Amazon Employee: We Shouldn’t Sell Facial Recognition Tech to Police

I'm not clear if you're predicting a drought of Computer Science students or a drought of Fortune 500 company computer science hiring candidates.

This fact is subtext to a lot of the flamewars that occurred between these "greybeard hackers" and mainframe programmers on the one hand, and Windows PC programmers on the other hand. In both cases, the greybeards saw their counterparts working at IBM or Microsoft, or working at other companies developing software for z/OS or Windows, as people who were willing to "sell out" and work for "the man." (Not "the man" as in Microsoft/IBM; "the man" as in the government's various three-letter agencies.)

This was also the original subtext of Microsoft and Apple's fight over personal-computer market mindshare. Bill Gates and Paul Allen came from money, met at a private prep school, and shared a vision of selling their software to companies like IBM, who would drop thousands of copies top-down into enterprise partners' offices. Steve Jobs and Steve Wozniak were introduced because of a shared interest in illegal pursuits like phone phreaking and LSD use, and wanted—sort of like the One Laptop Per Child project—to empower individuals by putting computers in their hands. (Consider: they were probably envisioning use-cases exactly like "being able to share copies of the Anarchist's Cookbook over the Internet.")

Apple and Microsoft today are more alike than different, but they started off with diametrically-opposed cultures and corporate visions; and which ecosystem you bought into (at least back when) could tell people a lot about who you were as a person. There was a reason Apple found its first beach-head in schools: the schoolteachers of the 1970s shared pretty much the same countercultural ideals.

Read through the Jargon File, or the fortune(1) database, and you'll quickly get a sense of what "hackers" as a culture were like. A modern Stanford-MBA-alum YC applicant, who fancies themselves a "hacker", might be a bit shocked :)

What's to expand upon? Anyone will admit that there are good elements and bad elements to the way our police force operates. This boils down to the question "Do you believe our police force as it exists today has a net-positive impact on our society?"

I suspect very few people would say they believe they have a net-negative impact. And even among the people that say they believe the police have a net-negative impact very few people actually act as if that is what they believe.

Naw, dawg, when things get ugly, you have to resist where you can. I'm guessing you can't agree, but there's a fair number of people who think we're in a Germany 1937 situation. We have a president who in all likelihood enlisted Russian interference in the election, who shares plenty of rhetorical similarity to actual fascists. We have a Congress that is moving as fast as it can to stack the judiciary, which in turn will hear future election cases. Things are getting grim. I've lived and worked in places where these processes have played out - like Hungary - and anyone who says it can't happen here is kidding themselves.

I also disagree that we should limit this discussion to political. Companies (esp the size of Amazon) are hardly affected by politics, they are often judged more harshly by the court of public opinion.

Folks who want to prevent what they see as disaster should use all the legal mechanisms possible to enable discourse and change the course of events.

If AWS is non-neutral, will that hurt their ability to attract clients? Absolutely. I'm the first one in line to abolish ICE, but if AWS starts kicking people they don't like off of AWS I'll also be the first one to stop using it. Sure, we agree on ICE, what about companies that make violent video games? Companies that make firearm accessories?

You need to consider how broad "increasing lethality" is. The military is more lethal when the troops are healthier. Boom, you "increased lethality" by being a provider of combat meals or training equipment. By sending them less error prone weapons. By coming up with more aerodynamic shapes for the planes that consume less fuel.

If you want e.g. a tighter civilian review of military decisions so that they face tougher and more reliable consequences for unjustified killings, or less foreign deployment, then say that. Focusing on the "lethality of military" is the wrong target.

With that said, it's worth considering that Eisenhower might not have agreed with this:

> no one should be profiting from “increasing the lethality” of the military

Accepting the utility of a military tends to mean accepting that it needs to be competitive in order to achieve the foreign policy goals that go with having a military. This often means incrementally increasing its lethality, and some people are going to make money along the way. Unless we think metal refineries and medical supplies producers and everything else in the supply chain should be operated on a strict non-profit basis.

In short, you are still completely correct! It might be worth considering being careful to avoid overinterpreting President Eisenhower is all.

A clearly and incredibly flawed statement, and dare I say a successful attempt at trolling. Nice to know this is what takes up the first 50% of the thread /sc

Eisenhower’s Farewell Address https://youtu.be/OyBNmecVtdU

That's a really silly idea. A military that's bad at its job is a military that can't act as an adequate deterrent and can't defend its country or its allies. You might as well just save the money and disband most of it, and let the countries who kept competent militaries have their way with you and your allies.

The goal of the military is to apply force with as much precision as possible, to preserve the perception of legitimacy for audiences at home and abroad as much as possible.

Being lethal isn’t necessary to apply force; effectively applying non-lethal force is just much harder, so current military doctrine emphasizes lethality against combatants.

But against irregular opponents, whose goal is often to sow hatred for the opposing, powerful military because of their heavy handed tactics, being able to consistently and safely apply non-lethal force would arguably improve military effectiveness.

I agree that engineering lethal tools only by people directly employed by the military is one possible way to make this argument internally consistent.

Another way is to just decide that non-lethal force is a useful and legitimate tool for the military.

I say this as someone who has ranged from medium to full-throated opposed to essentially every adventure undertaken by the US military and CIA in the past 50 years or so.

