26

Grimes and Elon Musk named their baby: X Æ A-12.

What are the risks of non-ASCII names?

For example, does the COBOL unemployment platform support non-ASCII names? Would it be possible to get a social security number (SSN) for the baby? Does it make their baby an easier target to impersonation attacks using Unicode manipulations? Could the baby name trigger a code injection attack?

Clearly pure-ASCII names might cause issues: too: \0, ^Z, NULL, etc.

16
  • 91
    You can have problems with ASCII names too. My cousin has an ASCII name and always has a lot of trouble because it contains a null character and two line feeds.
    – reed
    May 8, 2020 at 9:49
  • 27
    @reed My legal name starts with 0x04 EoT and it just keeps screwing me over. I'm glad to see I'm not the only one, but I do feel sorry for your cousin.
    – user163495
    May 8, 2020 at 12:19
  • 18
    Even sticking with just the Latin alphabet poses problems. Imagine someone named Null. Those people already exist and experience regular headaches when dealing with computer systems. May 8, 2020 at 19:07
  • 13
    They did not named their baby X Æ A-12. They named him (or her, I do not know) Kyle. This was just some supposedly funny or clever way to announce the name.
    – WoJ
    May 8, 2020 at 20:23
  • 15
    A couple or years ago I encountered a payment platform that would not accept the name "Grant" (It was blocking other SQL keywords too, so "Bobby Tables" would also be denied)
    – Jasen
    May 9, 2020 at 12:10

6 Answers 6

33

There are even Chinese people in the US. They name their children 李某. Would that be a problem? No. Some systems support these names, some use versions converted to ASCII through romanization (李某 → pinyin Lǐ Mǒu → Li Mou). The only non-ASCII character in X Æ A-12, Æ, is used e.g. in Danish names like Ægidius, converted to ASCII Aegidius.

A good example of such conversion is machine-readable passports: the last two lines on every passport contains only characters A–Z, 0–9 and the filler character <. For passports, every name in the world gets converted to ASCII.

Sometimes using mere ASCII may be more problematic than non-ASCII characters that are easily encoded in UTF-8:

  • Christopher Null is a real person who has reported some problems with his ASCII name.

  • My name is Logger, Startkeylogger.

  • Little Bobby Tables has a name as unique as the little baby Musk. This Robert doesn't have a social security number as he's a fictional character, but for real people the name isn't an obstacle.

    Exploits of a Mom

6
  • 6
    The child may be named "李某", but for everyday purposes, an appropriate transliteration into the Latin alphabet (probably Pinyin) will be used.
    – Mark
    May 8, 2020 at 21:58
  • 7
    For everyday purposes, the child will be called whatever their parents use (be that the official pronounciation of 李某 or not). It is when writing its name on a computer system that it will then often be adapted to whatever is suitable for the system and available input entry.
    – Ángel
    May 9, 2020 at 3:28
  • 5
    The ascii compatible romanization (already mentioned in the answer) is pinyin without tone marks (Li Mou), as the pinyin itself doesn't really help with the original problem of having non-ascii characters (Lǐ Mǒu). Neither does International Phonetic Alphabet IPA (lʲǐː mə̌u). May 9, 2020 at 4:36
  • 1
    Good point, there is however a huge difference between being Chinese and giving your child a Chinese name (which romanizes just fine, too), and giving it a name which is strong indicator that your full contractual capability (or sanity, in the widest sense) is to be doubted. A name for which the child probably could sue you for mental cruelty, once grown up. That's even moreso the case if as the main shareholder of a company, you make public statements such as "that company is way overvalued".
    – Damon
    May 10, 2020 at 19:00
  • The question is about whether it's problematic or not, and this answer is to demonstrate how it's not that unique after all, regarding information systems. Whether it's sane or not is not about information security, and should not be discussed here. May 10, 2020 at 19:09
20

I don't live in the US but I live in a street that has (if written correctly) 2 German Umlauts in their name.

