It's not meant to be a scientifically precise number and I used in a half serious way (hence "something like"), but I wouldn't be surprised if it were true. I

It's not meant to be a scientifically precise number and I used in a half serious way (hence "something like"), but I wouldn't be surprised if it were true. I should clarify: I'd have no issue working at start-up that used PHP for front-end web development, as long as they used a more appropriate language for backend/middleware. This is the approach taken by Yahoo and Facebook, although it's not one I'd take personally. I suspect most Facebookers would agree: there were several attempts to switch, but carrying them through would have meant halting site development; most wish Facebook chose a different language from the start. That's called technical debt and they're choosing to keep paying interest on it (writing a PHP to C++ translator) for the time being for very solid reasons (OTOH, PHP performance wasn't an issue as much at Yahoo as they made much heavier use of custom C/C++ extensions). In another way, PHP had a positive effect on Facebook: as it's a limited language, there's much greater motivation to create a distributed and polyglot architecture e.g., breaking the back and middle layers out into separate services and using Thrift RPC, as opposed to either keeping the architecture monolithic (creating a scalability bottleneck) or sending raw C structs or serialized Java Objects over the wire (assuming the network is homogenous is one of the fallacies of distributed computing). Sometimes worse is better. As for myself, I'm neither good at nor interested in web development (although I have tons of respect for great front-end people), which make me useless if 95% of a ten person start-up's engineering work consisted of web development. If the start-up was using PHP exclusively (without any custom C extensions) for tasks beyond web development (e.g., building the backend for what is essentially a recommender system), it would be a strong indicator of a technical team being ignorant and unaware of it (e.g., not realizing that what they're trying can be done elegantly through singular value matrix decomposition, something incredibly painful if not impossible to do purely in PHP) or being micro-managed. If you read Paul Graham, an advantage of early stage start-ups over bigger companies is greater freedom in technology choice, joining a start-up that restricts you to what you think is decidedly the wrong tool for the job is counter intuitive.