stunt, but WordPress is probably the other big player in the fight for the open web. It might actually benefit all of us if Automattic starts making a lot of noise about privacy.
And ultimately, its not as if anyone wants any of these tech giants to completely fail (well, maybe Facebook). What we want is to not have the nature of the web changed to suit the whims of a handful of companies.
This may be a case where the savings (i.e. money from selling data in return for a big pile of cash) are actually passed on to the customer - e.g. transaction costs are lower for Visa/Mastercard for merchants, which is why they sometimes don't accept Amex and Discover.
It would be interesting to see if Visa and Mastercard have some clause which prevents Google from buying data from AMEX and Discover (competitive advantage for them because then Amex and Discover's transaction costs aren't subsidized) in return for agreeing to part with their data. Said another way, what's preventing Google from also getting data from Amex or Discover? Its not as if Google is going to say no.
That's because they can't do much with this info since they aren't also a giant ad network. Now imagine the same data in the hands of a company which has built one. They should probably start charging more :-)
Indeed, it is amazing how many 'soft' things surrounding the software can be improved without requiring to 'get into the zone' as a developer. Of course, you also need to do the actual work and not get bogged down in these peripheral tasks.
This might actually be a pretty huge opportunity for a company which can hand hold the transitioning from Windows to Linux in an enterprise. After all, if the new Windows OS is provably non-compliant, shouldn't the enterprise customers be very willing to investigate this option? Are there already companies which do this?
I don't like AMP for no other reason than I don't believe there are any genuinely decentralized initiatives anymore.
"Oh, we open source it, you know", seems the common answer.
If an initiative is rolled out from the innards of one of the tech giants, and there are a bunch of other tech giants contributing to the initiative because it is open source, and most of the contributors and maintainers just accidentally happen to be also be employees of the tech giants, then stop and wonder about it for a while. And then reject it. That is, don't participate in the initiative.
Here is why: at the moment, the cost of open sourcing is minuscule for the giants but the benefits are enormous, and surprisingly often leads the entire tech sector down the path of greater oligopoly (Android being an excellent example). Another way to put it is, given none of the tech giants directly compete with each other in their respective core profit centers, open source is becoming a nice little platform (intentional or unintentional) for extending oligopolies.
There is no realistic chance that the open source code can be used by a competitor against the one who proposes the initiative (if you know of a counter-example, I would be happy to hear about it). But, there is every realistic chance that an initiative like AMP could extend a heavy toll on a genuine but small competitor in terms of code compliance (e.g. DuckDuckGo) and put them out of business.
But then, don't we all benefit from the nice byproducts of their technical innovations? In sum total, once you see the reduction in privacy, competition and decentralization of the web, probably not.
I don't disagree. If you have done this much research though, you could actually set up a little case study on this, write an eBook with nothing more than screenshots on how the price changes, and put it up for sale as something which will save people atleast $x per month. :-)
I wonder what traits will define the Facebook transcender like the Windows transcender [1] mentioned in PG's essay.
You can argue that since that essay (2005) the iPhone/iPad and Android and then eventually "all meaningful apps becoming web enabled" has mostly made Windows the last choice amongst the candidates rather than the first when it comes to selecting
personal computing devices. Right now, people who use Windows see it in the same way people see crusty old government bureaucracies - it is just one of those things you deal with because you have to, and move away from ASAP.
>> What guarantee do users have that they won't be individually selected to receive a higher price?
As long as Uber is not the only means of transport, this issue should take care of itself. Besides, as another comment points out, isn't this pretty much standard across industries?
Not sure what this has to do with Uber in particular which doesn't apply to any other company.
Also, think about what companies start doing when they start failing. And don't make implicit assumptions that these megaliths will never fail or stumble. What would you, as the CEO who is supposed to "maximize shareholder value" do? Are you going to say "Well, we could do some shady things to exploit the data we have and buy ourselves some more time, but it is a better idea to declare bankruptcy and close the company"? Imagine a company like Facebook nearing bankruptcy. You can bet on your life that they will offer the appropriate third parties greater control of their data mining prowess in return for a bailout. Is that also overall a Very Good Thing?
Also, notice how the 2008 financial crisis unfolded. None of the offending parties got anything more than a slap on the wrist, because they had the keys to an engineering construct of extraordinary complexity (the financial system). It was just "heads I win, tails you lose". The way our lives are becoming intertwined with these tech behemoths is no different, and I don't see it playing out any differently if there is a similar crisis in the tech sector. How can that be overall a Very Good Thing?