Tell HN: Airbnb now opts in your name and photo to data sharing
I received an email today indicating AirBnb will now automatically opt-in user's photo and first name in order to fight discrimination via their "Project Lighthouse"
There is no public content about this change, but if you are logged in you can view this page and opt out:
https://www.airbnb.com/account-settings/privacy-and-sharing/...
Project Lighthouse
If you leave this setting turned on, you’re helping us identify and prevent discrimination from happening on Airbnb—and you’re taking part in an initiative to better understand how and where discrimination happens on platforms like ours.
If you turn this setting off, we won’t use your information, and we’ll remove it from future studies. You can change your mind at any time.
122 comments
[ 4.9 ms ] story [ 200 ms ] threadAs do most people.
Notification text: "We’d like to use your information to help fight discrimination. Visit your Privacy settings to learn more or opt out."
/s
Essentially a good chunk of people do care so little about discrimination, that it's difficult for them to believe that others actually even care about it.
In this case, at it's face, to see a statement that Airbnb is collecting data to investigate if their platform is being used for discrimination - and then reach the conclusion that the most logical reason that Airbnb is collecting data is not because they want to know if their platform is being used to discriminate, but that instead it's just a ruse; they don't care and really just want to collect facial data in order to look like they care about discrimination to gain some side benefit.
Somehow via reasoning to think that that was the most logical conclusion is pretty amazing. It conveys so frequently as: I care so little about discrimination that I can't even imagine that Airbnb actually wants data and cares about discrimination - there must be some other reason.
And again, I can't make that conclusion about the statement above (because we've not discussed it), but from other discussions with people that's what surprisingly frequently ends up being the underlying thinking.
A guy getting a job at an animal shelter because he thinks it'll help him get laid is virtue signaling. Someone adhering to a religious practice by fasting is virtue signaling. Someone changing their facebook profile pic to have a logo for some cause is virtue signaling. A company prompted by recent events wanting to investigate if there is discrimination on their platform isn't.
Mhkay.
While I believe one can be genuine in either case, I find it more likely to be "virtue signalling" when it's a company doing something on a hot topic right now.
From your response, and others in this thread I'm starting to believe that a lot of people that use the phrase 'virtue signaling' may not actually know what it means.
If Wikipedia article is anything to go by. ;)
Just to pick a nit, this is only virtue signaling if they rub it in somebody's face. There's almost a whole chapter in the gospels about it, Matthew 6:16:
> And when you fast, do not look dismal, like the hypocrites, for they disfigure their faces that their fasting may be seen by men. Truly, I say to you, they have received their reward.
Do you really want to build a race database?
Do you want Airbnb making determinations about the intention of your actions based on a secret race database?
The timing isn't accidental; they even state as much in the announcement. This is driven by current events. I wonder if anyone would give a pass to 23andme if they started oversharing DNA data "to battle covid."
This is a good model of most human behavior though. Creating elaborate systems of social norms then covertly coordinating on how to evade them or politically argue about how some group gets punished for violating them, while another group gets to violate the norms but reap benefits.
The economist Robin Hanson has developed a pretty comprehensive social science body of theory under the name “homo hypocritus” for this and uses it to explain a lot of behaviors (see just four examples below).
It’s totally reasonable to disagree with Hanson’s theory, but I think you are going way too far to treat it almost like paranoia or conspiracy theories or something.
In fact I think given the way corporate scandals, dark patterns, rampant privacy violations, etc. are so egregiously common and often go unpunished, it should be the norm to assume harmful intent from corporations unless proven otherwise. Classic studies like Moral Mazes would seem to confirm this.
Examples
- https://www.cato-unbound.org/2011/07/13/robin-hanson/who-car...
- http://www.overcomingbias.com/2013/08/inequality-is-about-gr...
- http://mason.gmu.edu/~rhanson/showcare.pdf
- https://80000hours.org/podcast/episodes/robin-hanson-on-lyin...
