You can support some uses of facial recognition without wanting it to be used to say ticket people for jaywalking.
That's a huge assumption that many of us (opponents to widespread surveillance tech) just don't agree to. I don't think it's possible to hand government this kind of capability, then limit it to a specific set of uses. It always expands in scope, covering more and more use cases until the folks over at Vision Zero[0] make an impassioned plea, "Suppose you could prevent 10 pedestrian deaths a year by enforcing our jaywalking laws better? Why wouldn't you want facial recognition software to protect them?"
Or maybe the bias-free policing people[1] put forward their argument that removing human cops from jaywalking enforcement will increase Equity. It would be a decent proposition! And you end up with every minor thing being automatically caught and ticketed.
That would suck.
"Slippery slope" may be a fallacy in formal logic, but it's damn useful in resisting the march into a dystopian future. Nip this shit in the bud. Make it require some effort to enforce the law.
[0] http://www.seattle.gov/transportation/projects-and-programs/...
[1] https://council.seattle.gov/2017/07/20/council-president-har...
It's not if the slope is, in fact, slippery. At least around here, one fallacy more common than slippery slope is the slippery slope fallacy fallacy - calling out real slippery slopes as fallacious.
Edit: just to expand on this, here's a press conference from earlier today (sorry about the poor sound): https://twitter.com/DarwinBondGraha/status/14001715920642416...
Here, Oakland, California's Chief of Police admits that police claims about molotov cocktails and crowd demographics that were used to justify the deployment of tear gas and rubber bullets against protesters a year ago actually had no basis in fact. The Chief explained that he received information to that effect via the radio, and then went out and repeated it to the public (via the media) without making any effort to vet its accuracy. (For clarity, he has only been police chief since February; at the time last year, he was a acting as a spokesperson for the department). It's arguable that the decision to deploy tear gas escalated the protest into a full-fledged riot; even if you don't think so, it certainly misled the public about the behavior and composition of the protesters, inevitably impacting policy debates and so on.
I feel this is a good example of why the police cannot be trusted with facial recognition tools; false claims can be used to to designate large numbers of people as criminal suspects, and that suspect status can then be leveraged to intrude upon their rights. Were California's interim prohibition on facial recognition not in place, chances are that it would already have been deployed to identify large numbers of legitimate protesters on the basis of initial false allegations (ie, lies) made by individual police officers. 33 officers have since been disciplined and no doubt civil litigation will delve deeper into this, but at present the police officers who were disciplined do not have to be identified, despite the fact that they were quite happy to lie in order to violate the rights of that same public.
https://www.ktvu.com/news/oakland-police-chief-apologizes-is...
I also think eyewitness testimony has many failure modes. If anything, it's probably less accurate than current facial recognition tech and biased in ways that are harder to determine. Yet I wouldn't want to ban all use of eyewitness testimony.
Banning facial recognition seems like overkill. Instead, it makes more sense to restrict it so that we can get the good parts (catching violent criminals) without the bad parts (oppressive surveillance state). Instead of banning all fingerprinting and DNA analysis, we have rules for how police can use them. Why not have similar rules for facial recognition?
Ballistics comparisons in particular been found to be nearly complete bunk, yet experts continue to use them in court.[0]
Tire prints are also pretty bad.[1]
We live in an age of unprecedented HD video surveillance. We don't need automated facial recognition to bring violent criminals to justice.
[0] https://www.sciencemag.org/news/2016/03/reversing-legacy-jun... [1] https://www.apmreports.org/story/2016/09/27/questionable-sci...
Why would I want it?
I wouldn't want to use it in such a situation because of the likelihood that the person who committed a crime is going to go free while an innocent person might take the fall for it.
Do you really want every move you make logged? It's an incredibly powerful tool to use for oppression. If someone in the government doesn't like you, all they have to do is watch you for a few days. You'll be sure to commit a crime sooner or later, such as jaywalking, or maybe you looked at someone in a suspicious manner, and then they prosecute you to the max.
To find the person who appears on video footage filmed in the building, you need to spy on everyone and then match the face from the footage against the face of everyone else walking about. All of those other people did not break into the building, yet for this to work, their privacy is expunged by having their movement and faces filmed, tracked and catalogued so that the FR software can cross-match them.
I think there is a privacy argument here. If I didn't commit any crime, maybe it shouldn't be that my face gets recorded in some database alongside my latest location.
also, if you are part of a minority that is frequently misidentified by this “tech” and you end up being harassed by this pos tech, do you still want it used?
The potential for abuse is too high.