This thread made me register--hi! Quick background: played and DMed 1e back in the day (learned on OD&D in 1978!), kept extensive homebrew notes, am organizing them into the 1e Refit: clear rules & mechanics but maintaining a 1e feel.
One conclusion I have drawn from working on the Refit is to give the players choices and tradeoffs, and let those choices stand. For example, in the case of rolling for abilities and class ability minimums, I settled on a method that ultimately will let you play the class you want, but you will need to allow some risk to get better ability scores.
Adding a point buy to 1e is easy. Mine is a bit richer than 5e: 11-16 placed as desired. As has been noted, ability bonuses don't kick in until 15 or so in 1e, but that's a feature not a bug, part of that edition's feel. Now compare this to the "crazy" UA method of rolling 9d6 to 4d6 and dropping all but the highest three. This is less crazy than you might think if you do the (admittedly somewhat complex) math. For each of 7d6-9d6, the modal roll (single most likely) is 16, but the average roll is less than the mode in each case (14.9 for 7d6). In fact, the rounded averages for this method (9d6 to 4d6) are 16, 15, 15, 14, 14, 12. Higher, but not crazy higher, and I assure you (after lots of simulations) that this method can produce disappointing scores. Regardless of method, I then use a more or less 1 for 1 raising needed stats by lowering others (it's a bit more involved, but not much) to qualify for the desired class. The result is a mix of both higher and lower stats. I allow two other methods, one with less uncertainty than the UA method (DMG Method III but 5 rolls instead of 6) and one with much higher uncertainty and the possibility of some very high (and low!) stats.
Will this appeal to modern players? Given that one can always point buy, the uncertain options are add-ons and so should be neutral at worst. But I think a lot of 1e can be tweaked to give this result: give the players both choice and tradeoffs, not just choice, and I think you get a range of more interesting outcomes.