The militants on both sides seem really bad in Yemen.

Militaries exist to project power onto foreign entities. Defense of national territory is but one use for that projected power.

Keeping trade lanes open--such as those through the Straits of Hormuz or the South China Sea--is another use. Enforcing a trade blockade or embargo is another. The size and lethality of your military is measurable proof of your commitment to a particular political issue.

Securing access to raw materials is defending the homeland. I live on an island, patrolling the shipping lanes is absolutely something our military should be doing. Standing ready to assist our trading partners on whom we rely for food and fuel is absolutely something they should be doing.

Consider the following example. Calling someone to ask them on a date is normal acceptable behavior. Doing it 20 times an hour, 3 hours a day, every day, quickly becomes immoral and illegal.

[1] https://en.wikipedia.org/wiki/United_States_v._Jones

what I'm calling for is the individual to be "armed" with adversarial patches or whatever other tool it is that confounds facial recognition sytems. the only thing which is long term effective is a population which can and does resist. VPNs, encryption, cryptocurrency, and a bunch of other tech (existing and under research) will be the safeguard when all else fails.

Even dumber than the "Hey, why should we care about the environment since a few countries can decide not to"

yeah i said that.

"so who cares"

no i didn't say that.

I said the best defense isnt the benevolence of corporations willing to say no to gov contracts on principal. the best defense is a populace able to subvert these technological systems of oppression. For facial recognition this means adversarial patches. for Internet censorship this means VPNs. in other cases it means encryption and cryptocurrency.

So in that sense, yes, they'll keep looking and someone else absolutely will. As someone pointed out below, the real solution is to have them lobby the US Government to give up their pursuit, anything else is just a delaying tactic.

Chinas social credit is a perfect example of what I mean. technology is making it more and more practical to actually enforce restrictions on civil liberties down to the granular level.

Better surveillance would catch and/or possibly stop car thieves, bike thieves, muggers, vandals, robberies, some rapes, life threatening drivers, hit and run drivers, bad cops.

I get that there's no magic wand to make sure the tech is only used for good but that's true of pretty much everything. Knives can be used to cut food or to stab people.

The way I interpret 'bad mouthing' is saying things which are false, or misrepresenting what Amazon is does.

I don't think the author made any misrepresentations of what Amazon is doing with their involvement but rather bringing to light that employees do not agree with their policies, that to me anyways is in a different category than bad mouthing.

> I'm not seeing any sort of backlash against the military as Americans generally have a very positive view of it.

That might be true. But the military is many things, serving is in many ways honorable, creating weapons from behind a desk -- not so much.

Regardless, in the long run we are likely to see the end of armed conflicts. Investment in military tech will likely drop. That might take 20 years, or it might take 200 years.

But war doesn't create economic growth, why bother getting involved?

Thank gawd they can't just go out and purchase any security / surveillance technology they fancy.

Typically a trial / vetting period is used to test a given system for accuracy as well as ability to integrate with existing systems, plus the over all expense of operating the system over time.

This is when Rekognition falls down, because the accuracy is too wide, the system requires human operators to make up for the lack of accuracy. Then the monthly is expensive to use it is high. Plus there is a perceptual disconnect officers have with the FR system operators required to maintain Rekognition: these operators more or less become FR specialists, with salaries to match, but their actual moment to moment work is closer to a security guard, with their technological knowledge only needed for moments at a time. Police department Directors have a hard time justifying these new roles, in light of the multiple new high expenses required to use Rekognition.

Speaking of, Pixel 3's new enhancement feature[0] is now mainstream. If these end up mistaken for originals, it opens up an evidentiary Pandora's box. Then again, much poorer quality image enhancement has been around forever now (e.g. "digital zoom").

[0] https://ai.googleblog.com/2018/10/see-better-and-further-wit...

These are all publicly accepted problems by now, so I'm not going to do a writeup for all of them just because one person asks for it.

Your downvotes change nothing, by the way. This is not a sinister outlook, it's already happening right now and I'm provably right.

We get more police and they get more permissions, including broader surveillance, and they're looking to adopt tasers like in the U.S. The last few years did this and hardly anyone is arguing it anymore.

The surveillance state you don't want is coming, and you are the reason.

Limiting the polices technology and resources is an argument that doesn't make sense to me. Especially coming from open source advocates. Smells of hypocrisy.

Wasn't it, though? I mean, laws are ultimately decided by vote (directly or indirectly). People are free to agree or not with death penalty, but it's hard to deny that being personally affected by a crime can influence our opinions on the subject. That's why death penalty still exists in several states.

Same with facial recognition tech. Despite the legitimate concerns about potential abuse from a government, people can acknowledge that it is an extremely valuable tool for crime investigation and thus law enforcement.

So in the end, it all depends on how bad people want an efficient police, and that can definitely depend on our individual experiences with crime.

People forget that, in the USA at least, the police investigate and pursue crimes that have already been committed. They do NOT pursue crimes that have not happened and do NOT scan for people who have not committed a crime. As you all should know, license plate scanners, for example, are looking for specific plates of people who have already committed crimes.

My government actively engages in different military operations around the world, even if I may disagree. For "ethical" reason the bank managing my pension fund refuse to invest in companies supplying the military my government use for missions I disagree with. So I have mixed feelings about mixing ethics and business.