I never had a problem with my passport or with US authorities, visa waiver programs or anything that concerns the US government. They will accept it either the way it is or a "translation" into non-Umlauts (which even exists in Germany as a legal way to write the same word without Umlauts).

I have no experience with SSN or unemployment systems in the US, but I have no reason to believe they will behave differently from the civil administration I did come in contact with that handled it without blinking.

Now private businesses are a whole other can of worms. It has become a lot better in the last 3-5 years, but I had to fit a lot of square pegs through round holes to use my address in the US. For example, I own a credit card from a major US brand. And one would assume you can use that to pay online, right? My credit card is no different than your credit card. Well, yes and no. One will need to give the address to the merchant that in turn will send it with the card data to the card processor to reach a higher confidence that it was actually me doing the purchase, not someone who scammed just my card number. So the crappy merchant website would not accept umlauts. No problem, use the other accepted spelling. Then the merchant would accept it, but the card processor would barf and say "no sir, that's not their correct address, as written down here.". I think I dropped 25% of my purchases online because the merchants were literally to stupid to draw my money from a major US brand credit card.

Another example: bring an App into the Apple store. That was only 3 years ago. Apple is a major international company, surely they would not screw it up, right? Well, they did not. Directly. I could create an account. But they have a partner where you need to be registered to be accepted as a company. Guess who had never heard of Umlauts? I needed to talk to third level support (actually a developer that had database access) to get that done because even their own internal support interface would not let them.

So... I think you will be fine with all official, administrative tasks, but don't expect an easy life. Life is not just death and taxes, if you want to do something in between, maybe something that's fun or makes money, it will be a lot easier if your name is "Jake Brown" than "Jörg Oßten".

8
  • 6
    Using street names in authentication sounds horrible as even ascii street names could be written in several variations. There's also cultural differences other than the character sets, e.g. there's no first name and surname on every country. Authentication schemes should not rely on these kind of things, but unfortunately some do. May 8, 2020 at 7:51
  • @EsaJokinen It can be pretty bad. I had a problem once because Visa and my bank disagreed about the zip code of my debit card. But, what else can they do?
    – jpaugh
    May 8, 2020 at 16:11
  • 4
    If credit cards were to be invented today, everybody would laugh about the security. May 8, 2020 at 16:26
  • About the US part: they accepted all of ü, ue and u?
    – WoJ
    May 9, 2020 at 7:33
  • I know a phone company which refuses to allow apostrophes in names or addresses. In Ireland!
    – TRiG
    May 10, 2020 at 18:03
9

Posting this relevant link, because (to my surprise) no-one's mentioned it yet:

Falsehoods Programmers Believe About Names

I think it should be compulsory reading for anyone who works on systems handling personal data!  (Though I suspect it would be even more useful if it included examples of each case.)

As comments indicate, even plain printable ASCII can cause problems: far too many systems have trouble with names which include apostrophes (such as O'Connor and D'Artagnan), hyphens (Day-Lewis, Zeta-Jones), or embedded spaces (Lloyd Webber, Bonham Carter, de Vries), or embedded capitals (McDonald, FitzGerald).  So it shouldn't be a surprise that non-ASCII chars are even less widely supported… let alone mononyms, digits, extremely long names, names which include terms which are offensive in some language, all caps, all lower case, or change!

1
  • 1
    I'm not sure how this answers the question asked. This answers talks about what systems should do, not the current impact.
    – schroeder
    May 9, 2020 at 13:41
7

There are two answers to this question. One involves talking about "Is it possible?" the other involves talking about "What are the costs & concerns?".

The first will greatly enlighten the second.

The answer to the first question is somewhere from maybe to yes, but with workarounds.

The answer to the second depends on what you are trying to do and who you are. For someone like Musk, being a White Male Billionaire will smooth over many of the problems and costs.For many others, you may discover that having an ASCII name makes things much simpler.

