If your app isn't successful then there's nothing to worry about.

But what if your app is super successful? How would you ever be able to secure enough servers to meet the demand without burning a very deep hole in your pocket?

What we need is a Y-Combinator like incubator for Facebook Apps - a company which will host your app and help you scale it up for a chunk of equity.

(a) Economically hosting is a commodity. It's not an area where we can add any special value. We don't do hosting for the same reason we don't supply office or living space: it's better for founders just to buy these on the open market.

(b) Hosting is a technical problem, which it's good for founders to have to solve. We're happy to deal with all the paperwork of incorporation, because you really don't have to know about that stuff to run a company. (I haven't read half our legal documents myself.) But startups do have to understand how to manage servers.

- You don't get a static IP address, so if your instance fails and needs to reboot, you might end up with a different IP address.

- If your instance fails, it also loses all the data stored in it.

- There is no hardware load balancing.

With that said, you can probably use it to complement your other servers by doing some of the batch processing.

But PHP is not one of them, it's perfectly suited for the web. I personally find it easier to identify blub programmers than blub languages ...

As long as it resolves symbols (variable and function names) at run-time instead of compile-time, it can't be efficient.

That includes all dynamic languages i.e. Lisp, Ruby, Python, Perl, Smalltalk, JavaScript, Eiffel, Erlang, Forth, Lua et al.

No programming language is straight-forward unless you know all internals.

True, but as you say, that applies to any language. PHP is a very easy language for novices to pick up.

OOP ... gives nothing but an illusion of maintainability, while in reality you end up

In my experience OO code tends to be easier to maintain, and so long as you don't over do it, efficient.

Only good thing about PHP is that it's a dynamic language with dynamic strings and lists/arrays, it is embedded (in HTML) and is largely supported. And that's all you need for server programming on the Web.

That's why I recommended it as a good choice for web programming.

Ok, but you said PHP is efficient. I say, PHP is efficient as much as the language itself allows it to be. If there are many late-binding languages around and PHP at least doesn't look bad compared to them, it doesn't mean it's efficient. Late-binding offers some new possibilities in programming of course, but the price is high and programmers should be aware of that. And when it comes to computationally intensive tasks (encryption, graphics, any hardware-related task), you can't trust programmers who are entirely into dynamic languages, no matter what they say about their favorite ones.

That's why I recommended it as a good choice for web programming.

What other languages do you master as much as PHP?

I worked with Perl before PHP but never grew to like it (it's not consistent enough and not exactly easy on the eyes, either).

I do understand the tradeoff between efficiency and power, and PHP hits that sweet spot just right for web programming.

Worrying _too_ much about language efficiency when it comes to web programming is clearly a case of premature optimisation. PHP is efficient enough (probably Perl and Python as well, maybe not Ruby ?).

Well, sometimes it's a question of a choice between upgrading your hardware and optimizing your code (or even re-writing in some other language). Which is cheaper? I'm into it right now, haven't decided yet ;)