My suspicion is that the "young" developers they speak of (not all of them, obviously) are faster in the same way not writing tests is "faster". You're taking on a lot of debt that you'll have to pay off sooner or later. For most startups, the bad technical decisions younger developers will make won't bite them in the ass until much later - maybe never if they go out of business for not finding product/market fit - and so getting somewhere, anywhere as quickly as possible is perhaps better than having expertise at writing "great" software.
Also, Moore's law helps folks ignorant of Big-O, algorithm design, memory management, indexing, etc. push off needing to know that knowledge until much later - if ever. When you can spin up thousands of Amazon servers for pennies it really "doesn't matter" (to biz people) if you improve your search algorithm memory footprint (or performance) by 100%.
I think it's possible to have both - older developers that know when to take on debt and how to responsibly manage it can be just as fast as the cowboy younger kids, but that's even harder to find.
Not worrying about algorithm performance and memory management means one of two things 1) you are making a prototype and couldnt care less, 2) you are a shit developer.
I'm still in college and the only time I don't give a hoot about either of those things is for class work when I am not peanlized for it. But anything that goes to production has as small of a footprint as possible. But that may be because I know all of my stuff runs out of an ec2-micro and not on a $100/hr sever cluster.
Or 3) You are an excellent developer who understands the cost as well as the benefit behind optimising for performance and memory.
Is the time spent optimising better spent building a new feature (i.e. the money saved through efficiency will be less than the value gained by doing something else)? Is the optimised code going to be less flexible in the face of change? Is the optimised code going to be less comprehensible for future maintenance? And so on...
Folk who spend their time making sure that everything has as small of a footprint as possible are a royal pain in the many situations where "small footprint" is not the biggest cost of the system :-)
Also, Moore's law helps folks ignorant of Big-O, algorithm design, memory management, indexing, etc. push off needing to know that knowledge until much later - if ever. When you can spin up thousands of Amazon servers for pennies it really "doesn't matter" (to biz people) if you improve your search algorithm memory footprint (or performance) by 100%.
I think it's possible to have both - older developers that know when to take on debt and how to responsibly manage it can be just as fast as the cowboy younger kids, but that's even harder to find.