How would you spell Æ over the phone? What about the IRS Website that only allowed for UPPERCASE letters in street addresses, if a site doesn't even do case conversion, what the hope that they will work with Unicode? Many old COBOL systems on IBM Mainframe platforms use EBCDIC which stands for the Extended Binary Coded Decimal Interchange Code. A-Z, a-z, and 0-9 exist in EBCDIC (but {} and [] don't).

I'd say that unless you both very rich and a bit weird, don't burden your children with strange names. School is tough enough without that extra burden.

5

The Musk child was born in California. California does not allow non-ascii characters in names and the child's name as submitted may be rejected, according to this link.

Although it is not completely clear, the "Æ" may be the issue, as California requires the birth information to be entered into an electronic record. California adheres to the United States Standard Certificate of Live Birth (Section 102425 (f) (1) ) in that link.

The standard certificate is here.

These may be the rules for the standard certificate, although the whole matter seems to be far more elaborate than one would suspect, with the mother's cigarette smoking history included, etc. etc.

2
  • What do cigarettes have to do with anything?
    – TRiG
    May 10, 2020 at 18:07
  • Exactly. In all the government regulations about what can or must be included on a birth certificate, I was unable to find anything that actually answered the question. So, the news article was the best I could come up with.
    – Wastrel
    May 11, 2020 at 13:59
5

Speaking as someone who has a last name that, properly formed, would be "Ælwyn", and who happens to have a day job for a software company that writes, among other things, stuff for tracking "Land and Vital Records" ("Vital" being birth and death certificates, along with name changes)… in most places the kid would have problems, although there is pending and/or recent legislation (I can't recall which, offhand) that would require systems to cope with such names more effectively.

But in practice even then a lot of systems will transliterate it as a simple "AE" because most things treat it as a ligature rather than a distinct character, and decomposing a ligature is a typographical (presentation) change rather than a change of content.

But I wouldn't expect them to be appreciably more (or less) easily targeted by Unicode manipulations, really. Or at least if they are… it is going to be so far behind all the other headaches from systems that can't cope with them that they're not going to notice.

Edit:

Yes, having three non-familiar names will also cause at least some systems to struggle with it, though probably far fewer — several systems at least allow for multiple "middle" names. Before changing mine (yes, the Æ is self-inflicted, and yes, I knew it wouldn't translate well, which is why legally it is "Ae") I had two middle names, or a single middle name and a hyphenated last name, depending on which way you interpreted it, and several times it did cause "interesting" situations — but hyphenates are at least fairly well supported almost everywhere in the US now.

Edit 2:

To address the entirely valid point raised about Æ not strictly being a ligature: partially (or even mostly) correct. The character "ash" is not a ligature at all, and is in fact the proper form for words deriving from languages which used it (including my own last name). As for lexical vs. typographical ligature, all I can say is that even the Unicode standard can't seen to decide if it should be or not (see https://stackoverflow.com/questions/9376621/folding-normalizing-ligatures-e-g-%C3%86-to-ae-using-corefoundation for some discussion of this), but my actual point stands: in practice, a large amount of record-keeping software, if it can deal with it at all, will be prone to decomposing it (especially if anything like OCR software comes into play). To make it even more fun, unlike ẞ or the "Turkish I", there is a complete round-trip mapping between Æ←→æ.

3
  • I think there will be a problem with the spaces though? Won't the three elements probably be recognized as three first names? And out of curiosity, do you normally type your name as Aelwyn, or the other way? (sorry, on mobile so I don't have the right glyph)
    – WoJ
    May 9, 2020 at 7:35
  • 6
    My friend, Mr. Hämäläinen, was once converted into ASCII as Mr. Homoloinen (Finnish for "gay parasite") on a hotel management system. Having a name with umlauts can sometimes be embarrassing! May 9, 2020 at 7:58
  • Æ may be a ligature for Ae, or may be a single character in its own right, depending on the language. In Old English, Æ was thought of as a letter, not a ligature. Even in languages which do consider it a ligature, it's a lexical ligature, not a typographic one, so it really shouldn't be decomposed.
    – TRiG
    May 10, 2020 at 18:07

Not the answer you're looking for? Browse other questions tagged .