Some of us care somewhat (i.e. I'm proud to have convinced my boss to hire that female IT technician from another country who was mopping our floors. It took some convincing but he gave in. She got paid as much as us and got a career here in the west (I know because she later married a friend of mine). I've also tried elsewhere to help others to a career in IT, not sure how far I have succeeded in other cases, but in at least on case I gave away my well paid but boring job to make sure another less privileged man got a fighting chance. I visited him last year and he had usef his opportunity well. Oh, and I helped my neighbor down the street to a better job and tried to help his girlfriend. In the end she got a good job through other channels but I tried and I tried so in a way that was appreciated, not in a creepy way.)
All that said: that doesn't mean I won't say virtue signalling when something like this happens.
Oh. And the irony of me virtue signalling so heavily to get my point through :-) At least my identity is only known to me and Dan, so I can't cash out on any stupid internet points I get.
Almost everyone actually 'cares' about racism, at least in classical, crude forms. If someone calls a black person the 'n-word' - almost everyone 'cares' enough to know this is wrong.
But there are any number of people who Tweet this, say that, or even take some action that is utterly irrelevant to the material nature of advancing the cause, but a the same time, they are embellishing their own position or status.
'Personal Branding' is the living job of every actor, politician and political figure. That's 95% of their job. The statements they make are mostly about engendering to and audience.
A good example is the Gavin Newsom's wife, who has declared that she will be the 'First Partner' - and not the 'First Lady' of California because the term 'Lady' is 'exclusionary'.
I don't doubt there's materiality to her statement; she probably does 'care' about the issue on some level - but I doubt deeply that her actions matter and that this is mostly an opportunity for her to make a 'public statement' that the press will like, hop on, propagate, which endows here with 'progressive credibility' to possible voters. I significantly doubt she would ever do such a thing outside the context of the political lens. To boot - it's also counter productive because the term 'Wife' or 'First Lady' is in no way exclusionary. It's gendered yes, but no more so than we used gendered pronouns 'he' and 'she'. It's 'invented equality' that serves no purpose other than a nice bit of PR. That is 'virtue signalling'. [1]
A very excessive, fascist example of 'Virtue Signalling' from the Mayor of Oakland just this week. She initiated an FBI 'hate crime' investigation into some ropes hanging from a tree in the park. Those ropes were there to hold swings for kids, ironically put there by someone in the Black community. Nobody had complained or thought this 'kids equipment' was symbolic of anything, there was no public intrigue - but the Mayor took it upon herself to make a big fuss about it. To make the situation 'scary Orwellian' her public statement was that 'Intentions Don't Matter'. Read that again and consider the consequences. A swing set in a park in her view, is construed as a 'hate crime' irrespective of the fact that it's merely a swing set, not indicative of anything, placed there by a black man, in which nobody had any complaints. This is the Mayor of a major US city, who has initiated FBI investigations which could destroy people's lives, for absolutely no reason. [2]
The reason this is 'Virtue Signalling' - is because nobody outside of political theatre cares and the issue is irrelevant. Why was it only the 'mayor' who had to invent non-existent hate crimes? If there were any legitimacy at all to the situation, others would clearly be concerned. But there was no concern. The excessive, fascist reaction by the Mayor is therefore vapid, it's utterly political. It's an attempt to engender her credibility as a 'force against racism' - even at the cost of using the power of violence of the state over common sense. But like Trump supporters wouldn't condemn him 'if he shot someone' ... her supporters won't condemn her for her ridiculous display either.
A less extreme but more relevant example would be Alexandria Occasio-Cortez total rejection of Amazon's bid for a major HQ in NYC. Both the Democratic Governor, and 'far left' wing Democratic Mayor were both 'extremely' in favour of the opportunity, with Amazon to bring in thousands of very good, high paying jobs, along with so much increment...
And someone may rightly want to call out the people who are only doing it for the PR or the data collection, but we don't have visibility inside the company to know who they are in particular, so the blame goes to the abstraction of the corporation itself. Which it deserves if it hired those people even if it also hired some separate people with better motivations, because those people are still then going to capitalize on racism by marketing their claimed virtue to customers, or do bad things with the data.
A user can care about discrimination and still object to policies that overtly erode user privacy at the same time as they show solidarity with a certain movement.
