Simon’s Blog

It’s funny. Just a couple of days ago, I was reading a post on Scoble’s blog, entitled - Google, The World’s Largest Startup? The post was all about Google APIs and widgets, and how great they are. The comments section made for some interesting reading, especially comment #7 from Google’s Mark Lucovsky. Mark said, in response to someone questioning Google’s approach to API development:

Periodic small updates (to APIs) != inconsistency or broken apps or failures.

I was going to respond, saying that Mark was quite wrong… except I changed my mind. Why? Because, technically he was quite right! But, I still felt uneasy, because I felt he was right only in the letter of what he wrote. The spirit felt completely wrong - because, the truth is, things like this tend not to turn out well in the real world. To those who haven’t been around for too long in the software business, you might say Google’s method of kicking APIs out of the door as fast as possible is a stroke of genius. However, to those who’ve been around the block a few times, it looked like (or should have looked like) it might be an accident waiting to happen.

What’s the issue? Simple - it’s one of business commitment. If there’s there’s no good business reason for a company to keep a particular API going, they’re going to kill it. The company most likely won’t care how much time and money people outside the company spent building on that API. And, if a company keeps shoving APIs out the door, as fast as the company’s developers can implement new methods on the server-side, there’s clearly been very little thought given to the business reasons for implementing the API features. And that means, there’s no reason to expect any commitment from that company to maintain the API.

But hey, I’ve got an open mind, I thought, as I was reading Scoble’s blog. Google might be different. I’ll keep my mouth shut (for once!) and not respond But then, today, I read Tim Bray’s blog entry - the Unmashable Web? saying that Google just deprecated their entire Search API. And, in case you don’t know what deprecate means… it’s the first step on the road to killing something.

There’s lots more that could be said on this, but let’s cut to the chase… One thing we now know to be true: if you value your time, and your (or your investor’s) money, you need to think rather carefully before investing in building on Google APIs in future, because Google has shown us today that they are simply not commited to maintaining them.

I’m not saying Google is wrong to kill its search API. I’m sure the business reasons for doing so are sound. People just need to realise what they’re getting themselves into when they start to build on other people’s APIs… especially “software as a service” APIs, where you have zero control over the system.

[Update: you might want to follow the discussion over on Scoble]