Google+ names policy discriminates against American Native Indians
randomtechnicalstuff.blogspot.com
randomtechnicalstuff.blogspot.com
http://www.kalzumeus.com/2010/06/17/falsehoods-programmers-b... (was dead, now live, if dead again it's in google's cache)
However, until Google's (and Facebook's) absurd name restrictions (or even the idea that "proper" "official" names exist at all for humans) start seriously affecting their real customers (advertisers), they will not give a shit.
Another link, this on the consequences of real names policies: http://geekfeminism.wikia.com/wiki/Who_is_harmed_by_a_%22Rea...
E.g. if you want to be able to greet a user, ask them how they want to be greeted, rather than try to guess at which part of their name a user prefers to be addressed by and how to extract that part from their full name.
"People have, at this point in time, exactly one official name." <-- phrase is false
"There exist at least some people who will, at at least some instants in time, have either zero or two or more official names." <-- phrase is true
Now multiply that sort of circumlocution by 40.
I frequently receive the feedback "I wish you had identified a solution", which suggests to me maybe I missed the opportunity to title the post "Names are intractable hard and any programmer who thinks their program has solved them is wrong." I thought that was pretty explicit in the first paragraph but people seem to really WANT this problem to have a Ruby gem available such that they don't have to worry about it. I want dragons, but will live with my disappointment.
The problem with the google+ real names policy isn't that their algorithms aren't smart enough (though it seems that they could use some work). The problem is that they're too arrogant to recognize that their algorithms are unreliable and indifferent to the ways that this unreliability is bad for users.
No sane person thinks it's a good idea to just roll their own time zone implementation routinely, but people usually think rolling their own name handling system is a good idea. The purpose of patio11's advice is specifically to discourage people from doing that. And fortunately for names there are some reasonably good options, such as avoiding trying to handle names semantically at all and instead leaving them as more or less freeform user data.
12. People’s names are case sensitive.
13. People’s names are case insensitive.
32. People’s names are assigned at birth.
33. OK, maybe not at birth, but at least pretty close to birth.
34. Alright, alright, within a year or so of birth.
35. Five years?
36. You’re kidding me, right?Absolutely not!
Case sensitivity varies amongst countries.
One very simple occidental example: in France names are usually written LASTNAME Firstname
So if you type Lastname and your DB doesn't normalize the capitalization of names you won't match
So for the last name the search is case insensitive but for the first name not so much.
Another example is Turkish, where there are two capital I's, one with a dot, one without.
And not alphabets have a notion of case.
It's even more complicated: they have two letter Is, one with a dot in both lower and uppercase (İ/i) and one dotless (I/ı) in both cases. So any automatic case change needs to take this into account or the letter will get switched. This also make using fonts with fi ligatures problematic.
Take SQL as an example: are programs case-sensitive or case-insensitive? Neither!
For electronic systems, I pretend the apostrophe doesn't exist. I'll still use it in writing, but if it's going to be an entry in a database, I leave it out.
Some places I'm Sternal-Johnson, others SternalJohnson, others Johnson, others Sternal.
REALLY? Computers exist to bend to our needs, not the other way around!
... of course, that era was a long time ago. :)
My guess is neither. They're probably very smart people, and they may just arrogantly assume that their specification of their algorithms are correct, are correctly implemented, and therefore produce correct results. All three assumptions are clearly and demonstrably wrong.
That said, good intentions and brains can inadvertently produce racist and incompetent results.
If they want to maintain a real name policy for the whole of the planet? They're going to have to understand how people from many different cultures and nations identify themselves.
It doesn't help that Google simply refuses to let people talk to humans, and anything that doesn't meet with their algorithms is simply banned with no recourse other than to create bad publicity for them.
So, they supposedly know the problem they are creating for Native Americans, so can we assume further occurrences are racist or just more incompetence? Will someone with the last name "Afraid Of Ravens"[3] get their account suspended or not even able to create it in the future?
1) http://www.att.com/Common/about_us/public_policy/ATTLetter_F...
2) http://arstechnica.com/tech-policy/2009/10/google-voice-were...
3) it's an actual name if you switch the bird.
A lot of very common names are also dictionary words (Jack, Jon, Dick, Smith, Smart, Baker, LARRY PAGE, etc...) so someone clearly sat down and figured out an algorithm to allow or restrict certain words. Whoever came up with that algorithm should have quickly realized they didn't have a good way to distinguish real names from fake names.
My vote is incompetence.
The name seems pretty real to me. Could you please explain what about it would indicate it was fake?
Yes, using fancy methods to get a wrong result is still wrong. Let's also remember we are talking about a company that indexes the web. Perhaps they could use that data to test.
For all we know that might be a real name too, but all we can do is make an educated guess about whether or not it is. My point is that determining if a name is real or not is pretty arbitrary so if your objective is to verify if a name is valid or not, you simply can't be 100% accurate.
A wrong result is wrong, but making the best of what you have to try and get the right answer most of the time (while still being wrong sometimes) is in and of itself not wrong. Take any NP-hard problem, the best you can do is use heuristics to try and approximate a good solution, since the "right result" is simply impossible to compute.
