Hacker News Reader

Comment by a_shovel | Hacker News Reader

I think the lesson to take from these lists is that covering every edge case for many real-world situations becomes asymptotically difficult.

You can get 95% with no effort at all, 99% with a bit of thought, 99.9% with some effort, 99.99% with a smart design, some research, and a lot of effort, and anything further requires making it your full-time job indefinitely.

It gets thornier when it comes to personal things like names and gender. If you tell a group of people that their identity isn't supported by your system, some people will accept that and work around it, some people will be mad at you, and you'll just have to accept that the latter have a good point. You have to stop developing it at some point, so all you can do is try to cover as many people as possible.

That said, it is incredibly revealing of your personal biases if you design a system that breaks on things like names with "é", or "ñ", or punctuation; or if you implement gender as a boolean that also determines the user's pronouns.

View on HN

In regards to your last line, I feel like we do need to give people some leeway and not assume bigoted intent. Most of the things we can gig people for not implementing nowadays are only recently important in our cultural consciousness. English has traditionally shied away from using accents and one can reasonably assume that if a person signs up for my English language website, they won't be using Chinese characters in their name. As for the boolean determining pronouns thing, I can easily see someone thinking that was a clever implementation before people adopted the idea that there are more than two to be cognizant of. I also think using they/them is only recently considered grammatically acceptable. Going forward, we do want to support the things you mentioned.

Definitely old software that's obviously not getting updated can be forgiven. There's still 16-bit DOS software used in production that harkens from the time when ASCII was all that could be supported. I get it, it sucks when a legacy airline system written in the '70 removes the accent characters from your name but them's the breaks. But if you're designing new software nowadays and your requirements don't include the "new cultural consciousness," you're probably making a big mistake. I've seen software built within the last 5 years fail QA due to not displaying emoji.

Yeah, it's probably wrong to assume much about a lack of support for Chinese-character names on an English-language website, so I removed that. (But if you don't have a good reason not to, you should support it.)

Accents and punctuation are different. There are a lot of English-speaking people whose names don't fit in ASCII. A system that doesn't support them is at the "95% for no effort" stage, if even that.

Bias isn't the same as being bigoted. If when you find out you missed it you go "Oh, crap, you're right! We should fix that" you had a bias. If when you find out you missed it you start to defend why it's fine to not include it (not to admit you should, but maybe you don't have time/it isn't worth the cost), that's the point it's bigoted.

It's the difference between making a mistake, and getting defensive that it wasn't a mistake. Per old systems, they all fall under bias; it's only with how do the existing support personnel deal with it that you start looking at bigotry, a "I'm sorry, I know, it's just an old system and we don't have the money to update it" is totally understandable and I would argue okay. Digging in that it's not the system that is at fault but the individual wanting a different pronoun, or the person refusing to pick an anglicized name and start using it everywhere, is bigoted.

> If when you find out you missed it you start to defend why it's fine to not include it (not to admit you should, but maybe you don't have time/it isn't worth the cost), that's the point it's bigoted. > > It's the difference between making a mistake, and getting defensive that it wasn't a mistake.

I think that definition is pretty good, but it also reveals something about the popular usage of the term "bigot".

IIUC, an underlying assumption in the parent comment is a person is at fault for not accommodating the offended party's preferred taxonomy / categories / terminology. That a person is a "bigot" if they see their system's current taxonomy as better or more appropriate.

Merriam-Webster's definition is somewhat different [0]: "a person who is obstinately or intolerantly devoted to his or her own opinions and prejudices".

Note that with MW's definition, either or both persons in that "you missed it" conversation could be labelled a bigot, depending on how obstinately they held onto their views.

[0] https://www.merriam-webster.com/dictionary/bigot

You are skipping over the last two words in the definition. It isn't just holding to their own opinions. It is holding to their "opinions and prejudices". Of the two stances "I want to be referred to by the pronouns of my choosing" and "I want to refer to you be the pronouns I dictate for you" only one of those would qualify as prejudice and therfore that is the stance that qualifies as bigoted.

I wonder if we're assuming different definitions of "prejudices".

What I mean is having reached a conclusion without consideration of all the relevant details. I.e., deciding a matter prematurely. I think this lines up pretty well with the MW definition [0].

I don't see how that's related to the particular opposing stances in your example. Can you elaborate?

[0] https://www.merriam-webster.com/dictionary/prejudice

The person who refuses to use another person's preferred pronouns is obvious dictionary definition prejudice. It is someone prejudging the proper pronouns without the relevant details from the individual they are referencing. It is almost literally the person saying "I am deciding your gender without consulting you."