Ultimately it's my name and my likeness, and I don't want it to be shared without my knowledge and consent. A company can't say "but I'm fighting discrimination!" and expect me to be cool with the violation of my privacy.
Not to mention that the data will be used to classify users based on race, defining people by their appearance rather than their actions, beliefs, and character. There is no guarantee that such data will never be used against the very same users it purports to protect.
You can't criticise anything that is purportedly done in the name of fighting racism?
a) Saying this program was to fight discrimination, and
b) Saying they weren't collecting new data or doing anything nefarious.
What it left out was clearly explaining what they _were_ doing. I had to parse text carefully and use my own familiarity with the industry to understand that they were going to decide my race based on my photo/name.
Without clearly explaining what the new data use is, you can't expect people to make an informed decision about opting-out.
Yup. This is exactly what I took it as. Between this and their other data sharing policies I decided to delete my account.
“ We’ll only use information you already share This project will address discrimination that's based on perception—so we’ll use first names and profile photos from hosts and guests to help us understand the perceived race someone might associate with them.
We’ll use it to uncover patterns of discrimination We’ll use this information to help us understand when and where racial discrimination is happening on our platform. Any insights will be used to help develop new features and policies that create a more equitable experience for everyone.
Information won’t be tied to your specific account We know your privacy is important, so we analyze trends in bulk and Airbnb won’t associate perceived race information with your account. We won’t use this information in marketing or advertising, and it will only be used for anti-discrimination work.
We consulted with and solicited input from leading civil rights and privacy organizations to guide us We know how delicate this work is—so we developed this work with support and input from leading civil rights organizations like Color Of Change and Upturn, along with privacy organizations like Center for Democracy & Technology, to make sure our approach is both thoughtful and respectful of your privacy.
You can opt out in your Privacy Settimsg”
Bad form, Airbnb.
> Where permissible according to applicable law we may use certain limited personal information about you, such as your email address, to hash it and to share it with social media platforms, such as Facebook or Google, to generate leads, drive traffic to our websites or otherwise promote our products and services or the Airbnb Platform.
> Please note that you may, at any time ask Airbnb to cease processing your data for these direct marketing purposes by sending an e-mail to opt-out@airbnb.com.
> In some jurisdictions, applicable law may entitle you to require Airbnb and Airbnb Payments not to process your personal information for certain specific purposes (including profiling) where such processing is based on legitimate interest.
> Where your personal information is processed for direct marketing purposes, you may, at any time ask Airbnb to cease processing your data for these direct marketing purposes by sending an e-mail to opt-out@airbnb.com.
https://www.airbnb.com/terms/privacy_policy
Hi Please do not use my information for any marketing purposes, and do not sell my information. I want to opt out of all the data sharing involving my data.
I am sending you this email based on your documented policy referred below:
Please note that you may, at any time ask Airbnb to cease processing your data for these direct marketing purposes by sending an e-mail to opt-out@airbnb.com.
Very sad state of user friendliness. /rant Please, I beg you don't sell my info.
Subject: Opt out of marketing for user <your-email>
Please do not use my information for any marketing purposes, and do not sell my information.
I am sending you this email based on your documented policy quoted below:
> Please note that you may, at any time ask Airbnb to cease processing your data for these direct marketing purposes by sending an e-mail to opt-out@airbnb.com.
Edit: apparently it was a misunderstanding of the operator... account reinstated, luckily.
I was so upset I lost all my data, trips, and reviews!
Immediately (30') after I got the first e-mail saying that my account was deleted, I replied to the operator and told her I just wanted to opt-out, not having my account deleted. She replied quickly (another 30'), said that she was sorry for the misunderstanding, and reinstated my account.
Sidenote: Apparently my opting-out request got redirected to another operator who should have got in contact with me, but 17 days after I am still waiting.
I have to imagine they considered making this opt-in and realized the vast majority of users wouldn't take action to increase sharing of their personal data. Despite the good cause, it is incredibly shady that they would implement this as an opt-out.
IANAL, but I would think that this would be a flagrant violation of privacy laws in many regions outside the US (this policy is limited to US hosts and guests).
