I think there's a definitional question about compatible that probably bothers people more who came to the hobby late.
For those of us who played back in the days of the BECMI/Advanced D&D split, we knew that one system had AC start at 9 and go down, whereas the other had AC start at 10 and go down. Similarly, different attributes had different bonuses from one system to the other (BECMI didn't have percentile strength, which was a weird system anyway).
But if we chose to use a BECMI module with AD&D characters, or vice-versa, we just accepted those little differences because they were effectively inconsequential. The Moldvay Basic Rulebook "goblin" having slightly different stats from an AD&D Monster Manual "goblin warrior" didn't matter, because the stat blocks were close enough that they didn't even need to be fudged. You don't need to look up the stats for the goblins your PCs encounter in the Shadowdark rulebook, you can just use the stats exactly as they are written in Keep on the Borderlands, because the mathematical variance is insignificant. We're after playability here, not some platonic ideal of perfectly balanced game design.
Which I think is the major point Whizbang Dustyboots has been trying to make about Shadowdark. To a great degree, it's just as compatible with the monsters from both BECMI/AD&D/C&C and every other OSR game that 90% of the on-the-fly conversion is basically: "Change AC to ascending. Done."
Yes, I generally agree. If you look at the HP values for the "Ogre" in this post, they are 19, 30, and 59 (for b/x, shadowdark, and 5e respectively). My only claim here is that 19, 30, and 59 are, in fact, different numbers. Hot take! I did not expect that pointing out that different numbers are different would be a controversial claim!

If and when that variance becomes significant is subjective. But I am reminded that Gygax begins the 1e dmg with a discussion on dice probabilities; quixotic as that may be, the idea is that a DM should be roughly aware of these things while making rulings.