What are the relevant details when choosing your own preferred pronouns? Wouldn't all the relevant details be known by the person making the decision? I don't see how that can be prejudice.

Thanks for clarifying! I can see now how that could be an example of prejudice.

I had a different scenario in mind: person A indicates their preferred pronouns to person B, but person B refuses to use those pronouns for well-considered reasons and with a full understanding of the situation. It's this scenario that I was arguing to not be a case of bigotry nor prejudice.

Oh, I thought lostcolony had already eliminated that specific scenario as not being bigoted. It can be biased, but it wouldn't be bigoted if there are justifiable reasons for not making the switch. That isn't prejudice as it is a decision made after the relevant details are known.

Just to be clear, for the parent, 'person B refuses to use those pronouns for well-considered reasons and with a full understanding of the situation' may apply when we're talking about systems, but there are plenty of people who would use that sort of reasoning to be why they refuse to use someone's pronouns in conversation. No one doing so believes their reasons are poorly considered and/or that they have an incomplete understanding of the situation, after all.

So, yes, "justifiable", as in, "to change that in the system would be a lot of cost and risk and we simply can't justify it, but I agree it sucks that it's that way", not as in "You were assigned male at birth and I refuse to use anything other than he/him when referring to you, because I believe I understand the science, and your personal situation, better than you".

Not sure if "for the parent" was directed at me or the parent of my comment, but you and I are in complete agreement.

> I also think using they/them is only recently considered grammatically acceptable.

In general, as a singular gender neutral third person pronoun for a person of unknown, unclear, or unspecified gender, they/them is approximately as old as modern english itself, having been clearly used by one of the earliest practitioners of the language, Shakespeare.

What's modern is the idea that it's somehow incorrect to have a pronoun that is both plural and singular, which is funny because 'you' is both singular and plural and people use it every day without incident or confusion.

Use for a person of known but not-binary gender is newer than that, but probably older than you think, and neopronouns are actually probably also quite a bit older than you think too.

For what it's worth, I'm glad they/them is catching on and I use it all the time.

Growing up, though, I was taught to use "he or she" / "she or he" in English class and they/them was always used in the context of third person plural. I don't consider it a newfangled thing, but the embrace of it feels new.

"They" comes off impersonal though. As if you're using it because you don't know their gender. That's why I would much prefer a new word. Something that indicates that you know they are neither he nor she. (In a similar vein I'd also prefer a distinct singular "they" and plural "you")

I do know that realistically something like that would never catch on, but I can dream...

It certainly could; look how quickly the term "Ms." as an alternative to "Miss" or "Mrs." caught on. Though it has antecedents long before the 20th century, the term caught fire in the early 1970s with the first publication of "Ms." magazine and was in widespread usage within ten years.

I think the bigger problem is coming up with an accepted alternative; there are several floating around, none of which have had a Ms. magazine moment.

I hadn't thought before of how recent Ms. is. You've given me some hope now. Not holding my breath but keeping my fingers crossed at the very least.

> That said, it is incredibly revealing of your personal biases...if you implement gender as a boolean that also determines the user's pronouns.

It is probably revealing of your personal biases that you cannot imagine systems in languages other than English that need to do some of this since multiple parts of the language, including verbs, are necessarily gendered, with a neutral gender usually only reserved for inanimate things. Reference: I happen to speak one such language.

Problems like this are just hard due to generations of built-up cultural context and the associated effects. We can and should do better, but there is no need to get overly enthusiastic about accusing people of biases while we are at it.

There's a world of difference between "Which of these two options defines your gender identity and sex organs?" and "Which grammatical gender form should we use for words when we talk about you?". The former is a deeply personal question with unclear benefit; the latter is a specific question about how you want the site to do a specific thing it needs to do. (Does it really need to do that? Why exactly are you asking about gender?)

I have an anecdote that I think fits your assumption. It comes from my years in the Danish public sector where some of the systems are many decades old. In Denmark we’re all issued a social security number that is designed in such a way that the last number of it, is odd for males and even for females.

To make things short and somewhat anonymous, we had several systems that couldn’t handle people changing the “gender” of their social security number. They could handle you getting a new social security number, but it had to be of the same “gender” as your previous.

This makes no real sense, as it’s just a number and all these systems handle both types of numbers. But there we were, and it took us several years to get to a point where we could accommodate an employee changing the “gender” of their social security number.

So sometimes, developers do weird things because of their assumptions.

Ps. I’ve put “gender” in “”s because I’m unsure what the right word is in English.

