Lyft’s algorithm is trying to block people with names like ‘Dick’ and ‘Cummings’
theverge.com
theverge.com
With a userbase as big as Lyft's, I'm sure there were a ton of obvious true positives (anyone named "Fuck", ostensibly). I just can't believe they didn't notice a surname as relatively common as Cumming/Cummings. Yes, the apparent naivety of the regex is an issue, but this seems like a system for which a lot of testing on actual data would be easy and natural to do as part of the QA process.
I once read a story of a Vietnamese man named Hui. He had to go to the court to protect his right to be named this way in Ukraine where the word is used widely and has only one meaning - a dick.
There was a study linked to on HN a couple of years ago that found that babies recognize swear words even before they have language skills, and that people can sometimes recognize swear words in languages they don't speak. It has to do with the tone and sharpness of the sounds.
The baby wouldn't experience the father's name as a swear word unless the conversation was for example about the dicks new girlfriend and his wife's assessment of the matter.
Dick vs DIIIIIIICK
If I yell insults at you, you are saying that doesn't mean anything? And if you get insulted, it is your own fault?
No, words have meaning based on a common understanding between the person saying the word and the person hearing the word. If I insult you and you understand it, it is me who did the insulting, it is not 'your problem'
You should try this argument out in court: "No, your honor... I can't be guilty of threatening to kill someone.. sure I said, 'I am going to kill you tomorrow at 5pm', but those were just words! If the person felt threatened, that is on them!"
Yes. It's just sounds that you hear, then your brain recognizes and turns them into words with meaning. How you react to the meaning is entirely within your consciousness. This is how you might pick up words from languages you don't know if the sounds are similar, yet you wouldn't claim to be insulted by them.
> "If the person felt threatened, that is on them""
Yes, your feelings are your own, nobody else controls them. That's one of the fundamentals of free speech. There are some very limited situations (like threats) where it's overruled, not because of your feelings but because of the intentions being forecast. Even then, there's a very high bar to prove any criminal intent there.
Considering that it is often very easy to guess how people will “react to the meaning” of the sounds you make, it’s often considered polite to refrain from making them when they would probably offend.
In such a case I would just assume you are stupid or something and keep on my business. If you don't have capacity to ignore that kind of noise you probably need treatment. Threats, however, are a different story.
You shouldn't care much if some random people express their opinion which is you're a moron (unless you meet such people often - then it might be a good idea to actually make sure there is nothing wrong with you or you are not hanging out in a wrong place) - ignoring insults usually causes better consequences than reacting to them. At the same time you usually should be worried if somebody threatens to kill you, ignoring threats can often get you in a real trouble.
The entire discussion here has been about being offended by a word, even though that word is used in a context which is not offensive at all.
Some people are trying to insult and offend, so they set their names as offensive and insulting names... in an attempt to stop that, Lyft messed up and blocked non-offensive names that share the same words as the offensive ones.
That doesn't mean words can't be insulting and offensive, it just means context matters and algorithms have trouble with context.
Yes. You should check out Seneca, and/or Epictetus.
That's what you think the world should be like, not the way it is. If your name is Adolfhitler Smith, a perfectly legal name in the US, then you will have trouble getting hired.
That's a made up example, but there's evidence[1] that unusual (not necessarily obscene) names will lead to less success in job applications.
If you have a "bad" name, it's your problem well before anyone else's.
Good luck using the N-word in the US.
The point of the comment above is that the meaning is informed by the listener, you can import all the traumas and neuroses of those around you or you can listen mindfully and limit the impact to whatever is the speaker's intent and capacity to harm you.
Which makes sense. There's likely a lot of drivers who want to go by "Tom" instead of "Thomas" or whatever their full official name is – which is even more important for drivers with very uncommon names who want to go by something more familiar/pronounceable. There's probably a decent safety argument for not requiring drivers to have their full real names be accessible to the user, especially in cases where a driver has an uncommon (i.e. easy to doxx) name.
The privacy of the driver is outweighed by the need for the public to feel they will safely be taken to their destination in an enclosed vehicle totally controlled by another party.
You can definitely argue that they should show their real name, but it is not a good argument that because taxi requires it, lyft should as well.
https://data.cityofnewyork.us/Transportation/Medallion-Drive...
In any case, rules are different for different jurisdictions and entities. If NYC TLC did have that requirement, then it's possible that real name display is mandatory for Lyft drivers operating in NYC, given that TLC has some oversight over services like Lyft. However, TLC also has some say over livery cabs. It's been awhile since I've been in a livery cab, but I think they do operate under different requirements in terms of what's required for display.
