## Exact match

JW wrote to ask about the effects of Georgia's contested "Exact Match" law on people with non-ascii characters in their name:

How does this work out for Hispanic and other Latin alphabet diacritics? My Brazilian wife's full name includes the string "Lucía Mendonça" (í,ç). Many web forms, even in Spain, do not accept the diacritics. So her name will be spelled differently in different databases, from software flaws not errors in data entry. This affects not just Hispanics but naturalized Haitians, Poles, French Canadians, Swedes, etc.

And how does this work out for transliterations of names originally spelled in non-Latin alphabets (Vietnamese, Chinese, Korean, Hebrew, Arabic, Russian, etc)?

The answer depends on unknown (to me) details of the character sets and character coding used in the various databases involved; on the skills and habits of various people involved in various data entry processes; and on the details of the matching procedures used.

2.  The protocol codified by HB 268, and implemented by Georgia’s Secretary of State, Defendant Brian Kemp, requires county registrars to enter information from a voter registration form into Georgia’s statewide voter registration system known as “Enet.” That information is then matched against records on file with the Georgia Department of Drivers Services (DDS) or Social Security Administration (SSA). If the information entered into “Enet” does not exactly match the applicant’s identity data on file with DDS or SSA, the application is placed in “pending” status. HB 268 places the burden upon the applicant to then cure the no match result within 26 months. If this deadline is not met, or the application is cancelled, and the applicant must start the voter registration application process anew.

3.  Under this “exact match” protocol, the transposition of a single letter or number, deletion or addition of a hyphen or apostrophe, the accidental entry of an extra character or space, and the use of a familiar name like “Tom” instead of “Thomas” will cause a no match result. HB 268 imposes no requirement upon county registrars to check whether the information from the registration form was accurately entered into the “Enet” system or to perform any other quality review to determine whether the no match result was caused by a common error before relegating the application to “pending” status and putting the burden on the applicant to cure the no match—even when the no match result was caused through no fault of the applicant.

So the first question is what character coding scheme ENET uses, and what data-entry skills particular Georgia county registrars have. That is, does ENET allow characters like í and ç to be stored? And if so, do county registrars know how to enter them?

Then there are similar questions about the DDS and SSA databases.

And finally there's the question of how the matching algorithms work.

Apparently the quick answer to that last question is "badly". From a June 2009 SSA Inspector General's audit report on "Accuracy of the Help America Vote Verification Program Responses":

Our review found the HAVV program did not always provide States with accurate verification responses for individuals who were registering to vote. We determined the HAVV program had a significantly higher
no-match response rate when compared to other verification programs used by States and employers. HAVV’s no-match response rate was 31 percent, while the no-match response rate for other verification programs used by States and employers ranged from 6 to 15 percent. Additionally, we determined the HAVV program did not provide consistent verification responses to the States when the same applicant data were entered into the program. […]

We believe the high no-match response rate and the inconsistent verification responses can be attributed to the lack of (1) a unique identifier (full SSN), (2) flexible matching criteria, and (3) testing to assess the accuracy of the verification responses. Because of the limitations of the matching criteria established by the legislation, the HAVV program may indicate a no-match when a match does in fact exist in SSA records. […]

As stated previously, HAVV does not use a truly unique identifier, such as the full SSN to match voter information to its records. In addition, the HAVV program does not allow flexibility with matching the name and DoB to its records to compensate for typographical errors, other common database errors, and mistakes because it does not use the full SSN. Because of the limitations of the matching criteria established by the legislation, HAVV may be providing a high number of false negative responses to the States, which may lead to applicants having difficulty while registering to vote.

Diacritics are only one of many possible sources of error in this process. For example, on various government-supplied forms of identification, my name is listed at least 4 ways:

Mark Liberman
Mark Y Liberman
Mark Y. Liberman
Mark Yoffe Liberman

All of these are in some sense correct, but they're different.

Furthermore, some of these datasets have a three-field model (e.g. First Name – Middle Name — Last Name), while others (e.g. my U.S. passport) have a two-field model (Surname / Nom / Apellidos — Given Names / Prénoms / Nombres). To make it worse, other U.S. Government documents (e.g. the Global Entry system) vary the pluralization of those categories (Surname / Nom de famille / Apellido — Given name / Prénom / Nombre). A version of this innocent-seeming pluralization issue once resulting in a (luckily brief) visa problem for me, since my passport can be misinterpreted to say that my first name is "Mark Yoffe" rather than "Mark". (And after all, there are compound given names, like "Marc-Anthony" or "Jean-Claude", with hyphens or spaces or nothing in between…)

That sort of thing is much more likely to happen to people with Hispanic or Asian names, where the traditional grammar of names fits those database categories even less well.

