I do some part-time work at a non-profit that, ultimately, does super-international (30+ countries) surveys. As you might imagine, this results in a massive database and my main duty is wrangling the data into a .csv-friendly format.
One of the quick things we learned that "normalization" in the naive sense was an exercise in idiocy.
Breaking names into 'first name' and 'last name'? Hah, right.
Breaking locations into 'country', 'state', and 'city'? Hah, right.
The big solution: don't constrain what someone types in. Constrain how you work with it.
Funny that I hit this all the time with Japanese services online. Enter First Name in latin characters: "sorry, invalid characters" ... enter in full width latin characters .. "sorry, invalid characters"
I've even encountered sites which only allow hiragana for names - won't even accept katakana!
Until patio11 amends this post to contain suggestions about doing things right instead of merely telling people that they are doing it wrong, please stop recommending it as reading material. Thanks.
I remember signing up to websites in the late 90s that hadn't quite grokked internationalization. "Select your country (required)" , "United Kingdom" , "select your state (required)".
Apparently there is such a thing as Alabama , England. Who knew?
The first name and last name thing annoys me the most. I can't use my real name in most online services (yes, Facebook and Google, I am not using my real name, coz I can't!), because my official name is a mononym. When registering for PayPal, I had to sent at least 10 emails to get someone to respond to be able to use my real name for transactions.
Accurately and faithfully recording things as they are in the world often runs afoul of the preconceptions of data monkeys.
The usual examples of things that sometimes just have to be correctly dealt with are non specific gender [1], family trees with actual inbreeding and cycles (omg; it's a graph not a tree :/) and flying about Fiji means you can cross -179.99 to +179.99 and back again several times in an hour (oh shit, the plane's upside down again :/)
If you go back far enough, beware February 30th, 1712. It was a real date in some parts of the world, and somebody's great^n grandfather was no doubt born on that day.
"If you go back far enough, beware February 30th, 1712"
OK, my girlfriend and I looked it up. All I can say is this: experience matters. These are not things they teach in school (not any school I have been too, anyway).
I claim I was born in 1910 on Facebook without any issues. I backdated my age and then started posting pictures changing the dates to the 1910s and 1920s (of things like me in period clothing standing next to older buildings during the filming of an older film; with cars from the era parked out in front, a fruit vendor on the side of the street, etc.)
This is because Facebook doesn't allow you to backdate prior to your birth - they really don't understand personas at all.
If I want to post a picture of me buying a WW2 era newspaper from a paperboy and claim it was taken in 1941 then why can't I?
It's high quality content that people enjoy (compared to someone posting a picture of say, a hamburger) - isn't that what's going to make people continue to have FB Time and not have tumblr Time - the quality of the content?
Youtube also has problems with this - I don't want my name, I want an invented thematic identity; a character with its own soul. Thank you.
Is it just me or the title is a bit sensationalist? Surely they could done it a bit better: "Facebook registration doesn't allow dates before x" - which is somewhat understandable, and yes this does indeed force her to "lie" about her age, but to make it on the Times with such an accusatory article makes me question the motives/credibility of the author/reporter.
You know, at first I thought this was a lot fluffy for here, but, it is amusing to see yet another case of a company that demands you use your real info turning around and preventing just that. I almost hope facebook kicks her off because of lying, just for the shitstorm it would raise.
So she was born in 1908, but Facebook changes it to 1928, yet lists her as 99 years old? Something doesn't add up there...
Sounds like some BS journalistic license has been taken.
Given that Facebook doomsayers consider it "over" when FB loses the teenage/hip crowd, perhaps this is a broken-by-design feature to keep the elderly from making Facebook uncool. Kind of like a digital bouncer that keeps the uncool people from taking valuable space in a crowded club.
(I'm kidding, though I wonder when digital services really will do this at some point)
I have an Excel spreadsheet with rainfall records dating back to the 1860s. It turns out Excel does not accept dates before January 1, 1900 as valid, while dates far, far into the future (up to December 31, 9999) are perfectly ok. Nobody thought someone might want to enter a date before 1900? It's just bizarre.
Even if not technically-required, I would advise centenarians to roll back their reported age a few decades, anywhere it may be used to target ads. The "meet other 100+ singles!" ads are really obnoxious.
One of the quick things we learned that "normalization" in the naive sense was an exercise in idiocy.
Breaking names into 'first name' and 'last name'? Hah, right.
Breaking locations into 'country', 'state', and 'city'? Hah, right.
The big solution: don't constrain what someone types in. Constrain how you work with it.