it just never occurred to me (it definitely should have) that that would result in potential "lock-in".
I don't buy into this, there's no lock-in to me. The user will fill his needs whatever the mean, either by using something provided (almost) freely by the SoC, or by buying additional HW.
So what can be argued is: will the few things added to every boards to provide the SoC specific functionalities be compensated by the things not needed for those who will get it anyways. That's a trade-of, additional price will also be a factor.
But providing additional interface on the far end of the card is no philosophical problem to me. Especially for ubiquitous things like ethernet, I can't think of this being a locking-in feature. After all these boards are designed to have a long lifespan, so the need to upgrade is not as important.
And if (and only if) a future upgrade does not provide it, then you always be able to switch to an USB one, if (and only if) you still need it...
i always thought, y'know, things like HDMI would be ubiquitous / common enough that you would always be able to find an upgraded computer card with USB-OTG and HDMI: the impact of specialisation at the *user-facing* end never hit my tiny brain :) huh.
In my computer life I've seen 2 wired net techs (BNC & eth), whereas I've used 5 for display (VGA, BNC, DVI, HDMI, DP). I'm OK with USB because of its backward compatibility (I'm still not sure it will stay that way) So I'm not sure trying too hard to guess the future is mandated here, just being pragmatic. Which I can (and have) be(en) convinced EOMA68 is...
But I'm convinced that providing as much of the SoC functionalities as pragmatically possible is better. Either by enabling more widespread adoption, or by avoiding the need for USB-based addons.