And so far, we're assuming that some version of the information was entered correctly. For a couple of years earlier in my life, my drivers license had an incorrect birth year, due to a data-entry error by someone at the state Department of Transportation. And I always have to pay attention ensure that I'm Mark rather than Marc, and Liberman rather than Lieberman or Liebermann or whatever — again, I've occasionally run into temporary difficulties because a letter of invitation or a reimbursement check used a wrongly-spelled version of my name. So I can well imagine how large numbers of Georgia citizens can end up with differently-spelled versions of their name in different databases, especially if they're not used to checking up on such things.

And according to the previously-cited court filing,

68. A preliminary review of data produced by the Georgia Secretary of State’s Office on July 4, 2018 indicates that approximately 51,111 voter registration applicants are in “pending” status for reasons related to the failure to verify against DDS or SSA identity or citizenship data. Approximately 80.15% of those pending applications were submitted by African-American, Latino and Asian-American applicants. Only 9.83% of the “pending” for failure to verify applications were submitted by applicants identifying as White.

Just as intended, apparently.

1. ### Philip Taylor said,

October 24, 2018 @ 7:21 am

"Exact match" requirements are not restricted to America; my wife (Âu Dương Lệ Khanh — two family names followed by two given names) recently had enormous difficulty in verifying her identity in the U.K., as her name appears differently on her passport, on her bank accounts, on her credit cards and so on …

2. ### Norman Smith said,

October 24, 2018 @ 8:24 am

Database errors can happen and persist even with common or garden variety names, like mine. For 9 years, I worked at Canadian federal agency which recorded my first name as "Normand". When I asked to have this corrected (which I did about once each year), I was told it would be too dificult to do so – too many records to fix. When I moved to a different Canadian federal agency, I asked again and was told "Sure, no problem!" and the change was effected in about 15 seconds.

3. ### Michele Sharik said,

October 24, 2018 @ 9:19 am

My full legal name is:
First name: MICHELE
Middle name: DIANA
Last name: NIEHE SHARIK PITULEY

I’ve had enormous difficulty getting all 3 of my last names into government databases correctly. Every damn year I get a notice from the IRS asking me to send in a copy of my SS card to verify my name on my tax return. I even went to the SSA office to verify it was correct in their database and they said yes. So I don’t know what the problem is with the IRS.

Plus, the given name I go by is Michèle because it’s French (French-Swiss, actually, as my first and middle names are after my German mom’s girlfriends she used to ski in the Alps with when she worked in Switzerland. So yes, my middle name is pronounced dee-ah-na not dye-anna.) and so there are a lot of places that use the accent mark. But sometimes they use é rather than è! I’ve even seen Ë.

And this doesn’t even count the number of times my first name has been Anglicized to MICHELLE in one database or another.

So, bottom line is that I would probably be affected by the GA law, too. Sigh.

4. ### David Marjanović said,

October 24, 2018 @ 10:33 am

Why is it even legal that a candidate is allowed to make election laws that affect his own election? Doesn't it, like, automatically follow from nemo iudex in causa sua that it can't possibly be!?!

5. ### J.W. Brewer said,

October 24, 2018 @ 10:37 am

The statistical allegations in the recently-filed complaint all seem to lump together "African-American, Latino and Asian-American" would-be voters as an undifferentiated mass. There are no doubt more factors at play here than databases having trouble with non-ASCII orthography and/or non-Anglophone naming conventions, but it only takes a modest amount of cynicism to speculate about what breaking out African-Americans separately from the other minority groups would show. (African-American surnames are overweight British-origin surnames as compared to the general non-Hispanic white population and at least impressionistically, I have not noticed first names that seem distinctively African-American having any notable tendency to include non-ASCII characters.)

6. ### CP said,

October 24, 2018 @ 10:47 am

I used to receive three copies of each Capital One ad sent by mail. One addressed to FirstName LastName, one to FirstName MiddleInitial LastName, and one as FirstName MiddleName LastName. These turned out to correspond to the three ways in which the three major Credit Bureaus are tracking me.

