[washadjeffmad]

Then imagine (magnetic) external contacts, like the Apple Pencil.

Wireless data rates (eg- Wifi6) currently exceed wired by a good margin. The only barriers to fully utilizing that are vendor-manufactured and can be solved with both custom firmware and more open app stores.

[Nextgrid]

> Wireless data rates (eg- Wifi6) currently exceed wired by a good margin

Maybe in lab tests and based on theoretical data rates from datasheets. In practice they don't; most people are in a noisy environment and use shitty consumer-grade equipment. Even if you wanted to pay to make it work, finding good equipment is not easy and often involves trial and error until you find a product that's good enough.

In contrast, even the cheapest Ethernet cable will pretty much always work, and when it doesn't it's very obvious where as Wi-Fi has so much potential for "kinda working" where it looks like it works on most speedtests (because TCP corrects for packet loss) but completely craps out on a real-time application such as a video call (where you can't conceal packet loss).

[washadjeffmad]

I'd hazard that the usage rate of RNDIS on iPhones is rather low, though.

And you may be right for certain contexts, but ubiquity counts for a lot.

"Gosh this bus is making a lot of stops, let me just whip out this USB-C/Lightning to Ethernet adapter and plug into my battery powered switch to access this content on my Raspi file sever" is something I've only said a few times, followed by "why didn't I make this thing ad hoc?" or "why'd I make this service internal only again?".

I do however have Wifi6 APs at home and get better than USB HiSpeed data rates from my 10GbE network. The whole network cost around $500, which is probably near what people spend on replacing crappier all-in-one consumer gear over a similar lifecycle. It seems unrealistic that anyone would be required to spend remotely close to that much to make a reliable video call, though.