It’s not my position that not allowing any random line noise is automatically discriminatory- but see my example above that restricting allowable characters to Unicode is discriminatory, or at least can be.
Or maybe we could accept that one’s legal name should not be a core part of one’s personal identity, any more than a person’s social security number or any other random identifier. Just restrict it to the unaccented uppercase Latin alphabet and let people pick some mapping that can be remembered. Nothing stops people from using their preferred name in practice.
Heck, maybe we shouldn’t have legal names at all, just an invariant ID number.
There are a finite number of living languages with written traditions; they use a finite number of characters. It should not be difficult to come up with something that allows them all (accepting all Unicode characters would certainly do it). That someone in the future might come up with an ad hoc name is a separate argument, and kind of reductio ad absurdem.
This is not a difficult problem to solve with technology: unicode and Windows etc. has done that already. The problem to solve is that the powers that be are unwilling to recognize the problem.
What is the problem? A person’s legal name may be different from their preferred name, but that is not an actual problem, since the only purpose for a legal name is for the government to be able to identify you. There’s no value in supporting a person’s personal identity via their name and a lot of reasons to disregard it.
Arguably, the major problem to be solved–as evidenced by the problems in this thread–is that there should be a clear, simple, reproducible way to reference a person’s legal name. It’s not just computers; humans are not going to get that right either, whether due to missing accents or characters that they’ve never seen or otherwise. There’s just as much of a problem on paper as on a computer. After all, computers follow the rules that the programmers set up. The programmers are probably better informed about these matters than the general public, so if they get things wrong, the public will too.
It’s much better to simplify as much as possible to eliminate ambiguity and confusion. Delete everything but A-Z.
You’re right. There’s no problem at all in a system that treats the names of people of English ancestry different from people of other ancestries. Heck, why don’t we just make sure the whole world speaks English and gets their nourishment from Starbucks and McDonald’s? Diversity creates ambiguity and confusion. Let’s just stamp it out.
What does any of that have to do with how someone’s legal name is stored in a database? How people actually identify themselves in person is a totally different thing. Prince or anyone else is welcome to go by weird symbols or whatever else they might come up with. But it needs to be possible for someone to type a thing into a computer, or write down on paper unambiguously and with minimum error.
That is one of the reasons the Unicode standard keeps receiving updates (and has so many idiosyncrasies): to make that more possible.
Once, in the early 1990s, I did someone a favor and helped him produce a document on the computer. You should have seen his jaw drop when I produced an “É”; he had not conceived that was even within the realm of possibility.
There are dozens, probably hundreds, of legacy software systems that have significantly worse problems than multi-byte character encoding. The problem is fixing/replacing all these systems while keeping all the systems running, and doing all the other new development that needs to happen, and with a budget that barely keeps the hardware up and running. The problems are getting fixed as those legacy systems are made end of life; new systems generally work well.
Yes, TPTB should prioritize updating old systems. It’s a perennial problem, but it’s hard to allocate funds to fix things that generally work even though it inconveniences many. With an unlimited budget and unlimited number of engineers you could fix all the problems. Unfortunately, that’s not the world we live in.
Unicode is wholly incompatible with human interaction (even at a programming level). It’s utterly impossible for a human to unambiguously translate a generic piece of printed text to Unicode. It’s non-trivial even in simple cases, and the cases won’t always be simple.
For human compatibility, you need to reduce the character set, even below what ASCII supports.
I don’t think anyone ever claimed it was. There are various normalization and equivalence tables that are supposed to help with application programming.
It does what it does, which is supposed to be to subsume all those “legacy” character encodings in a centralized way.
I already mentioned I got lists of students composed from A–Z , no lower case either 
(But it was not my problem to interact with electronic records in any way, shape, or form. If the lists had been handwritten it would have made no difference to me, I just need to know who is enrolled in the class.)
Yes. And that’s a good thing if you want a common format for storing books or displaying websites or the like. But a very bad thing if you want identifiers that can reliably cross system boundaries, including from digital to paper or even spoken over the phone.
Perfect. It’s impossible to know in advance if the lowercase is relevant or not. Perhaps it is, in the case of McLastname. But some people have weird printing habits and mix case for no reason. Just standardize on uppercase. They can still type in their preferred case. For places that display names publicly (social media, etc.) most have separate screen names. That can be arbitrary Unicode or something even more all-encompassing.
I actually have no idea (re. my story). It was never my job to log into the mainframe and do anything (my part in all this would have been the same 200 years ago with handwritten lists). All I can tell you is that there was never any ambiguity who was who, and every record has a unique ID number anyway, the printed name being merely for human consumption (I suppose)
I mean, yeah, I kinda agree there. If you want to argue for every human having a unique identification number or something and that being our legal identifier, I basically agree, that would solve the issue you’re talking about.
