That Viaweb was competitive because of blub, does not imply that blub is a necessary condition for any company to be competitive, or that not-blub would impede Facebook from becoming successful.

For example, if fast new features are only possible with blub, and new features are not as essential to Facebook's success as they were to Viaweb's, Facebook's use of not-blub (or perhaps, non-use of blub) would not materially impact Facebook's success.

One way in which this might happen is: Viaweb is B2B, and its customers value new features; whilst Facebook is B2C and a social network i.e. the value is in the number of people on the platform, not its new features.

Another explanation might actually just be that there was no high enough level language in Viaweb's days to compete with Lispers, and plenty in the PHP era. To an extent we witness this today with the myriad frameworks available which in theory eliminate any edge for fast prototyping except in safety-critical or performance-critical applications.

[1] https://facebook.github.io/react/blog/2016/01/08/A-implies-B...

I agree with pg that lisp was probably the best tool he had available at the time for his specific problem, ViaWeb was a system that generated systems, sounds like a lispy problem. And it also worth of note that not all of it was in lisp, so they mostly used languages that were good at each specific problem.

If your problem involves shipping quickly and iterating, php, the platform is very powerful, especially at the time facebook started.

So PHP may have been facebook's lisp, remember, rails and django were released on 2005, facebook started on 2004. So was php a bad choice? I don't think so. It might have been the best choice at the time.

What you should take from pg's essays on lisp is not that lisp is the holy grail of programming, but that picking the best tool for the problem may play a big part on your success.

Think about it, erlang is just perfect for whatsapp, so is lisp for ViaWeb's metaprogramming needs, and I think facebook didn't really have any big, complex problems at first, so picking the simplest, quickest platform to develop on was probably the best bet they could have made.

So, yes, PHP was more powerful, or better put, more empowering than lisp for facebook's specific problem.

Friendster infamously was left in the dust because its VC laden board was more interested in striking cool deals than allocating the resources necessary to make the site fast enough, so that e.g. simply logging in didn't take a minute or more.

They, not a founder, were calling the shots, and while I don't know the Myspace story, I note they were bought by the monster sized News Corporation a year and a half after Facebook was founded, and during the period where their relative market positions flipped, so blame it on Rupert Murdoch if you want ^_^, but the chances they had nimble management after the acquisition are very very low.

Getting back to Friendster, as a DEC published history of microcomputer companies in the '70s put it, the ones who survived at minimum did an adequate job of everything essential, from all the different bits of hardware (no one ever bought DEC for the quality/price-performance of their disk drives), to documentation, to support etc. So no matter how powerful your language is, if you don't, for example, run it on enough servers, or architect your system so that's even possible, you're not going to succeed.

From http://www.businessinsider.com/how-moskovitz-became-facebook...

>Dustin was like, 'Hey, I want to help out. I want to help you do the expansion.'

>Zuckerberg told Moskovitz, "That's pretty cool. But you don't know any PHP."

>"That weekend he went home, bought the book 'PERL for Dummies,' came back, and was like, 'OK, I'm ready to go,'" Zuck said, and told him, "The site's written in PHP, not PERL, dude."