Nope, I grew up around people with adjectives in their name. It is very, very common among the plain tribes[1]. I tend to be suspicious of people named "Smith" or "Johnson" since it seems like a common dodge on forms.
I am troubled that you think having an adjective in a family name should be cause to trouble someone. I realize European names have suffixes added in place of adjectives, but that should not blind people from other ways names are assembled. I am also particularly unhappy with an American company that has heuristics that flag a specific ethnic group in the USA.
"the best you can do is use heuristics to try and approximate a good solution, since the "right result" is simply impossible to compute."
Then don't friggin do it. Google is a SEARCH company and has indexed a huge amount of the web. Running names through their damn heuristics should have shown a problem. If you google "Yellow Horse" you see multiple examples of people with that family name on the first page of results.
Google especially shouldn't try it since they don't have support needed to handle the exceptions without people going to the press.
1) not sure about south or east or 1st Nations
The reason it is a stupid policy is Google has no intention of making sure every person signs up with their "real name."
So they create automated rules about what a "real name" is. And that's dumb.
Adolf Hitler is not his real birth name..even than me Hitler is somewhat a non-existent name as its not spelled that way in German anyway.
People's names are imprecise due to the language and cultures underlying that naming.
Interestingly, it's somewhat undone by one of their other bad ideas: merging Youtube and Google+ accounts. Because nobody had a real name for their Youtube account, and suddenly Google creates a bunch of Google+ accounts that nobody asked for and are obviously pseudonyms.
So, I think it's a fairly common thing that people, with absolutely no intention of doing so, accidentally create a system that disadvantages some group of people compared to others. My question is, what term should we use to describe this action? I want to be able to bring this problem to the attention of the people performing it, without making it seem that I am accusing them of ill intent.
Quite a few people (myself included) are/were of the belief that "discrimination" is the correct term to apply in this context. (To us) the word "discrimination" is a description of effect, not intent. In a technical sense it only means treating two groups differently (it's a common term in systems analysis), without implying judgment.
Apparently, some people think that the word "discrimination" does imply intent. That's fine, words mean different things to different people, and I'm glad I learn that. But, that leaves me in a bit of a quandary: if discrimination is not the word to use here, than what is?
Right now I'm leaning towards compound terms like "accidental discrimination." I've seen "indirect discrimination" used elsewhere (as a UK legal term), but I don't like that one. If "discrimination" implies motive, adding "indirect" to me implies being sneaky about it as well.
(Some part of me thinks we should just give everyone a GUID or something and be done with it...)
I guess my big problem is the thought that "Yellow Horse" triggered your automated systems in the first place. It's not like "Yellow Horse" is that uncommon a name with the Lakota. You would think with the obvious resources of a search company that had to scoop up tons of addresses and vCards from the internet, Google would have found out some Native American style names. Did some programmer think they were just fantasy names and ignore it or did they not test this thing.
It was bad enough with the Google Voice crap, but this is just too far. Has Google made an official comment on it?
I've forgotten the multiple names academia has come up with for this situation. But in a nutshell, Google, you are conflating two "problems" you are trying to solve: "True" and "Names".
We could and do debate the merits of these separately. E.g., I think your "true" initiative is kind of full of shit -- based upon my and others' experience on the Web.
But even if you want to and decide to pursue "true", that doesn't give you any authority to judge and decide what is, or isn't, a "name".
Perhaps your overwhelming mountain of data and analysis capabilities might enable you to establish a leading position in some de facto opinions on the matter. But, as the OP example demonstrates, even there you are most obviously failing -- it appears to us, the general public, to be in good part due to a lack of real effort to analytically and functionally back up your "initiative".
So... as a result, once again your "initiative" and "policy" looks to be full of shit.
Just drop it. Treat names like the name qualifier in email -- or I guess that is now, for Google, a field within Contacts: People enter a string, and that's the name. Maybe some subsequent processes try -- more or less successfully -- to parse it into components. But at the base level, it's an arbitrary string.
There, that was simple, wasn't it?
(Now, focus some effort on keeping that name from unintentionally bleeding across properties via Google+, etc.)
Of course it's not malicious. This guy is not good people.
http://www.equality-law.co.uk/news/106/66/Types-of-discrimin...
>staff are either incompetent, or racist. I hope for incompetence. Either is bad, but racism is much worse than incompetence.
>protected characteristic
http://www.legislation.gov.uk/ukpga/2010/15/section/4
http://www.equalityhumanrights.com/advice-and-guidance/new-e...
Does that include names?
> Does that include names?
IANAL, but the specific discrimination against Native American or First Nation names could fall under race. Race in terms of the Equality Act is more than just skin colour: it includes nationality, or ethnic origin.
Or even belief, which "need not include faith or worship of a god or gods, but must affect how a person lives their life or perceives the world", according to this page from the EHRC:
http://www.equalityhumanrights.com/advice-and-guidance/furth...
Our names do "affect how we live our lives": we have to use them every day. They are a way of defining who we are.