Albeit, this is rarely used to block payments, because if they stopped payments when customers didn't put their middle name the exact way it's written on their card, nobody would be able to pay online.
This guy would disagree:
http://mw.eco.br/ig/prof/ReinhardtAdolfoFuck.htm
Along the same lines,
They are as dumb and undesirable as incomplete regex email filters in web forms.
Every single one ever written will have this problem. And ironically, absolutely nobody reading a fart joke will ever be offended. But the millions of people who are denied service and not twitter famous enough to reach a dot-com employee are very much offended.
...Also, the single reason to add those auto-censorship features is to please advertisers, not to protect users/drivers from fart jokes.
Expect Advertisement on lyft very, very soon.
You do see their first/given name for identification on pickup.
In general, just because someone is providing a service to you (e.g. driving you, providing customer support, etc) doesn't mean they should be unnecessarily exposed to offensive language.
The flower Fuchsia is named after someone called Fuchs, and therefore the correct pronunciation is "fook-seeah", not "foo-shah".
BTW, isn't it insulting to those people to suggest that their names are insulting?
Implying devs test their code thoroughly. lol. After working at numerous companies (some with high bars of entry) I've noticed many devs push changes without testing properly. Or, if they test, they rarely consider edge cases or implications of what they're doing and only test the obvious base cases.
"I'm so sorry my name is offensive!"
To this day, emails from that account have "Ryan <lastname>" in the header.
This is no the only use, of course, but that doesn't matter. Good luck being a Hindu and displaying a swastika.
Regardless, your point still stands. The product manager should have gone through whatever process was needed to pull down a limited set of production data to test against. Or the engineers could have logged exceptions in prod before making the filter active.
The Simpsons figured this out decades ago: https://www.youtube.com/watch?v=M1EjcWU6sEk
What for?
And how does any other environment where people introduce themselves to each other fix this? I don't think they see it as a problem in practice.
There are many terrible things in the world. People naming themselves Mike Rotch isn't one of them. Certainly not as "terrible" as denying hundreds of thousands of people a service.
I've never heard of this. Where did you learn this? I'm not saying you're wrong either, as it makes sense. I've just never heard of it.
It's useful for busier locations, where more people would call for a taxi at the same time.
He had his own flying car, in 1961!
They could just generate the list of questionable names and get support to give them a call about it or institute a rider identity verification programme.
The intentions can be noble, such as a Lyft employee seeing an user named “F* N*” and thinking, hey that shouldn’t be okay.
The actual consequences can be meh.
It always amazes me when US Americans talk in terms of human races like it's the most normal thing in the world. And no, “Cummings“ is not a “Caucasian name“ (whatever that is supposed to mean), it's an English name. I don't want sound too harsh, it's just that as a Central European with all our recent history, things like this really rub me the wrong way.
Because that's exactly how some people teach the definition to mean.[0] I've always been told that to the people in Persia, the people that came across the Caucasus were much more pale skinned. Teachings also say those same people helped populate Europe. [0] https://en.wikipedia.org/wiki/White_people
I question the wisdom of telling native speakers that their language uses a word wrong/that doesn’t mean what “they assume it means”. Stuff in AE (and other language) means exactly what AE speakers (and speakers of other languages) assume it means.
Cau·ca·sian /kôˈkāZHən/ NORTH AMERICAN white-skinned; of European origin.
¹ https://en.wikipedia.org/wiki/File:Portable_radio_SCR536.png
Namely: Why would it be called "white EUROPEAN" when basically none of the people it is applied to are European in any reasonable sense of the word? They may have great-great-great-grandparents who were European, but that does not make them European.
Americans do this all the time - they call themselves "german", "irish", or "italian", even when they are not born there, do not speak the language, have never even visited the country, and have zero exposure to the culture of that country.
In America, a nation of immigrants, people are often proud of where their family came from. This can trickle down a few generations. There is an increasing number of people that identify as "American" instead of where their family originally immigrated from, particularly in Appalachia and the South, so this could eventually change.
I understand the frustration from a European point of view, but I just wanted to provide some context to the phenomenon.
Even when that’s true—-and it often isn’t--they usually mean that they’re part of a specific American subculture.
Italian-Americans, for example, traditionally celebrate Christmas Eve with a “Feast of the Seven Fishes.” This dinner isn’t traditional in Italy, but it’s also not common in other American cultures either. An Irish-American family might have a turkey instead but they’re more likely to “observe” St. Patrick’s Day by eating soda bread and corned beef.
