There is no magic, and the article makes that clear. It might win on the constant because of the potential parallelism, but even that's not clear. But if it d

There is no magic, and the article makes that clear. It might win on the constant because of the potential parallelism, but even that's not clear. But if it did win on the constant, the amount of potential parallelism is huge! It's just that we never seem to be able to take advantage of it "properly." But the thing is that this idea keeps getting re-discovered, with people getting excited all over again. This article is a clear debunking of the hype, and has value in being able to point people at it and say "Look, it's been done, it doesn't work, and here's the explanation. If you've done something really new, then you need to make it clear why it's different from this 20 year old work." Having said all that, I'm not sure what it is you're claiming doesn't make logical sense.