Joel on Google

I was reading this piece from Joel Spolsky on why he thinks Google is the new IBM, and not in a good way. The new IBM poised to be toppled by an upstart new Microsoft. It’s an interesting set of work, but I don’t agree with almost anything he says.

The big analogy is, more or less, that Google’s efforts these days are like IBM’s Lotus 1-2-3 back in the day, which was dominant all the way up to the instant when it was irrelevant. He tries to draw a bunch of parallels between Google and IBM from back then and put forth his theory that Google could be in trouble.

I have a lot of nits and a few real issues. There are a bunch of things he says that I don’t agree with but are more or less quibbles and differences of opinion. For instance the success of Java. To him the success of java was that it was more portable than C. To me that portability was the initial marketing hype, but really the success was driven by the fact that it was a “compiled” language which gave it the distinction of being a “real programming” language instead of a “scripting” language - which, for some reason, was important to people back then. Not only that but it was an OO language that didn’t have pointers and did have built in garbage collection that made it an easier and safer language to work with than what was currently out there.

He talks about the futility of optimizing code for current hardware and that the right approach is to simply keep adding features. While I think he’s right in principle, another factor to bear in mind is that we’re getting a whole new crop of mobile devices that have small CPU’s and small memory, these will increasingly be used to access web apps, so to some degree, optimizing code lets current machines run it well and these new mobile devices will catch up quicker in the future. Going with big bloat now will mean those devices will never get to the point of being able to run them.

But really those are just opinions and his views are pretty interesting. The thing that really bugs me about the piece is that he assigns to Google a stunning stupidity and blindness that, as far as I can tell, is completely unwarranted. He puts Google into some hypothetical situation and then makes up dumb outcomes to support his theory. Here’s my example of what he’s doing. “So in this situation a rich sugar-mama decides she wants to give Felix a billion dollars. Felix sitting smugly at his desk thinks he’s rich enough and tells her to step off. So she gives it to someone else who is now really rich and Felix feels dumb for having turned down the money.” Well, yes, if that happened I’d feel dumb, except for the fact that I’m not a moron and if anyone out there wants to shower me with money they should feel secure in the knowledge that I would happily accept it.

That’s exactly what he does when he invents his “NewSDK” which has a built in magic wand to do everything really cool and cross platform. Google, in his scenario, strokes it’s collective moustache and laughs it off. Then the magic SDK takes off, everyone starts using it immediately and Google is completely blindsided. What? Does he genuinely think that Google doesn’t monitor and evaluate everything that’s going on in the space? That this company of engineers simply sits on their thumbs all day watching porn and playing Halo 3, not necessarily in that order? It seems about as likely as me turning down a billion dollars. (Seriously, if you want to give me the money get in touch, I won’t say no..) Far from it, it is completely within the realm of conception that Google could create and distribute this magic SDK which would come built in with easy integration with all the Google apps there are.

The other thing he’s decided is that Google is not forward looking at all. That is, they don’t understand about how quickly hardware and bandwidth will become available and are not at all prepared for the future. A company that has built these crazy data centers out of commodity hardware so that each bit is replaceable by another. He thinks they wouldn’t be aware of the future capacity of hardware, right… because they don’t monitor and forecast these things at all. Why would they? It’s only crucial to their business. And bandwidth, they got no ideas on that, it’s not like they’re buying up every last scrap of available capacity and laying down more when there isn’t enough. Oh wait, what? They are? Oh.

It’s an interesting piece and he lists some interesting general thoughts on how people should approach developing. But the tie in with Google is tenuous and contrived - I kinda think the whole piece was written as linkbait. Which obviously worked on me. What a weak-minded fool I am. So, linkbait? Or what am I missing?

Sep 27th, 2007 · Comments RSS · · · · · Tags: development · google