And alongside that, we should have a field to insert a name, which can take any characters, so that people can write their name to be referred to in whatever way they want, with whatever characters they choose.
That’s how a computer scientist would solve the name problem. 
Many of us computer scientists have read Falsehoods Programmers Believe About Names, which I see DPRK has already posted. It’s enough to drive a person mad, and it’s obviously impossible for a system of any kind to survive all of those broken assumptions. The character encoding is just a small part of it!
But there is a solution. Well, two. One is the ID. The other is to just force everyone to come up with some canonical name expressible in the Latin alphabet (and less than 64 characters, etc.). It does not have to be their “real” name, whatever that might be. It’s probably best if it bears some resemblance if that’s possible, but if not then do whatever, as long as it’s not offensive. The only purpose is so you can be identified for government services. It is not required that it has personal meaning to you.
Regarding the falsehoods programmers assume: I must admit I certainly assumed the first four points as well as the last one - variations of “People have one full name”. That blog author glibly states that he could provide examples of people which don’t, but then doesn’t.
I’ve tried searching, asking the AI masterminds and even thinking about it myself. Yet I still cannot come up with someone or some culture with more than one full name - or less than one.
Can anyone think of any cases, excluding stage or pen names?
Sorta depends on how much anglicization you consider different. Garry Tan of Y Combinator has a Chinese name of 陳嘉興, which romanizes to Tân Ka-heng. So, the Garry is all new, while the Tan somewhat resembles the original. I think this happens a lot, where immigrants take on a fully different Western first name and then a last name that bears some passing resemblance to the original but is still pretty distorted. And the orthography is obviously totally different as well.
My sister alternately goes by her maiden name or not, depending on the circumstances. If she’d changed her first name as well she might well have two full names that she goes by.
I changed both my first name and my last name when i married, making my middle name (which is the name people called me) my first name, and adding my husband’s name as a fourth (and last) name. So i went from
Name1 Name2 Name3
To
Name2 Name1 Name3 Name4
And that’s what’s on my passport.
Professionally, i go by Name2 Name3 Name4
But the state didn’t do middle names on the driver’s license, so my driver’s license, which is the ID i use most often, was
Name2 I3 Name4
And just to confuse things, the bank messed up when they created my account, and have me listed as Name2 Name3_name4. That is, they gave me a compound last name that starts with I3 and has a space in it. They can never find me because I’m alphabetized wrong. But let’s ignore that, as it was actually an error. Still, complicates real ID …
I also have a full Hebrew name that i use for signing religious documents, like my marriage contract. It is
Name5 daughter of Name-father and Name-mother
What’s my unique full name?
update
I was able to get real ID this year, and the clerk apologized, saying she would have to change the name on my driver’s license to match my passport. I’m delighted. I will no longer have to decide when i buy an airline ticket what id i plan to show when i travel. So i kinda have a unique full name now. But it doesn’t match my birth certificate.
Plenty of people have more than one full name that they use- in addition to stage/pen names , you have people who use a “ized” version of their first or personal name - " Marco" becomes “Marc” or vice versa, people who have two different unrelated names ( my kids have Chinese names in addition to the ones on their birth certificates) , people who use a version of their name in a different writing system (I’m not sure exactly what my name would become in China, but I’m 100% sure it won’t be in Latin letters), people who sometimes but not always use their middle name or initial, people who use different last names depending on the situation ( one for work, another socially and a third at the kids’ school).
The thing is, all that business about programmers believing people have “one full name that they go by” doesn’t really matter . I can go by ten different variations of my name and it doesn’t matter as long as I maintain a certain level of consistency between entities that interact with each other. Most of the time it doesn’t matter if the name I use differs from the name on my ID because one has a middle name/initial and the other doesn’t. It might matter if my kid’s school/doctor has something that matches the name on my ID in their official records but that isn’t necessarily the completely different name they use to address me . Where I might run into trouble is when I want to use different versions of my name in connected situations - I will have trouble if I buy plane tickets in Name 1 but my ID is in Name 2 , or if my paycheck comes in Name 1 but my bank account is in Name 2. Or if I want my job at a huge entity to use one name for my email address, logins, directory listings , business cards, etc. but a different one for payroll and tax records. ( That sometimes works at a small organization).
Programmers don’t necessarily assume that people have “one full name they go by” - they might believe people should choose a single “taxpayer” * or “government” name if they want to avoid complications but that’s a slightly different belief.
*Something I heard when a tour guide in Alaska introduced himself by his “taxpayer name”.
Getting slightly off topic, but you’d think it wouldn’t be that hard to develop systems that recognized that some people go by their middle names.
I’m a lefthanded person, and being a lefty in a righthanded world is no big deal compared to the hassles of being a middle name person in a first name world.