“(c) 2010” is already on the web for massively scalable services? How?

As someone avidly interested in web services and their path from concept in a vision planning room, to the point where they leave their impact on the world, I find it fascinating that Facebook, Microsoft, and Google have changed their (c) 2009 to (c) 2010 on their front pages, but Yahoo, Twitter, Amazon, and even Yelp still read (c) 2009. I have my own reasons for finding that interesting, but that’s not important. What IS important is when the flip happened, and the deep understandings of each that the answer to ‘When?’ represents.

Sure, I don’t expect most people to care. This might seem like an utterly trivial, ‘who the hell cares?’ question to many, but when you think about it, at any given point, every piece of the a page’s visible design and text has to pass through one person’s yes/no decision, and one computer (likely unencrypted). Since there are so few software companies operating websites at distributed scales like this, there isn’t a lot of non-proprietary software that can manage these gargantuan services. This probably means the service management software isn’t very well updated with the latest data visualization software. Every change that might occur is either a bug, or something that was approved in a very deliberate fashion. Which is why when I see that several sites have already updated their copyright year to 2010, I realized that they must’ve had a team of people that discussed the importance of changing to 2010 immediately after the new year.

What kind of mindset is needed to get this resolved? I mean, at some point, close to the end of the year probably, some person had to stand up and say, “Hey, if we don’t do this when the year runs around, then we’re gonna look stupid.” And because of the yes/no process that ended in agreement with this guy, someone had to stay up and edit the year in, sync it across all required web-facing servers hosting the site, and then finally, produce it on demand, to the world. What I’m really curious about is speculations on the drama that could’ve surrounded this process.

I wonder… Did the engineer really stay up that late at work (on New Year’s Eve, no less) just for this? Or was there a decent chunk of dev/test resource time spent to make it happen with server-side code? Could he just pay Chinese gold farmers a couple bucks to press the button at the right time?

Let’s remove one simple but hugely complicated problem just to make the scope of the 2009->2010 challenge easier to understand than it actually is, and assume that there’s only one timezone to care about. (Actually, this might be more true than not, since most sites only have a centralized team managed page). Or maybe there was a legal team that warned the product groups that there was a potential for lawsuits if the year was misrepresented on a site with their logo. Nevermind the ridiculous legal precedent that must’ve been quoted to give the lawyers fear in the first place, or the BS they must’ve used to convince the PG it was important enough (heck, maybe the legal department was just playing a practical joke). Okay, fine, that last one is a little too unlikely.

I’m going to bet that this miraculous change was the decision of a single engineer, with a religiously singular internal desire to see this ridiculously low-pri task completed, and sitting on the cloud by 1/1 12:00. Probably a program manager, but with the collaboration of someone with world-server access. They probably worked with someone in operations that snuck it in at the last second before the RTW VS. [This could never happen at any of these, now-massive corporations… or could it? 😉 ]

Of course, my now immersed-in-the-web mind wonders some more: How was this change done? I wonder if a clever engineer decided to hardcode a date change script into the next build. What did this code do?
Would it still do it’s thing when the year rolls to 2011… and not throw some unexpected exception and take down the livesite? Could it really have been a Find-Replace that just changes all instances of “2009” to “2010”? Nah, it wouldn’t be that stupid, or we’d be seeing at least a few random instances of 2009 (that is specific to 2009) turn to 2010.
Maybe he/she made the change and had a script upload the new version at the right moment? Would they be able to secure buy-in from the deployment team for that?

[to be continued…]