I don't think it's a stretch either to say that many cultures place a great deal of emphasis and tradition on names. I'm no anthropologist, but variations on naming ceremonies exist throughout the world -- christenings, for example.
Also see the other six points under belief on the EHRC page above. They arguably protect the right of Native Americans to use their own names, and not be denied service because of them.
Also, continuation of accidental discrimination after being made aware of it becomes purposeful discrimination. If google doesn't fix it they can't use the "oops, we didn't know" defense for much longer.
I agree that it's a stupid policy, but calling things which are not really racism racism only dilutes actual racism and victims of racism.
"Google+ naming filter thinks American Native Indian style names are not real names"
For example, consider the study "Are Emily and Greg More Employable than Lakisha and Jamal?" [1] They found that resumes with black-sounding names were less employable than the same resumes with white-sounding names. Is that because a large number of hiring managers are also KKK members who put on their hoods before sorting applications? Probably not.
A more likely explanation is that, as Harvard's Project Implicit [2] demonstrates, a lot of people are consciously anti-racist but still have racist attitudes and reactions. If you live in a segregated society, it need not even be a bias against blacks, so much as a bias for people who are more familiar to you. Is that effectively racism? If I were somebody not getting hired because of the color of my skin, I'd sure think so.
So in the case of Google Plus, I doubt the mainly white, mainly male, mainly millionaires sitting in a suburban office park were consciously saying, "At last, our revenge for Custer!" But them saying, "Looks good to us!" and then enforcing it is effectively saying, "We believe our personal biases and local cultural norms are good enough for the entire world." That might not be explicitly racist in intention. But I don't think that's required to create a racist system.
[1] http://www.nber.org/papers/w9873 [2] https://www.projectimplicit.net/index.html
I don't know your racial category, but people of color frequently mention how white people aggressively like to redefine racism so it doesn't include their own racism. This particular argument is common among them. Because after all, racism is one of those -isms which are a fundamental part of many cultures (like the US).
And white people do the same thing about people of color, pointing to the not-infrequently-made redefinition of racism that anyone who isn't in the currently-dominant race cannot, ipso facto, be "racist" [1].
The fact that people of all races can point to self-serving and ridiculous attempts from people of other races to redefine their own racism as "not racism" does not, however, make the very idea that the term must have some boundaries to be useful invalid.
[1] for an example of this argument see, e.g., http://wwsword.blogspot.com/2008/03/why-black-people-cant-be...
"It's just a joke." - In defense of the slew of racist jokes that appear every time a POC is mentioned or seen.
"I don't hate black people, I hate black culture." - Every time a POC does something bad, or merely annoying. Mysteriously, nobody hates white culture.
Man, that's been the exact opposite of my experience.
Obviously unscientific, but it's interesting to compare the difference between https://twitter.com/search?q=white%20people&f=realtime and https://twitter.com/search?q=black%20people&f=realtime .
If you do that for names of German origin (with a German word list) there will be hit after hit
Google apparently doesn't know much about names.
Discrimination does not require intent.
It's why stop and search that targets a large number of young black men is seen as racist.
It might be incompetence the first time but this has been going on for such a long time now that that is wearing a bit thin as an excuse.
Not to mention
Admiral Sir Reginald D'Arcy ffolkes-O'Harris
And in college I knew someone with a name very similar to the name in the article, a "normal" first name, and family name <adjective> <animal>.
There are enough past incidents of “unusual” names being erroneously flagged to suggest it's no about irrational human discrimination but about imperfect mostly automated processes.
The policy is about using “real names”, the automated system was obviously tripped and the appeals fell through, imperfect systems don’t imply malice.
Seeing that eventually the issue was resolved and I assume they’ll be more mindful of native American names from now on, is there really a need for name calling and further escalation?
... but they can still discriminate. In some jurisdictions, such as the UK, intention or malice is not required. If Google has a policy or system that disadvantages a group of people who share a protected characteristic, it is discriminating. This is defined in law.
http://www.equality-law.co.uk/news/106/66/Types-of-discrimin...
Not to make a statement about your racial category (because I don't know it), but more privileged people irrationally fear claims of racism because it's one of the few ways they feel vaguely limited. (Of course, the minor social consequences they fear are nothing in comparison to less privileged groups who are regularly imprisoned/invaded/starved.)
I hope that someone from Google is reading and takes note of this. The Real Names policy has actively discriminated against many people in the past. They were meant to sort this out, but they have not. It behooves Google, who want real names, to accept real names. And to understand that even if the names are a little strange, they are all too often actually real names.
I once heard of a couple in Kent who called their child "Depressed Cupboard Cheesecake". The point being is that there are a number of oddball names out there. How is Google going to tell what is fake and what is not fake?
Perhaps, but this isn't Google's first screw-up where tribes are involved[1].
> imperfect systems don’t imply malice
Imperfect system do imply malice when the only appeal that works is embarrassing a company via PR.
> I assume they’ll be more mindful of native American names
That is a rather large assumption given their customer care history.
1) http://www.att.com/Common/about_us/public_policy/ATTLetter_F...