7. ### J.W. Brewer said,

October 24, 2018 @ 10:59 am

David M.: I hate to break it to you, but laws about electoral procedure are, at least in the U.S., invariably enacted by actually-existing legislatures whose actually-existing members are by and large expecting to seek reelection themselves and thus invariably have a keen interest in whether any particular change in electoral procedure will advantage them or disadvantage them and/or their political allies. Here, the lawsuit appears to primarily if not exclusively challenge a new Georgia statute enacted by majorities of both houses of the Georgia legislature and signed by the current governor (who is fwiw not himself seeking reelection – is the AmEng idiom "lame duck" known internationally?).

In other words, the grievance doesn't seem to be that the statute itself is fine or at least tolerable but Mr. Kemp has abused his administrative discretion by implementing it in a fashion the statute itself doesn't require that is suspiciously beneficial to his own current campaign for higher office; the grievance is rather about the statute itself. It is quite plausible, of course, that the legislative majorities that voted for the statute do prefer Mr. Kemp over his opponent in the forthcoming election, but that just gets us back to the underlying problem that legislative majorities are rarely if ever blind to the potential real-world electoral consequences of any electoral procedural reform they are considering.

8. ### Zeppelin said,

October 24, 2018 @ 10:59 am

Assuming incompetence before malice on the bureaucracy's part: African-American first names may cause issues due to variant spellings? Many seem to be "made-up" non-traditional names that don't have an established spelling, or non-traditional spellings of traditional names, so clerks would be prone to entering them wrong.

9. ### J.W. Brewer said,

October 24, 2018 @ 11:17 am

@Zeppelin. Unusual first names and/or unusual variant spellings of common first names may increase the risk of transcription errors for databases that still rely on low-level workers looking at hand-written forms and then keyboarding in their content, but to the extent those are more prevalent among African-Americans that would be offset to some extent by African-Americans being less likely to have unusual surnames ("My last name? Szilyagi. What? Oh, it's just spelled the usual way.") Although certain sorts of "ethnic" white surnames may be less common in Georgia than in other parts of the U.S.

10. ### Jerry Friedman said,

October 24, 2018 @ 11:29 am

J. W. Brewer: The article MYL linked to in his first sentence says, "The match rates using exact matching are nine and six percentage points lower for black and Hispanic voters, respectively, than for white voters." You'd have to read the article more carefully than I did to see how the author determined that. He doesn't suggest a reason. I'm wondering whether it has to do with first names that contain capital letters or apostrophes or spaces, such as D'Sean and La Tanya.

11. ### J.W. Brewer said,

October 24, 2018 @ 11:48 am

@Jerry Friedman, thanks. I had previously just clicked through to the complaint in the lawsuit, not the separate article by Mr. Enamorado. I see that although all racial/ethnic groups do better under his preferred algorithm than "exact match" in his study, whites still do better than all non-white groups. I think you would need to see much rawer data (including representative examples of names that failed to match under either/both approaches) to get a sense of what might be driving the differences, but I am skeptical that non-ASCII characters are the primary factor. As to the apostrophe factor, you'd think that Irish-Americans would have had enough political clout for long enough to get that adequately sorted out for surnames and that any other racial/ethnic group that was more inclined to use apostrophes in given names would be able to free-ride on that.

12. ### Ellen K. said,

October 24, 2018 @ 12:08 pm

I could be missing something, but I don't think there's any claim that African Americans have issues with non-ascii characters. However, non-alphabetic characters (spaces, apostrophes, dashes) will still cause problems even if they are ascii because they may be left out. And as has been mentioned, African Americans are much more likely to have unusual or unusually spelled first names, even when they don't have non-alphabetic characters.

13. ### David Marjanović said,

October 24, 2018 @ 1:18 pm

David M.: I hate to break it to you

I know, I know – I've been following US politics (the most important in the world, after all) closely enough since 1999 to know e.g. what a "lame duck" is. But I'm no less outraged, and still can't quite grasp where more Americans aren't.

Sorry for the oversimplification in "make […] laws".

14. ### Gwen Katz said,

October 24, 2018 @ 6:08 pm

It's also possible that black people in the US receive disproportionately poor customer service, leading to more mistakes on their documents. Studies have demonstrated this in other industries, like restaurants.

