[fivefives55555]

> I think the difference between automated image recognition and security guards with clipboards is one of scale. It would simply be impossible to have images of lawyers from across 90 firms on clipboards and would take long enough to be infeasible for every single patron coming through the doors. Automation makes this possible and easy and thus it’s worth having the discussion.

Sure, I definitely see that. What I am having a hard time understanding is what, then, are people who are against the use of facial recognition in this case but are not against businesses being able to bar access to unprotected cases in general, arguing here, exactly? That businesses should be allowed to bar as many people as they can manually?

Basically the position that "businesses shouldn't use tech to help them bar people, but should still be able to bar people" seems confusing to me, because then obviously bigger businesses like MSG can still do even manual barring 'at scale' (e.g. hire more security guards and distribute lists of barred individuals across them) that small businesses wouldn't be able to do, etc...

[donmcronald]

> Basically the position that "businesses shouldn't use tech to help them bar people, but should still be able to bar people" seems confusing to me.

It's a combination of things. The high scalability and lower cost make it much easier to ban people. Big companies are already demonstrating they're willing to ban people for unreasonable things and the continual reduction in competition between businesses means people that get banned have fewer alternatives, so being banned is a much bigger deal than it used to be.

Take it to an absurd extreme and imagine if you walk into a store where they have facial recognition identify you, look up your bank account balance, and decide you don't have enough money to be worth letting in. The potential profit is less than the average cost of serving someone from your demographic, so it makes business sense to refuse service.

The venue matters too. There probably wouldn't be a lot of complaining if a Rolex store discriminated against poor people, but what if grocery stores did it?

I don't want to have some algorithm giving me a pass fail score that determines where I can go and what I can do and it sure feels like that's the direction we're heading.

[vlovich123]

> then obviously bigger businesses like MSG can still do even manual barring 'at scale' (e.g. hire more security guards and distribute lists of barred individuals across them)

Are you sure about that? Image recognition isn't a horizontally scalable task if you're talking about guys with clipboards. And the larger that clipboard, the lower the accuracy gets. Think of it as you have to match each incoming patron against a database. If each security guard has the entire database, they're doing a costly time-intensive, & inaccurate search through the book (security guards are not only paid too little to care about this, but humans in general are going to perform poorly at a task of "does a picture of this person exist in these other 200 images of banned people taken in alternate lighting conditions, clothing, facial hair, etc". And remember, as a business your goal is to let in all the people who came in a timely fashion. Otherwise your business dies / people complain / laws get changed. So to do this, you'd basically have to have a close to 1:1 relationship between person entering and security guard screening. That doesn't scale, even for larger businesses.

Now you could invert the problem. You have the security guard take an image and distribute it to other guards who have their portion of their database that they're combing through and match against a subset of that. Still, that's a heck of a lot of extra man power that scales with the amount of bans you hand out and there's a natural backpressure for your business.

With automation though, you can scale this for pennies. So you don't have to be very discriminant about who you're banning which is a meaningful distinction. This isn't a bad PR approach. You probably can't muster enough political will to amend the laws to force venue owners to only ban people from public venues who themselves are being a public nuisance of some kind (which I think solves the problems a heck of a lot better than protected classes). But you can make it financially infeasible for them to do this by banning the new tech that lets them go after you.

[fivefives55555]

> If each security guard has the entire database

What I meant was that each security guard has a certain subset of the database, so they're only responsible for scanning for those people. Say, instead of having 10 security guards each with a list of all 100 banned people, each security guard has a list of 10 people.

> Now you could invert the problem. You have the security guard take an image and distribute it to other guards who have their portion of their database that they're combing through and match against a subset of that. Still, that's a heck of a lot of extra man power that scales with the amount of bans you hand out and there's a natural backpressure for your business.

I think that's then an open problem of 1) how many people can one security guard reliably track, 2) how many people are on the ban list, 3) is it cheaper to hire that necessary number of security guards for each event than it is to invest in facial recognition tech? 4) How much cheaper or more expensive is it, exactly?

But my main question wasn't about the technical implementation, it was once again about what the actual argument is. Is it that businesses should be allowed to bar people as long as they don't use technology to do so? How would that account for different businesses having wildly different enforcement capabilities?