It's a weird mindset, and it's definitely tied in to the toxic streak of white supremacy that runs through all of American culture.
Yes, but it has been completely abandoned by the scientific and political mainstream for decades now. Using the word "Caucasian" will definitely give you some strange looks ;-)
> You're not wrong though, we should use White European instead since that's what it means today.
I agree, because it would make it clear what it is really about: lumping people together by their skin color. This is one of those examples where PC speech only helps to hide a problem instead of solving it.
There are a number of these things, though. I personally get a little irritated with the way many people here use terms like "Hispanic" to contrast with "white people". To me, "Hispanic" is a wide spectrum of ethnicities but really more of a linguistic, cultural and national/political specifier, with lots of overlap with "white". People who are trying to be correct about this will pull out complicated phrases like "non-hispanic whites" which may be correct but are also clumsy, reflecting the absurdity of the way we categorize this country's social boundaries in the first place. "White" itself has had a fluid definition in the US over time, previously having excluded southern or eastern Europe, Jews, or even the Irish.
I know that in my own recent ancestry I have at least 4 European ethnicities, some of which would have been ridiculed by other Americans in the past. But this nuance washes away and I am a "white guy" which lumps me together with cultures I don't particularly identify with. There is no sense in denying, for example, that I am less discriminated against by appearance than other groups. But I do not think either that there is one homogeneous "white people" living in the United States.
Therefore I find it pretty odd that in the USA the idea of a uniform European "ethnicity" - largely defined by skin color! - is still very common and even officially practiced. I remember how I had to fill a form when entering the USA which asked for my "ethnicity". I was flabbergasted and just ticked "unknown" or something like that because I didn't want to identify myself as "Caucasian".
Every form I can remember seeing in the US that asks whether you are Hispanic says that it includes people of any race.
So how do you like this election season?
Yes, that is the "correct" thing that goes on the form, written by people who have bothered to look into this question. It isn't how people talk in casual speech.
I personally do not. Nor do those forms you are talking about. But it something I hear a lot out in the world.
How very dare you. I'll have you know I come from a long line of Fuckers.
I learnt that Apple has a validation endpoint. Since the engraving service was recently updated, I took a profanity wordlist and checked it against the endpoint just for fun. The results are...counterintuitive: https://pastebin.com/mzpECiQw (NSFW language, obviously)
What happened was, the German localisation of the app description included the word “Knopf”. Knopf is not a rude word, according to any German I’ve discussed this with — it is one translation of “knob” in the sense of “button”, but Apple’s naughty word detector clearly thought it was “knob” in the sense of the euphemism for a body part.
It didn’t stop the app passing review, but the automatic warning was still a regular part of updates for that particular app.
[0] Back in 2012
1. People who use them for that just come up with new ones all the time
2. People still use the word in the regular sense, like would I now have to come up with a euphemisms to describe a door knob ?!
3. How I wore is my decision. As long as I don't hurry anyone intentionally or knowingly no person and even less company had the (moralic) right to constraint me. (Through wrt. Minirs, Y least younger ones, their parents opinion matters, too.)
[0] curl -X POST https://accounts.api.playstation.com/api/v1/accounts/onlineI... -H "Content-Type: application/json" -d '{"onlineId":"hitle123987","reserveIfAvailable":true}'
Anyway, I switched languages and I'm "zed0134" on their network. The whole thing struck me as a bit odd.
? Seems like a poor reason but most profanity filters are inconsistent as hell anyway so its possible.
I can explain that one to you.
curl -d '{"onlineId":"zeta0134","reserveIfAvailable":true}' \
-X POST https://accounts.api.playstation.com/api/v1/accounts/onlineIds \
-H "Content-Type: application/json"
edit: As other repliers have said, I did think that it could have something to do with Los Zetas, and maybe that was the case in the past. However, as of now, `zeta0134`, `zetas0134`, and `loszetas0134` are all valid and available names, according to the endpoint.
Interestingly, `pabloescobar0134` is considered improper. However, `elchapo0134` is not – which supports the argument that Sony's blacklist is (obviously) manually curated and perhaps subject to change. Maybe enough people who wanted "Zeta" complained for Sony to realize it was a word that had many more uses than in Los Zetas
I don't know Mitch Hedberg, but that's also my reasoning. I'm not superstitious, but if I were and I had a problem with number 13, I totally wouldn't want to sit on row "14". There is a supernatural power that makes 13 bad and you can fool it by printing a "14" on top of what's clearly the 13th row? The spirit, deity or force of nature that causes bad luck looks at the labels? Come on!
