An email came in to the creativematch support desk last week from a rather bemused June Dobbs who was having a problem adding a post to the forum. June has a collection of old design magazines looking for a good home. The error message said the posting contained offensive material, however neither June or I could see anything remotely offensive in what she had entered.
After studying the forum code for a while, I found the problem was that the magazines had accumulated. The real cause though was that, many years ago, someone had added a check for offensive words after we had been suffering from an attack by a spambot. It was a quick fix and, if you'll excuse the pun, a bit crude. The result was that part of the word accumulated was being match to one of the filter's bad words. It reminded me of a problem I'd heard of before, and in fact it turns out that this bug is now known at The Scunthorpe Problem.
It also reminded me of a story I heard last year where the Christian right organisation, the American Family Association, ran a news site on which they automatically replaced the word gay with the word homosexual, with rather unfortunate consequences for an article on the sprinter Tyson Gay. They have fixed the problem, but luckily you can see a screengrab on this article on the dangers of auto-replace.
Again, there are numerous examples of auto-replace problems, try typing Scarlett Johansson with autoreplace on and you'll probably lose one of the t's. Our branding guidelines say that creativematch is one word with a lowercase c, but I've given up trying to stop Outlook from making it uppercase. A more annoying example was when someone was emailing me a case-sensitive password which they put at the start of a line, and their email client was uppercasing the first letter.
The problems aren't particularly new either, I remember a story I heard when I was at college of a computer translating "Hydraulic Ram" into "Water Sheep". I searched the internet for more details, but it seems this may be an apocryphal story, or might not have even been a computer translation. I did discover a story about an expensive program developed by the EU that translated "vis à vis les fermiers normands" as "screw the Norman farmers" and "nous avions" as "we aeroplanes".
After a small tweak to the algorithm I managed to add the posting, so if you're interested in a collection of design magazines or know anyone else who might be, please contact June via the forum . If you have any other amusing tales of mistranslation and correction let me know.
Posted By Adrian Marshall at 9:57 PM in Category:
Programming
Link directly to this article.