15. ### Scott Robinson said,

October 24, 2018 @ 7:41 pm

16. ### Garrett Wollman said,

October 24, 2018 @ 8:29 pm

One potential issue that hasn't been mentioned in comments yet: at least some of these databases probably live, or started life, in IBM mainframes, using the EBCDIC character set, and name forms in these databases may have experienced multiple generations of "code page"-related transformations even if they're stored in ISO 10646 now (which last is still highly unlikely). In the case of things like passport data and travel-industry passenger name records, I believe it's required by international treaty that all names be squashed to ISO 646-IRV, whatever damage that might do in the view of the culture of origin.

I'm lucky in that my whole name can easily be represented in the modern, unaccented Latin alphabet — but there are three different common spellings of my given name, and about eight of my family name, so it's not all joy.

I've also run into issues with travel-industry standards that represent passenger names as FAMILY/ALLGIVENNAMESSQUASHEDHONORIFIC which seems surprisingly difficult for many professionals in that industry to parse. And of course that is the form on the boarding pass that is _supposed to_ exactly match your identity document at the security checkpoint!

17. ### Andreas Johansson said,

October 25, 2018 @ 12:04 am

My full name is Håkan Andreas Johansson: the 2nd name is the name of address, or "first name". This tends to confuse anglospheric computer systems, and has among other things left me waiting for a long time to check in manually at airports because the automatic systems don't accept my passport.

The non-ASCII character doesn't help, but seems to be causing less issues in practice.

18. ### DWalker07 said,

October 25, 2018 @ 9:04 am

For African-American names — at least the subset of those who play professional sorts — there are often apostrophes in first names (like D'Sean which was mentioned above, or D'Andre).

That could be entered into various databases as a "straight quote" or a left-leaning quote or a right-leaning quote, with or without a space on one or both sides — which makes for about 3 x 3 or 9 variants right off the bat. And I'll bet none of those are exact matches with each other.

For this law to work as it SHOULD, there needs to be better rules around "details of the matching procedures used". But I'll bet the law is working exactly as it was intended by its framers.

19. ### J.W. Brewer said,

October 25, 2018 @ 11:23 am

Again, a first name like "D'Andre" would lead to mismatches only to the extent a (characteristically white) surname like "O'Neill" would. And a two-piece name like "La Tanya" — to use an example from upthread — would lead to mismatches only to the extent a (characteristically white) surname like "Van Buren" would. So one should be on the lookout for Irishly or Dutchly surnamed white residents of Georgia who have difficulty registering to vote in order to test the hypothesis.

I expect, fwiw, that it's going to be hard to get good statistical data on the actual prevalence of orthographically unusual first names that look characteristically black, because many government datasets just ignore apostrophes. So "O'Neill" comes out in a Census Bureau list of surnames by frequency as ONEILL, and "D'Andre" comes out in an SSA list of baby names by frequency as "Dandre." As long as the databases that are supposed to match up with each other all treat apostrophes the same way (especially if they do it by pretending they're not there) there's no problem. If they do it differently, then there's a problem.

Georgia is apparently also one of the minority of states that requires most private-sector employers to use the federal "e-verify" program to check the bona fides of new hires, i.e. to confirm that the name etc. they're getting hired under matches up with that of a real person who is either a U.S. citizen or a non-citizen with an immigration status permitting him or her to work legally in the U.S. I would think greater attention ought to be paid to any recurrent mismatch issues (especially for people with names of a particular ethnic background), and how they could be ameliorated, in that context, since being wrongfully denied the ability to hold a job and make a living day-in and day-out is imho a bigger deal than being wrongfully denied the right to vote every year or two. Of course, if the e-verify system has a lower mismatch rate and/or better protocols for resolving initial mismatches than whatever Georgia is doing for voter registration does, that's a helpful example that the technical ability exists to do it better, which might raise questions about why Georgia doesn't, in the voting context, use whatever approach yields those better results.

20. ### Michele Sharik said,

October 25, 2018 @ 11:37 am

@Garrett Wollman: “I've also run into issues with travel-industry standards that represent passenger names as FAMILY/ALLGIVENNAMESSQUASHEDHONORIFIC which seems surprisingly difficult for many professionals in that industry to parse. And of course that is the form on the boarding pass that is _supposed to_ exactly match your identity document at the security checkpoint!”

OMG yes! I run into field length problems all the time. Once when flying via American Airlines, they have me a boarding pass saying NIEHESHARIKPITUL/MICHE (or similar – it was truncated).