I imagine the way you're approaching this is not unrelated to the fact that you're not superstitious. That is to say, if you were superstitious, it's possible you wouldn't feel this way about the fake 14th row. Not having to see and think about the number 13 may be enough.
Maybe it's that.
I can't imagine Jobs putting up with trying to swear at someone over text and repeatedly getting "ducking."
Android is the superior experience in almost every UX category imo, and text handling is the best example of that.
sent from my iPhone.
you may have to do it a few times before it works.
On Android the latency of the screen has always bothered me when typing so much I can't notice anything else, but I haven't had a high-end Android device in years where that could be remedied (I was very tempted by the OnePlus 7 but didn't want to fork over the cash just to try).
By the way Semen is the way the Russian equivalent to Simon is spelled officially. There is quite a number of Russian men who have just that for their first name in their passports. I once met one, he worked as a developer and people were making fun of his name.
That only works for a very small subset of the world population. Even in the US, try sueing tictok.
Proof: https://www.transliteratciia.ru/transliteraciya-imeni-i-fami...
https://wireframe.raspberrypi.org/articles/banned-character-...
We see the same phenomena with encryption, authentication, and password security. People just tend to have a myopic understanding of the topic and vastly overestimate their ability in something they've barely even studied, which is the worst kind of ignorance. And they very much deserve to be named and shamed when they inevitably screw it up.
When dealing with names & addresses, consult an expert.
And phone numbers, and dates.
the scunthorpe problem will never die.
As a programmer, it's difficult how little of the world follows consistent "rules" that you need to build straightforward conditional logic.
Really, the only surprising thing is that Lyft didn't cross this bridge years ago.
I ended up taking my name off (the emails and website) of an open source project I maintain just to please the spam filters: https://twitter.com/j__st/status/1210559583934189568?s=19
I can appreciate the funny side of it, but it's also profoundly saddening somehow. I was born with this name, what do you expect me to do?
And who is bothered by this, really?
I find it hard to imagine that it's the drivers calling themselves Dick Cummings out of immaturity. So what does it really matter what the customers call themselves?
Edit: Actually, it's worse bc if you get it wrong you open up your employer to getting called out for cultural insensitivity.
You should never try to define 'appropriate language' and then impose your definition of that on to the world. Especially if you want to do business.
That was my thought too. It's one thing if you're using obscenity in an anonymous online alias, as a great many Redditors seem to love doing, but I'm struggling to believe that it's something people do in signing up for real-life ride-sharing apps, either as a driver or rider.
Obviously there must be some amount of people doing it, otherwise Lyft wouldn't bother with this at all, but it's just so... unexpected.
There's no data that this is true. It could simply be a project manager deciding that it's a good thing or the result of a tiny fraction of complaints.
https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-...
Discussed most recently just 63 days ago:
Edit: People seem to think I'm offended by those words. I'm not.
See also:
Then why defend people who are offended by words? That's their problem, not yours.
> some are apparently offended by others being offended.
O_o
Those while thing was hypocritical and goofy.
Obviously context and tone play a lot into it and both "cunt" and "mate" can be friendly or threatening/derogatory depending on delivery and circumstance, but I imagine it would be remarkably difficult for people who aren't immersed into Australian culture to figure out which is which based on text alone
I hope to one day be able to see the feature spec and QA that led to these types of hiccups.
Sample usage: "AS is sometimes an unintended result obtained when insanely smart people work with vast resources on very complex problems for long periods of time."
I expect I will be receiving a similar notification from Lyft soon as the email address I am registered with definitely has a name which would incorrectly trigger this algorithm.
When are we going to grow up and realize that the mere appearance of words like "dick" and "cum" don't hurt anybody?
I remember getting sent home for using the word "Penis". My parents both went to the school for a parent-teacher interview. That principal had brought the teacher in and were serious about my need for detention etc. The whole thing ended in farce as my father laughed in the principal's face then angrily demanded no consequence for me.
I doubt today's generation of teachers would tolerate this. Probably react to being humiliated like this by calling it harassment or similar nonsense. But I also think my father if he was part of this generation would have twittered the whole incident instead and let the twitterverse do it for him.
https://www.travelettes.net/funny-town-names-from-around-the...
In this case for 'trademark' reasons, but the problem is the same. Filter for a set of things, and eventually legitimate examples of said thing are going to get caught up in it.
Also, is it really "the algorithm" or does Lyft just maintain a list of entries of banned names? I know "the algorithm" is the hip new thing, but I have my suspicion it's the latter.
https://www.dw.com/en/german-grammar-in-ok-boomer-tweet-gets...