It could help an e-commerce site predict what kind of clothes to recommend, or otherwise segment you for marketing

Everyone has biases, that's just a fact. Something can only be unbiased in comparison to something else that has the same biases. Having biases doesn't make you evil.

Yeah, I only considered English language sites, not other languages that don't support gender-neutrality as easily. I think I should add something about them, though. I don't see how any language could possibly require gender to be implemented as a boolean. If you've got an English-language site you're translating to another language, and you've implemented a selector for "he", "she", and "they" pronouns without direct reference to gender in the English version, you could just re-use the same implementation in your language with different options.

(Also, I can guarantee that there's someone who speaks your language that prefers to be referred to in the neutral/inanimate form. Doesn't matter which language. It's fine not to support that on the site, just wanted to point it out.)

> It gets thornier when it comes to personal things like names and gender. If you tell a group of people that their identity isn't supported by your system, some people will accept that and work around it, some people will be mad at you, and you'll just have to accept that the latter have a good point. You have to stop developing it at some point, so all you can do is try to cover as many people as possible.

That’s a pretty big if, which itself reveals a bias. While it’s certainly not going to cover every single identity, the following will likely cover vastly more than added effort:

1. Do you need to ask for the data at all?

2. If 1, do you need to require the data?

3. If 1, unless you have to comply with arbitrary external requirements (eg law), use a free form text field.

4. Test that the text field and accompanying information is accessible to assistive tools. This usually comes for free if you use built in tools.

5. If not 2, affirmatively indicate that entering anything is optional.

6. Ask for feedback on how you can do it better, again making this obviously optional. If the feedback is meaningful (i.e. “I cannot use this”, rather than “I would prefer this be less inclusive”), accommodate it.

> ...it is incredibly revealing of your personal biases.

People make mistakes. I think you're opening a can of worms if you try to conflate a lack of robustness in user experience accessibility with some underlying anti-inclusivity bias.

> It gets thornier when it comes to personal things like names and gender.

In the case of name, I read an article sometime back that mentioned the problem of getting people's fullname right. Not everyone has a "first name" and a "last name" distinction. So it is better to just have one field of "fullname" and let people enter it as per their locale and tradition instead of having "first name" and "last name" fields. I think gender also falls in the same category.

This is the first "falsehoods programmers believe about X" article I read and what I think kicked off the trend. In fact this GitHub page references the same as "the article that started it all".

https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-...

> ...it is incredibly revealing of your personal biases...

I think that is an unfair accusation.

I've been coding for over 20 years, and I have only had to enable UTF8 to cover all corners of use cases (web based data/content) to special characters. I work with companies in multiple foreign countries, and my biases are irrelevant to technically working systems.

The bias in this is whatever belief system one might have to say "ASCII is good enough". Or whatever level of ignorance to believe that ASCII is good enough without understanding that it cannot be. Etc.

But what if ASCII is completely sufficient for all the customers I work with?

This is where the accusation crosses into unreasonable for me.

> The bias in this is whatever belief system one might have to say "ASCII is good enough".

Could there also be bias in the statement "ASCII is not good enough."?

I wouldn't call it bias. You can make a claim "I know my use-cases so well that I can guarantee that ASCII will cover it". I'd call that foolishness, not bias.

Sounds like you have never tried sorting or searching.

Where in the alphabet does "ö" or "aa" go? Well, it depends on the language AND the word.

Is ø the same as ø? Unicode can encode it in 2 different ways, so you need to normalize all encoding before comparing.

Is ø the same as o when searching? For Americans it is, for others not.

But how is it a personal bias if I never run into these technical issues to deal with in the first place?

You have never sorted any texts? Never made a list of names?

Never implemented a search function?

These aspects can easily creep into an app without you even noticing. Saying "I haven't seem them" is not proof for them not being there.

E.g. web servers had some surprise security bugs because they didn't think about overly long utf-8 codes.

The default Android keyboard (Google keyboard) had/has an issue like this. It really likes to auto-complete a to å and ä even though the keys are at opposite ends of the (Swedish) keyboard. If I wanted to write "här" (here) instead of "har" (have/has) I would've used ä...

I have done all of those things. But my customers simply were not affected negatively by the technical choices I made. (and I have great relationships and very open communication with the people I've served)

I implemented single fields for personal names, single textareas for addresses and some very basic content format selections and collation and my simple solutions have solved _all_ of my customers needs in these areas.

It seems to be some bizarre assumption that _everyone_ has a bad experience because some code wasn't done they way they would do it. It's absurd to be accused of something that has simply not been a problem for my users.

Are you sure most systems even need to record gender or pronouns? Outside of medicine sex is rarely significant, and if by gender you mean masculine or feminine traits then again (other to the person themselves) it’s largely irrelevant I’d have thought.

