StorJ is an awesome idea, but at this point it's not realistic. What makes people think it would be able to compete with services like Mega.co.nz which has developers working all the time to improve the product?
Sure, StorJ could create a bounty to have a developer improve it, but what happens when the developer scams the bot? Well the bot could have unit tests, and the code would have to pass the unit tests before the developer is paid. The bot can't write those unit tests, so a human would have to. So that human could create the unit tests in such a way that the future scammer developer's code passes the test even though the code sucks.
Don't forget the fact that APIs change and many bitcoin services go out of business. StorJ would be completely crippled by something as small as an API change.
It comes down to the fact that a human can, and does, run web services better than a robot does. So why use the robot? A lot of people commenting in that thread are clearly not programmers. They say things like "the bot could get a degree from an online school". Yeah, maybe in the distant future when AI is much, much more advanced. But if you're a programmer, it's plainly obvious that this isn't realistic right now.
Don't get me wrong, it's a totally cool idea, and someone could build it. But we need way better AI for it to work well.
StorJ requires a lot of component parts, but I think people are missing some of the easy solutions. Lets take your API change case. Well yes this would be crippling to an old StorJ(and make it fail most of its unit test), but using good and annotated unit tests StorJ would easily understand which particular API is failing. If everything was double implemented it could just switch to an alternative service. If not it would seek out a StorJ instance that has claims to have fixed the problem. It would use several methods to verify the code and then implement it.