And the TSA people wanted me to go back and get a new one that didn’t truncate my name! I asked how they could do that since it seemed to be an issue with field length? Any new pass they gave me would just be truncates some other way. The agent got their supervisor who eventually passed me through. But what a hassle!

21. ### James Wimberley said,

October 25, 2018 @ 1:44 pm

Since I raised the issue with Mark I declare a proprietorial interest in the thread. I'm delighted at the number and quality of the comments it has aroused!

J.W. Brewer : I did not suggest that the character set issue is at the heart of Kemp's outrageous voter suppression ploy using exact match. It does however point to a different legal vulnerability in his scheme. If it can be shown not to be fit for purpose because of issues with character sets (especially if differences are imposed by the design of the databases), then opponents may not have to prove discriminatory intent or effect.

22. ### Jerry Friedman said,

October 25, 2018 @ 1:46 pm

J. W. Brewer: Again, a first name like "D'Andre" would lead to mismatches only to the extent a (characteristically white) surname like "O'Neill" would. And a two-piece name like "La Tanya" — to use an example from upthread — would lead to mismatches only to the extent a (characteristically white) surname like "Van Buren" would. So one should be on the lookout for Irishly or Dutchly surnamed white residents of Georgia who have difficulty registering to vote in order to test the hypothesis.

Indeed one should. But in addition, clerks involved in the process might have a better success rate on familiar names such as O'Neill than on relatively new ones like D'Andre, especially if it's pronounced like "DeAndre" or "De'Andre"—there's no "ONeill" (that I know of) to confuse "O'Neill" with. And if white people get better service in county clerks' offices than people of other races, as Gwen Katz speculated, that might add to the problem.

Also, illiteracy rates are probably still higher among black and Hispanic Americans than among whites, and accent mismatches are possible especially with Hispanic (and other) immigrants, so there may be more people in those groups who have trouble communicating their exact names to clerks. Again see Gwen Katz's speculation.

It might even be possible that shortsighted programmers allowed for features in surnames that they didn't allow for in given names. All just speculation.

23. ### James Wimberley said,

October 25, 2018 @ 1:57 pm

Andreas Johansson: you should be pleased on several grounds that the Danish energy company DONG (a comical acronym from Danish Oil and Natural Gas) has changed its name to Ørsted to reflect a strategic shift from fossil fuels to wind energy. It's sufficiently large and newsworthy that the name is generally written correctly on the clean tech blogs I frequent. In the company logo, the Ø is actually punningly replaced by the common on/off switch icon.

24. ### J.W. Brewer said,

October 25, 2018 @ 2:25 pm

I'm sure not everyone does this (and the rate at which people do do it may not be consistent across racial/ethnic groups), but Georgia is up-to-date enough that you can get the voter registration form on the government's website in a version that lets you type in the relevant information and then print it out and sign it. Presumably the typed version can be scanned/OCR'd reasonably accurately, and this approach thus eliminates errors introduced by a bureaucrat trying to read the name on a form filled out by hand and mistranscribing it. Although if you type in non-English characters like "ä" (which the website didn't block me from doing when I tried) there may still be trouble when the scanned/OCR'd info from the form gets fed into the database.

25. ### J.W. Brewer said,

October 25, 2018 @ 2:57 pm

Update on a related issue in another country that I think was covered on LL some years back. As of last year the Lithuanian courts were finally taking the position that the Lithuanian government could not refuse to issue identity documents with "foreign" spellings using exotic non-LIthuanian characters like "W" and "J." http://en.efhr.eu/2017/03/07/final-verdict-letter-w-identity-card/

26. ### BZ said,

October 25, 2018 @ 3:49 pm

I think people with "foreign" names (such as mine) who are eligible to vote in the US have been in the country long enough to come to some sort of standard (for them) spelling of their names. Pronunciation is a bit harder, though

27. ### James Wimberley said,

October 25, 2018 @ 6:00 pm

The SSN is a near-perfect unique identifier. Duplicate numbers areas unlikely as duplicate email addresses. In the UK, the equivalent would be the NHS number – issued at birth, before it's even registered. The trouble with these when misused as proofs of identity, for which they were not intended, is that they are weakly coupled to other available information. For the SSN, to show that the number xxxxx is actually attached to the John Smith who claims it, you would have to look at the contribution record and match that to an employment history. SFIK that would not be available without a court order. Using the NHS number is even worse because the relevant records are confidential medical files.

