See, when you said
> All of these have been traced back to Proto-Indo-European and have different origins:
that could be taken to mean that the PIE terms have different roots or no roots at all.
"ob" and "eu" go back to * h₁epi- and * h₁es- ... and evidently obs-, obc- is frequent so there's the assumption that * h₁epis > * h₁epi- must (might) have been a thing; while you already noted that loss of b before a consonant is a thing, so that * h₁ebs- > * h₁es- were not too unlikely either. Even closer now, I'll stop there.
But to think that terms as basic and simple as "yes" would be innovations instead of variations of varieties that existed for millennia, warrants a lot of doubt. Just pointing at a few made up roots doesn't impress anyone, really.
Historical linguistics is clinging to the "single mother" hypothesis. It's a useful abstraction, but as programmers we should know better. Simply speaking, it appears to me that the sheer amount of material to work with is too intimidating to researchers (manually) to even attempt to favor the assumption of chaos. So if it can't be supported by historic artifacts it's labeled mere coincidence. And reliable artifacts are sparse. So everything is uncertain. OK, I don't actually know much about the archeology, but as far as I know, Celtic and Germanic is generally not well preserved as text. I agree, I guess, the threshold to validate a claim is rather high. And I do err at times, quite trivially. Perhaps that's why I present my argument as a joke. But the threshold counts all the same against falsification. You can't just arm chair a "nope" like that. That was really inconsiderate.
Instead of roots, take a step back and consider basic semantics. What's closer to yes: "true, good, to be, " or "that, there"?
And for a region, that happens to lie in the west of roma, what's more likely to be it's name "west" or "the country in which they say oc"?