> This is a false dichotomy. OP can be wrong with their new prediction too.
It's not "I was wrong, therefore CRDTs are the future" but "CRDTs are the future, therefore I was wrong." As you say, this new prediction, too, can be wrong, but I don't see where any sort of false dichotomy comes into it.
You're right that OP doesn't directly make that correlation, but I deduced it from his insistence on that one thing can either be "the future" or not. That kind of thinking usually stems from a dualistic perspective. I wanted to emphasize that. I find such extreme stances generally unhelpful and likely to turn out wrong again.
We really have three OK-ish options for rich collaboration:
* CRDT's
* OT
* Maybe differential synchronization
If a key OT partisan now backs CRDT, that's a significant development. It means CRDT's arguments have become compelling enough to convince people on the other side.
Now, maybe something new and wonderful will pop up and supplant all of the above...
In CoCalc, I use a fourth approach that I think isn’t in your list. https://blog.cocalc.com/2018/10/11/collaborative-editing.htm...
I don’t claim it is better than any other approach, except it is easy to understand. Also I’m not sure that it isn’t technically an extreme special case of OT.
> That kind of thinking usually stems from a dualistic perspective. I wanted to emphasize that. I find such extreme stances generally unhelpful and likely to turn out wrong again.
Author here. Bold stances make for better reading. Its fun to stake your reputation on a perspective you hold. Being bold and sometimes wrong is better on most metrics than never saying anything at all. The best way to optimize for "least likely to be wrong" is to never be noticed, or do anything notable. That sounds like a waste of a life to me.
I'm not optimizing for "least wrong". I'm optimizing for "most movement forward".
If I took your advice, perhaps I could have titled my article "CRDTs perform a bit better than I thought. I have a hypothesis that the choice-gradient should shift by a moderate amount toward CRDTs in situations where both OT and CRDTs would be appropriate." Yikes! I believe that if I said that, my article wouldn't have hit the front page at all. Let alone multiple times.
You may have preferred that article had you read it (though I roll to doubt even that). But I suspect you wouldn't have had that chance, because you would never have found it.
Of course reality is complicated. In this case, there is probably room for both OT and CRDT based systems. (Or even, CRDT based systems which provide an OT frontend). But making a new possibility for the future depends on a communicated vision. You can't communicate that vision if you bore everyone to death whenever you start talking.
It's not "I was wrong, therefore CRDTs are the future" but "CRDTs are the future, therefore I was wrong." As you say, this new prediction, too, can be wrong, but I don't see where any sort of false dichotomy comes into it.