28. ### James Wimberley said,

October 25, 2018 @ 6:21 pm

J.W. Brewer: The Georgia pdf registration form did accept the Vietnamese string chữ Quốc ngữ (which Wikipedia says is the name of the script). The chances of Brian Kemp's software dealing with these characters correctly are, I fancy, rather slim.

The form includes this delightful White Knight section:
"Are you a citizen of the United States of America? Check One: Yes/No
Will you be 18 years of age on or before election day? Check One: Yes/No
If you checked “No” in response to either of these questions, do not complete this form."

29. ### J.W. Brewer said,

October 25, 2018 @ 7:45 pm

James Wimberley: one irony is that the problem arises in part because of historical Anglo-American reticence about truly universal/mandatory domestic identity documents, which somehow give off a tyrannical vibe to certain Anglo-American sensibilities. If there's one universal Panopticon database, and/or everyone has barcodes tattooed onto their arms or chips implanted into their heads, these mismatch issues are unlikely to arise, yet somehow that seems too dystopian a solution.

A pretty large majority of Georgia citizens over the age of 18 will have a driver's license and some of those who don't drive will have the state-issued ID card issued by the same agency (such an id being useful for non-driving activities like buying beer, boarding airplanes, and the like). If you have either of those, you have a unique- identifier license number in the DDS database, which is tied to: a) some version of your name; b) your date of birth; and c) a photo (not necessarily recent). One would hope that the DDS database, being maintained by a Georgia state agency, would be fully compatible in how it organizes/codes things with the database run by the separate Georgia state agency that handles voter registration, although maybe that's too optimistic. But if you've got a DDS-issued identity document, you can look at it to see which version of your name is on it, which is presumably the version of your name in the DDS database, so if you use that exact version of your name (if there's a typo, just copy it … and do whatever it does or doesn't do with apostrophes and umlauts and what have you) on the voter registration form your risk of mismatch would hopefully be quite low.

But because Georgia does not compel 100% of the population walking the streets of Georgia to have obtained a DDS-issued identity document, the DDS database is incomplete, and therein lies the problem. Similarly, if the U.S. took the approach traditionally followed by certain less liberty-oriented foreign regimes and required parents to pick names for their children off a pre-approved list, many of the other difficulties referenced above could be avoided. But we don't do that, because we value the right of parents to give their children names of their own choosing, even when those names are so exotic as to predictably doom the kids to a lifetime of bureaucratic hassles and confusion.

30. ### Gwen Katz said,

October 25, 2018 @ 10:48 pm

The SSN is a near-perfect unique identifier. Duplicate numbers areas unlikely as duplicate email addresses. In the UK, the equivalent would be the NHS number – issued at birth, before it's even registered. The trouble with these when misused as proofs of identity, for which they were not intended, is that they are weakly coupled to other available information. For the SSN, to show that the number xxxxx is actually attached to the John Smith who claims it, you would have to look at the contribution record and match that to an employment history. SFIK that would not be available without a court order. Using the NHS number is even worse because the relevant records are confidential medical files.

The other problem being that, while everyone who can legally vote is (as far as I know) eligible to get a social security number, no one is required to have one. Yes, it's essentially just fringe sovereign citizens who don't–but they're legal voters too, and a system that bars even one rightful voter, or requires them to do something that isn't a legal voting requirement, is a system that needs to be improved.

31. ### Andreas Johansson said,

October 25, 2018 @ 11:50 pm

James Wimberley wrote:
Andreas Johansson: you should be pleased on several grounds that the Danish energy company DONG (a comical acronym from Danish Oil and Natural Gas) has changed its name to Ørsted to reflect a strategic shift from fossil fuels to wind energy.

What would those grounds be? I mean, they're hardly the first or most notable company with a non-ASCII letter in their name, and it's not even the same one as in my name. It's not even one that exists in Swedish.

32. ### Andreas Johansson said,

October 25, 2018 @ 11:57 pm

Seems I just had a post eaten, apologies if it turns up another one saying much the same as this one. Anyway:

I'm not sure why James Wimberley thinks that I, particularly, should be pleased at DONG is changing its name to Ørsted. It's not likely to improve the way my name is handled bureaucratically anymore than any other company name that contains a different non-ASCII character from the one in my name.

(In case the characters don't show up properly for someone, the second letter of Håkan is an 'a' with a ring above, whereas the first of Ørsted is an 'O' with a diagonal slash across.)

