Scheitern durch technik

[Dave Birch] An unexpected start to the year. With the Year 2000 well behind us, I hadn't heard any technical experts warning us about the Year 2010. But there has been a Y2.01K bug.

German banks computers think it is 2016. NO credit or ATM cards work!

[From German banks computers think it is 2016. NO credit or ATM cards work!]

Card transactions in a number of countries failed because of some software that couldn't handle the year 2010. As I understand it, it's something to do with binary-coded-decimal, or BCD, a prehistoric means of encoding numbers in computer systems that is used in the EMV standard). Germany was one of the places where things went bonkers: the POS terminals thought that it was 2016 and therefore some local cards wouldn't work properly because they thought the date was after their expiry date (Visa, MasterCard and Amex are not affected because their versions of the EMV application don't process the incoming date from the terminal).

A "Year 2010" software problem triggered by the change of decade has affected 30m German debit and credit cards and left bank customers unable to make purchases or make cash withdrawals.

[From FT.com / UK - 'Year 2010' software glitch hits German bank cards]

The same problem cropped up in Australia, so the banks there went for a quick hack (which I imagine was was to tell the terminals to fed an old date to the cards) to get things moving again.

The computer bug that brought Bank of Queensland's Eftpos transactions to a grinding halt in the first days of the New Year has hit other banks - including the Commonwealth Bank-owned BankWest. The glitch, which started on January 1, left Australian retailers struggling to perform routine electronic point-of-sale transactions... As of today, transactions are currently being processed at their normal volumes, said Keycorp, the Australian-provider of the EFTPOS machines. However, the receipts are still dated 2016.

[From Welcome to 2016: Eftpos glitch spreads | BoQ | EFTPOS | BankWest]

Note again: it's only the local cards that are affected, not the Visa/MC/Amex cards. It turns out that payment systems developers aren't the only people who get confused writing software to deal with complicated things like dates.

WM Experts reports that, text messages received after 1/1/2010 may be dated as 2016. This bug is reported in Windows Mobie 6.1, Windows Mobie 6.5 and even in some non-Windows Mobile phones. Only some mobile carriers seem to be experiencing this bug. WM Experts say that HTC and LG are aware, but no official response yet.

[From Year 2016 (Y2K16) Computer Bug Hits Texting in US and Banking in Australia]

The source of all these problems? "10" in hexadecimal is "16" in decimal. My guess is that the terminal software has a date field that is in BCD (which is the what EMV specification requires on the cards), but it's being read as hex. For 0 through 9 that doesn't matter, but come 2010 the discrepancy means that the terminal reads "10" as 16 and now thinks it is 2016.

bugs and mistakes, EMV, retail

It's all a bit embarrassing for a technology-centric industry at the beginning of a new decade. In Germany, the authorities were on hand to provide advice.

The German Savings Banks and Giro Association (DSGV) ...did recommend that people keep enough cash on hand just in case the cards don't work.

[From Carry Cash as Well as a Card, Says German Bank Association - PC World Business Center]

Unsurprisingly, a few people e-mailed this link to me to highlight the absurdity of my position that society would be better off without cash, but I just don't make this connection. For one thing, when cash does disappear from retailers in the future, they won't rely on a single (eg, EC Card) system to replace it: there will be a variety of choice ranging from "traditional" cards to mobile phones to online and so forth. I could well imagine being in a shop where the EMV terminal has failed, so I take out my iPhone and PayPal the guy the money instead. Yes, that will take a minute instead of 10 seconds, but since it will only happen once every few years, so what? The fact that someone wrote some buggy software and that someone else didn't test it properly (presumably because they were under pressure to cut the cost of the product) shouldn't undermine the more general point: e-payments are a good thing.

These opinions are my own (I think) and presented solely in my capacity as an interested member of the general public [posted with ecto]