Hector Martin

I am complaining about real world implementation of those standards by products I can buy.

USB's biggest crap-factor is that it's cheap. BOM cost correlates to profit, and profit correlates with the ability to R&D. The low cost of USB devices leads to under-engineered, under-tested crappy implementations of (relatively) complex USB protocols/products.

I.e. for every good {USB hub, video capture, UART, mass storage, ...} chipset, a cheaper&more crappy version will eventually take over the market. It's a race to the bottom. The good stuff still exists, but not in a meaningful quantity anymore.

USB (1/2) is designed to be as hostile as possible to any audio system, being impossible to galvanically isolate without massive expense, and introducing common mode ground currents at precisely the most sensitive part of the human audible spectrum. Fuck USB.

No joke USB is single-handedly responsible for a massive fraction of horrible interference and background noise in amateur and professional audio productions alike. Any time you hear a 1kHz whine in the background, that's USB.

USB3 fixed this problem, but neglected to specify USB2 to USB3 transaction translation, which is a massive lost opportunity to fix the galvanic isolation problem for USB2 cheaply, and nobody puts USB3 on low channel count audio interfaces, so we're still fucked.

So yeah, no, sorry, the USB design isn't "quite good", it's complete shit. It's not differential (not fully), it's not extensible/virtualizable/tunnelable (not properly), utilization sucks, rate control sucks, speed is data-dependent, it's overcomplicated, etc etc etc.