33. ### ajay said,

October 26, 2018 @ 3:55 am

especially if it's pronounced like "DeAndre" or "De'Andre"—there's no "ONeill" (that I know of) to confuse "O'Neill" with.

I've never met an Oneill, but there are certainly Omalleys around in the UK.

And, a bit of googling shows, the US has at least a few Oneills.

34. ### DWalker07 said,

October 26, 2018 @ 9:13 am

@J.W. Brewer: I wondered if someone was going to mention the countries that require a baby's name to be selected from a pre-approved list. I *think* that many hospitals in the U.S. did this many years ago, but no longer. And of course, there's no reason to require names to be selected from a list. Hmmm.. I wonder if my next child can be named 3947598345? Or #$%#$%%#?

35. ### Gwen Katz said,

October 26, 2018 @ 1:04 pm

I knew a guy named Eric who changed his name's official spelling to 3ric. Boy did he crash a lot of computers.

36. ### Grover Jones said,

October 26, 2018 @ 1:52 pm

Keep in mind that those with "pending" status can still vote, simply by displaying a driver's license or ID card. That would be a regular ballot, too, not provisional.

One theory for African-Americans' being 70% of the "pending" category. A lot of that can be tied back to Stacey Abrams' New Georgia Project, a minority-voter-registration group, which has run into trouble with forged applications. Further, the group does not use online registration; only paper ballots.

37. ### GALESL said,

October 27, 2018 @ 10:35 am

On the issue of civil registries in the US handling African-Americans properly even when their names have no non-ASCII characters, perhaps we could ask Condolcezza Rice and Orpah Winfrey.

38. ### Michele Sharik said,

October 27, 2018 @ 12:30 pm

@JW Brewer: “ But if you've got a DDS-issued identity document, you can look at it to see which version of your name is on it, which is presumably the version of your name in the DDS database, so if you use that exact version of your name (if there's a typo, just copy it … and do whatever it does or doesn't do with apostrophes and umlauts and what have you) on the voter registration form your risk of mismatch would hopefully be quite low.”

That wouldn’t work for me. My California Driver License displays my last name as “NIEHE SHARIK PITULEY”. However, my last name in the DMV database is listed as NIEHESHARIKPITULEY, which I only know because the DMV clerk told me about the workaround she has to use when I first got my license in that name. It seems their DB doesn’t allow spaces in names. If I had used hyphens to separate my names, it would’ve been fine.

So typing in my name as it appears on my DL would not produce an exact match with the DMV DB. (Unless they also compare it to the (presumably saved in the same DB) display name.)

39. ### Rachael said,

October 27, 2018 @ 2:26 pm

This is why we named our daughter Zoe rather than Zoë – the latter would be likely to cause her a lot of hassle from badly-written databases when she's older.

40. ### Kenny Easwaran said,

October 27, 2018 @ 3:40 pm

Georgia has a moderate white population of Irish ancestry (and they are probably heavily represented in that small fraction of white people who aren't getting exact matches). But it's notable that a lot of black people have Irish last names (From Shaquille O'Neill to Kwame Kilpatrick to Deval Patrick), in addition to often having more creative and less familiar first names.

41. ### James Wimberley said,

October 28, 2018 @ 7:49 am

Andreas Johansson:
General grounds for welcoming the change of name from DONG to Ørsted:
– DONG sounds silly.
– Wind energy is important in the energy transition; we want its proponents to succeed and gain public recognition.
– Wider recognition of the cultural importance of diacritics is valuable, and as Ørsted is newsworthy, the change may promote this.
– The new name recognizes a great scientist.

Grounds why someone with your name might welcome the change:
– Reason 3 above becomes personal.

42. ### James Wimberley said,

October 28, 2018 @ 8:01 am

Child names subthread: What countries still insist on a closed list of acceptable names? France used to, but this was abandoned several decades ago. The list had become absurd – it included numerous worthy saints nobody has heard of, and excluded Rose and Florence.

There was IIRC a cause célèbre involving a Breton nationalist with a large brood of unregistered children named Gorboduc and the like. Was she required to send them to school? How should the school record them? Family allowances?

Mind you, complete parental freedom over naming can tip over into child abuse.

43. ### Andreas Johansson said,

October 29, 2018 @ 5:08 am

@James Wimberley:

Thanks for the explanation. I'm afraid I'm deeply skeptical, though, that Ørsted's new name will make any appreciable difference to the handling of diacritics.