Also my all time favorite when Facebook banned people posting about faggots which is a traditional food in the UK Midlands
https://www.telegraph.co.uk/technology/facebook/10419598/Man...
Life seems to be peppered with problems with people with limited vocabularies who would rather make themselves "offended" by something, rather than think.
See also: https://en.wikipedia.org/wiki/Controversies_about_the_word_n...
It's not an algorithm, it's some overjealous dev/PM with not many important tasks on hand except for a little incident which he made big deal out of it by downloading all dirty words and do string contains.
</speculation>
Example: https://en.wikipedia.org/wiki/Marijuana_Pepsi_Vandyck
Not a good reason to block them though...
It's not clear to me why Dick is the standard nickname for Richard.
Compare manikin, a little man.
Apparently unrelated to French Jacques, despite the similarity.
https://metro.co.uk/2015/03/10/people-living-on-butthole-lan...
Dick Pick created an operation system named GIRLS !
https://www.worldatlas.com/articles/the-strict-name-laws-of-...
The underlying story is getting tragic, however: "you can't blacklist impropriety" is apparently not learnable, despite decades of failed attempts.
How many times must we go through "real names" idiocies before these supposed geniuses stop being imbeciles
Follow up: Also appears there are some misgiving regarding crude language within the Hindu religion as well. [0]
[0]: https://www.hinduismtoday.com/modules/smartsection/item.php?...
As far as I can tell, he was a member of an Abrahamic sect that has since died out. But yes, one that accepted Christ as a prophet.
Which?
"Shooter" is a word that's acceptable to use in almost any company, even though the verb at it's root part of some abhorrent acts.
"Fuck" is a distasteful word for an act with near-universal approval.
Near-universal approval and near-universal distaste. A curious phenomenon, isn't it?
"making love" is something partners do as a healthy and (re)productive part of their relationship, "fucking" is what 'animals' do, purely done for pleasure. It is a bad word for the same reason gluttony, excessive alcohol and drugs are a sin. A society where everyone mindlessly chases dopamine from carnal pleasures is less productive, especially when this leads to unrest and violence (and I guess that's where forced sex comes into play).
This reasoning is inconsistent with the rejection of "Cummings."
See eg "shirtgate" where an astronomer was castigated by feminists for wearing a shirt with skimpily clad figures, on the basis that it was offensive to women. Ironically, the shirt had been given to him by a feminist woman friend who thought it was liberating.
> Ok, let's call that puritan values of abrahamic religions.
Opposition to bad language is a feature of every society with more than one social class, which is all civilized societies and a lot of uncivilized ones.
Example: the United States.
Oh wait, that one goes the other way.
Although you are correct about nudity not being a big deal there, they actually have specific laws about the names that newborn children are allowed to be given.
How do you know it's not secular? Are you sure the people who made the decision at Lyft were religious, or did it for religious reasons?
I'm not Christian but I don't think most Christians consider a night out a club to be debauchery.
We detached this subthread from https://news.ycombinator.com/item?id=22022195 and marked it off-topic.
Is it because we're medieval scribes and we really want to use the Latin faeces to indicate our scholarly intentions, instead of the Anglo-Saxon shit that everyone around us uses?
There is no reason I can think of that comports with a technological or innovative worldview.
Including cybersecurity. ("Shit" is just an identifier or a guessable string with 13 bits of "chaos" completely equivalent to "Dave" or "Phil.")
Edit: Well I see small-minded superstitious dunderheads have shown up to downvote me, so FUCK CUNT SHIT ASS
In pretty much every community, we seek to reduce the friction between members, and as a part of that, different communities would often choose a point on that chance-to-offend spectrum and discourage use of anything more (potentially) offensive than that point. Despite myself generally preferring more open (and more offensive communities), I find it very rational for others to set the bar at other points.
Edit: sorry, digressed so far that I forgot to finish. By "priorities" I mean, Lyft, pull your people off of pointless projects like messing with people's identities, and put them on QA of basic functionality like accepting a credit card payment via your website. Last and only time I tried it, it failed utterly. Now I don't know how much anti-sales other people need to de-persuade them before they'll not-buy, but in my case that was enough. The person for whom I was trying to buy a gift card, got something nicer with a lifecycle that doesn't burn such an obscene/offensive quantity of fossil fuels.
When you see a comment that ought to have been moderated and hasn't been, the likeliest explanation is that we haven't seen it yet. We can't moderate what we don't see.
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
https://news.ycombinator.com/newsguidelines.html
Also, please don't go on tilt like this.