Pronouns can be useful in any system that generates text that refers to people in the third person, so makes sense to ask -- but it also makes sense to just let people put in whatever they want unless you are trying to aggregate data from it for marketing purposes, in which case imo you should stop doing that.

> Outside of medicine sex is rarely significant

Frankly, even inside of medicine it's usually not terribly relevant, and is often a poor proxy for the information they really want out of it. Half the time a medical form asks you that it's literally just so that cis men won't get mad at you for asking if they're pregnant on a later step.

I won't be at all surprised if I get some kind of kneejerk reaction to this statement suggesting that it might be important in an emergency or something, but for the most part that's not really the case, and even where it is, if it's that important (ie. allergy level), it's not sufficient to avoid harm because you will have situations where neither "sex" or "gender" as commonly formulated on these forms will tell you what you really need to know (hormone balance, organ configuration, medical history). Any one of those could be out of alignment with whichever social construct of sexed traits you pick.

> Frankly, even inside of medicine it's usually not terribly relevant, and is often a poor proxy for the information they really want out of it.

Sex is incredibly important inside medicine. It's probably the most important distinguishing characteristic when determining how to diagnose and treat someone (maybe weight is more).

Here are a few examples:

- 99% of breast cancer occurs in women

- any diseases affecting the female[male] reproductive systems can only impact women[men]

- 4x as many women suffer from osteoporosis

- Roughly 70% of those living with autoimmune diseases are female

- stomach cancer (2:1 men) oesophageal cancer (3:1 men) liver cancer (2:1 to 4:1 men)

https://en.wikipedia.org/wiki/Sex_differences_in_medicine

Yep, called the kneejerk.

These are statistical differences. If you want to know if they apply to an individual you need to know more. A checkbox on a form is not sufficient.

If you're usefully aggregating data for some purpose, say for a study, then go off ask all the relevant indicators you want.

If it's an intake form for an eye doctor appointment, it won't tell you dick all. If a patient came in for a breast exam, you already know they're at risk of cancer. If you're their family doctor figuring out risk factors, you already know their medical history.

And if they're trans or intersex, and you're assuming that either their assigned sex or their self-declared gender tells you which bucket they fall in, you know a lot more than anyone else about how those traits affect those things because they are woefully understudied, and almost certainly complicated by literally everything in their medical history.

> If it's an intake form for an eye doctor appointment, it won't tell you dick all.

Where are you getting your information?

Results: Sex and sex hormones influence the lacrimal system, eyelids and blinking, corneal anatomy and disease, aqueous humor dynamics and glaucoma, crystalline lens and cataract, uveitis and retinal disease, ocular circulation, and optic nerve anatomy and disease. Systemic conditions, particularly autoimmune disease, and conditions that are unique to women, such as pregnancy and menopause, further illustrate the effects of sex hormones on the eye. Gender-based differences in ocular conditions and disease should be considered within the context of the underlying physical and social environment.

[0] https://pubmed.ncbi.nlm.nih.gov/19811761/

Skimmed the paper on Scihub. It looks like it's mostly a review of other studies that either show statistical correlations or are about hormone levels. I.e. it directly supports their point that sex is just being used as a proxy for the thing that you really want to know. (And most of it seems irrelevant for someone with healthy eyes getting a new pair of glasses.)

Proxies for the thing you really want to know are pretty useful when you can check the proxy cheaply and the thing you really want to know is (comparatively) expensive to test.

Like, I don't think the world would be better if we routinely gave cismen pregnancy tests, would it?

Again, we're talking about forms here. If you have a reason to want to know if someone has a uterus or might be or become pregnant, you can just ask that. Forms for things where this is important usually do anyways. Sometimes gated on an "if you answered female to the above..." which actually compounds the problem for some people who are not neatly categorized that way.

Likewise, asking a cis man if he is or might be pregnant should not be a horrible thing to do, nor a problem for record keeping, though like I alluded to in my initial reply here, it does sometimes cause confusion or even confrontation to ask them.

Those statistical differences are used to more quickly reduce the search space of symptoms when the cause is unknown. So yeah, they are still important, even for individuals.

In the marketing world, it's vastly predictive of behavior, similar to income level or political party. Men and women tend to have highly significant differences in shopping and purchasing behavior.

So, whether they 'need' gender or not, many places 'want' gender.

Looking back at this, yeah, I agree with the criticism that the last line is overly harsh. I try to be lenient to people with unconscious bias, and this doesn't help with that.

Reply on news.ycombinator.com