"... Why doesn't he focus on web.py and promoting it? ..."

The focus is there. I've been lurking the web.py mail list for a long time (Jan 19, 2006 ~ http://www.flickr.com/photos/bootload/88347878/ ) and there has been a steady progression of code since then. So in my view it is being promoted, worked on and improved.

   "The programmers were in the corner doing what they were told. 
    That's one reason they were so easy to outsource. If a programmer 
    never really talks to the customer, never thinks, just solves little 
    puzzles, well that's a perfect candidate for something to offshore." 
    ~ Founders at Work, Ch24, Philip Greenspun, P325, Livingston, 2007.

But instead of just promoting just another bit of technology, maybe it's better to do something with it. A lot of technology is commodity that can be replicated by some programmer in the corner. Build useful things with it, creating value is a lot harder to commoditise.

Some interesting and useful things have to be done as companies rather than open-source projects.

And some interesting and useful things have to be done as open source projects.

Like programming languages:-)

What are those, in the software world?

I can only think of projects that involve a lot of drudge work that nobody wants to do unless they're paid to. I suppose this includes most services. What else has to be done by a company?

Any software project, no matter how cool it seems, is at least 50% drudge work. I'd be willing to argue that the inverse-pseudo-pareto-principle applies... 20% of any non-trivial software project is fun, and the other 80% is drudge work.

Even certain open source projects require paid staff. The only reason FireFox exists is because the Mozilla foundation can pay programmers to work on the ugly parts.

OK, so that would prove that open source can get drudge work done, right? Or do you think 80% of open source work is sponsored by companies? For all I know, that might be the case, but are sponsored efforts any less open source?

The sponsored efforts are indeed open source. However, the only reason the drudge work got done was because someone was paid to do it. People don't like doing crappy stuff if they aren't getting paid.

I can think of other reasons why people would do boring stuff. This is by no means an orthogonal breakdown!

- Militance: They believe in a cause and do some sacrifices to further it. Like translating software and docs into an obscure language. This can be done by zealots of the software, of the language, or both.

- Self-image (somewhat related to the above): After having completed the sexy parts, you care about the appearance of your project as a whole, so you fill in the boring parts.

- Pleasing users (somewhat related to the above): You get hooked in a virtuous feedback cycle with your users. Some things that would be boring in themselves get done in anticipation of the positive feedback.

- Competitiveness: a mix of the two above. You want to be the best at what you do, and the boring stuff offers opportunities to differentiate.

- Scratch a personal itch: I think many Linux device drivers get done because of this. Also most of the external bug reporting (and most of the business sponsorship.)

- The long tail of vocation: what is boring to most is probably someone's joy.

- Retribution: You love an open source product as an user, so you give back.

- Social acceptance: You want to become part of the community of contributors. You aren't qualified to do the hard stuff yet.

In MMOG's, lots of people volunteer for Game Masters, which is an euphemism for unpaid customer service. They do very boring, repetitive work for the benefit of a for-profit company that gives them absolutely nothing in return (indeed, it's legally dangerous to give them anything in return).

So don't underestimate people's ability to do drudge work for free.

With regards to people working as Game Masters in MMOG's, I think it's mostly done for the feeling of godlike power.

In the game I worked in, most had no power whatsoever in the game. The only power some of them had was moderator role in the help chat channels, and the ability to teleport their GM characters (not their playing ones) to check out problems. They were watched closely for abuse, so you can hardly call that power at all.

You can say they were motivated by a feeling of notoriety. In my area we have problems with people that would sneak and set woods on fire only so they can feel important helping to extinguish the fire. They get no tangible benefit from it, not even power, and they risk shame and jail if they get caught. Feeling relevant is a strong motivation for some people.

Google probably has to, for example.

Yes, Google is a service. But then again, Debian repositories are a service, as is Wikipedia, as is Usenet (although you said 'open source', I'm assuming your argument extends to donated effort in general). So there must be some other difficulty to Google.

There are high stakes, business wise, in web search. But then again, that was just as true for operating systems and web browsers. So cross that one too.

Google is resource intensive. Open source is good at providing manpower, but not so good at sustaining huge running costs of other type. But then again, Google started quite small, and I'd guess Wikipedia, Usenet or Debian need more resources than many small companies. So, is it only a matter of size? In any event, this doesn't apply to Aaron's new startup.

Anything else?

Just came to mind: an area where open source doesn't even try to go, AFAIK, is services that require you to entrust them with sensitive personal data. For example, I don't think there's any volunteer run service that asks you for your credit card info, or that hosts your personal email.

I don't think it's only (or mainly) a matter of accountability. In general, volunteer services are given for the benefit of a community. Services targeted to individuals (except casual social interactions like answering questions) don't attract much volunteer effort, and are generally provided by for-profit companies.

Anything that creates obligations essentially has to be done as a company. Open-source programmers typically tend to be leery of taking on projects where they have to do something, with good reason. Also, contracts generally aren't enforceable if there is no money (or other "consideration") changing hands. Anything where you want to be sure that something gets done in the future basically needs to go through the real economy.

The examples cited so far are basically special cases of this. Google needs to be able to enter into contracts for hosting, bandwidth, power, uptime, etc, so it needs to be a real corporation. When you entrust sensitive personal data, you want an ongoing obligation to keep that data secure, which you can't get with open-source. Additional examples might be corporate support contracts, support for certain platforms, and any service dealing with payment or finances.

Basically, open-source is good for sharing what people have done, while corporations are good for determining what people will do.

Very lucid, thanks.

I don't think there is much point in telling him what would be more fulfilling to him.

Ohh but how your perception will change when you see his site achieve a state of perfection:)

Hah, anyway I myself do enjoy webpy and wouldn't mind giving something back to it, but if I were him I'd probably rather do a startup.