Edit: I also did not receive an email, but found the toggle active in my account preferences after reading this post.
Whatever good intention might be behind this, it's really off-putting to me. They send your photo and name to a third party who is tasked with evaluating those to determine your race.
"will look at these photos and first names and indicate their perceptions—or, what race they think the information we shared suggests. They’ll share these perceptions with a specialized team at Airbnb exclusively for anti-discrimination work"
Presumably that'll be correlated to booking data to identify rejections. I can't think of any other use case.
Gotta say, if someone truly doesn't want me to stay somewhere because of my race, I'll take that rejection rather than force myself into a shitty situation.
FAQ: https://www.airbnb.com/resources/hosting-homes/a/a-new-way-w...
What's better:
1) agree to their new terms and then opt-out
2) specifically disagree to the new terms but not be able to opt-out?
?
I imagine they make it as hard as possible to do that.
From the email: "We know your privacy is important"
Airbnb should prove it by making this opt-in instead of opt-out (and not hiding the choice with a bunch of boilerplate). I expect these dark patterns from a shadier website, not Airbnb.
Edit: If you go through "Account settings" you are only given a choice to "deactivate" your account. To actually remove your information, go here: https://www.airbnb.com/privacy/manage-your-data
Most companies don't delete accounts when you request to delete account, they only disable your access to account. So, before cancelling I prefer to go into account, remove as much info as possible and change all other info before requesting to delete account.
AirBnb wouldn't even let me access my account until I accepted their new T&C and upon declining they just moved me to cancel account with no chance to make any changes to existing info.
But yeah, a lot of companies don't even do what the GDPR actually requires, let alone what people think it does.
"OK, send more data!"
Few people just opt in to having their rights violated
It's really only done because they know few people will voluntarily opt-in to something like tracking or profiling.
Given the usual "Your privacy is important to us" guff this makes their privacy terms look even shadier.
As far as the pandemic goes, Airbnb has a lot less turnover than hotels and there are mandated days between guests to ensure the virus cannot live on any surfaces. That’s much safer than you’re going to get at a hotel with a lobby that has hundreds of guests coming in and out every day. Half the things you pay extra for at a hotel are now useless (gym, room service) whereas the benefits of an Airbnb like having a real kitchen and more private space are worth more now.
MIT just had to remove a major AI dataset that was used to teach hundreds of AIs. It had numerous examples of black people and monkeys labeled with the N-word and has trained systematic bias into our algorithms. This type of bias cannot be fought by hiding from the problem. What Airbnb is doing is courageous and you are free to opt-out if you don’t like it, but taking a big public stand against an organization for actually trying to improve things for minorities in 2020 is pretty tone deaf.
I don’t think anyone would complain if it was implemented as an opt-in.
> but taking a big public stand against an organization for actually trying to improve things for minorities in 2020 is pretty tone deaf
From the email: "we’ll use first names and profile photos from hosts and guests to help us understand the perceived race someone might associate with them...We’ll use this information to help us understand when and where racial discrimination is happening on our platform."
So if I get rejected for trying to book a room, my brown face will be labeled "minority" and be used as a datapoint to help improve the company. I have every right to disagree with how Airbnb is going about this especially from a privacy perspective, and do so vocally.
Oh wow, first time a guy on the internet calls me a white dude. If you saw me in real life, or made any inferences based on my HN username, you would know better.
I apologize for being rude but to be honest it’s pretty frustrating that people would be so selfish as to make a big stink about it instead of just opting out. Do you know how many times Facebook or LinkedIn has opted me into research and sharing that I didn’t opt in to and no one even bats an eye most the time. Now that the research is about discrimination instead of advertising profits it’s some big invasion of privacy that we need to delete our accounts over. Not to mention that Facebook and LinkedIn typically don’t let you opt out at all.
Man, this comment and some similar sibling comments betray this mentality where racism is the ultimate evil and, thus, anything done in the name of fighting racism is justified.
In particular, multiple people seem to think you're not allowed to criticise Airbnb's approach to handling private data _because Airbnb purportedly is using it to somehow fight discrimination_. Nevermind that:
* There's no transparency about how this 3rd party will use the data
* There's no data on how effective their approach will be.
