Hector Martin

Look, if you're complaining that not all USB 3.2 devices can do 20Gbps, I hope you also complain that not all 802.11ac devices can do 3.6Gbps, or that not all HDMI 1.4 devices can do 4Kp24, or that not all LTE networks will give you 1Gbps down.

This is how standards work. The standard defines the maximum capabilities, which not all devices need to implement. The standard *may* define some minimum capabilities for *certain* devices, like USB 2.0 requires hubs and hosts to support 480mbps (but not devices).

I feel like the complaint is primarily about hosts, and the perception that a company could take a computer equipped with a 3.1 chipset and with zero modifications slap a 3.2 sticker on it, even though it could never achieve more than the 3.1 it was allotted.

Is anyone actually doing that? I've looked at the 3.2 spec and I can't find a hard requirement for hosts, though it *does* say that *hubs* have to support all speeds. USB 2.0 *did* require hosts to support 480mbps. This could be an oversight in the spec.

I honestly have no idea! But apparently, at the least, some people think they are, or could. At least, that was the sense that I got from the Ars article.

The author of the article just had a back-and-forth with me on twitter but he didn't produce any examples of *hosts*, just of devices, so I dunno...

Fair enough; I don’t even know. Juuuuust seems like an unnecessarily confusing mess, probably moreso than other versioned specs. But I don’t really know the realities of the marketplace. So *shrug*.