[stephc_int13]

I am pretty sure that Kodak or Nokia had pretty healthy results when the wind turned for them.

Those kinds of ships are so big that it is very easy for managers, investors and employees to convince themselves that everything is _fine_.

The biggest problem, for Meta, in my opinion, is not the brand, not the business model and not the competition. The problem is much worse because it is internal and probably too late to fix.

Their engineering culture as a whole is a one trick pony.

[miketery]

Can you expand more on the one truck pony? My understanding is that they have some of the best build tools, AB testing infra, and global scalability.

[stephc_int13]

Exactly.

[jpeter]

What is wrong with their engineering culture? They gave us React and PyTorch. Pretty good track record

[stephc_int13]

You could also say that they gave us Zstd and LZ4, but this is simply not true, they hired someone who made it.

There are plenty of brilliant coders, engineers and scientists at Meta, no question about it.

[rmvt]

> Their engineering culture as a whole is a one trick pony.

what do you mean by this? what's their one trick?

[stephc_int13]

The most important component of a technology company is its engineering culture.

This is something that is built early, most of the time with a strong influence from the founders/CTO.

Once it is built and scaled up, it can still change but very slowly.

You can hire extremely talented outsiders (John Carmack) but it won't do much.

A company like Meta is like an overfitted AI model, they can do one thing very well and almost everything else very bad.

Take a look at horizon universe, does it look like state of the art for you?