* The default is opt-in, not opt-out.
* Airbnb has done ethically dubious things before.
Yet, according to some comments, if you don't let Airbnb use your data, you're supporting the racists/being tone deaf.
https://www.airbnb.com/resources/hosting-homes/a/a-new-way-w...
I would also consider reading the white paper if you have some time:
https://news.airbnb.com/wp-content/uploads/sites/4/2020/06/P...
TLDR is that that information is stripped of Airbnb identifying information, asymmetrically encrypted before perceived race is determined, then noise is added when the data is returned to Airbnb to prevent identifying individuals. This data is only used for identifying acceptance rate disparity.
Isn't that going to make it basically impossible to investigate false positives? This is already a huge problem with racial disparity numbers in general. If you don't account for things like income or education level you see huge racial disparities everywhere, but if you do they either get a lot smaller or go away entirely. It would be naive not to expect similar factors affecting housing determinations.
You also have the problem that the average guest is going to stay for a few days, so even a listing that stays fully booked is going to have maybe a hundred guests a year. ~12% of the population is black, so that's about 12 people. How are they expecting to draw any meaningful conclusions from a sample size that small?
You can just A/B test it with random changes until it goes away, but that doesn't mean you solved the problem, only that you made the metric a target.
If the problem is that some of the hosts are overtly racist, you could make the disparity go away by creating an algorithmic bias the other way, but that's only increasing the overall unfairness. The racists are still there harming the people they harm, then you introduce an additional harm to some entirely different innocent people in unrelated transactions. That doesn't help the people originally being harmed because the undue beneficiaries of your changes are completely different people who just happen to be the same race as the original victims, which makes the numbers balance even though overall unfairness has gone up rather than down. The equivalent of "solving" cops killing disproportionately many innocent black people by having the cops kill more innocent white people.
Meanwhile if the problem was algorithmic bias to begin with then you still have to understand the specific means by which the bias operates, otherwise you're susceptible to doing the same thing. Your algorithm was improperly disadvantaging Chris to the benefit of Chaz, you modify it to additionally improperly disadvantage Anna to the benefit of Alicia, and now your numbers balance because Chaz and Anna are the same race and the harms cancel out for your metric, even though they don't cancel out in real life for the people affected because they're different people.
And if the problem was that you weren't correctly accounting for confounders then there wasn't a real racial disparity there to begin with, and by making the un-adjusted numbers balance you created one.
You have to actually understand the causes and mechanisms before you can devise a solution. Just having aggregate statistics doesn't do it.
I'm inclined to think that it actually makes it worse, because then people only care about the statistics. But the statistics can show a disparity when everything is fine because it's just confounders, and the statistics can show balance when everything is not fine when you're making multiple errors in different directions that sum to zero in aggregate but not for the people affected.
https://arstechnica.com/tech-policy/2009/09/your-secrets-liv...
Plus they essentially admit that anonymizing the data makes it less useful and their claimed solution is to use a larger sample size. Which is like offering "buy more fuel" as a solution to poor fuel economy. It's no solution to the efficiency loss (at any given sample size it's still worse) and you may not always be able to get a larger sample size.
None of which addresses the problem I identified anyway, which is related to identifying the cause of the disparity once one is discovered, which is already almost intractably hard even without anonymized data.
The much better solution is to identify specific instances of discrimination and address them (and the mechanism of discrimination they represent) regardless of what the statistics say because, again, aggregate statistics can both say that something is wrong when it's not and that nothing is wrong when it is, and the only way to tell is by looking at the individual cases.
https://www.airbnb.com/privacy/manage-your-data
I theorized about Airbnb inferring demographic information from Facebook likes in 2015 because it was trivially easy: https://caseysoftware.com/blog/social-apis-for-social-evil
It was confirmed they were doing it in 2018. If you used social login with them at all, they have all that info and they're supposed to delete it (according to Facebook's AUP) but effectively impossible to verify.
https://i.imgur.com/6U8JiER.jpg