It's not a big mystery. Linus released a primitive kernel that worked, at the right time, with the right license, and then diligently kept rolling up contributions and releasing the result.
This is all true and important, but I think it's leaves out the really important part. Linus has good judgement in two critical areas; policy and people.
You and many others are correct about the timing, license and Linus's willingness to accept contributions without preconditions, and that part of it accounts for the early days. But it could have gone so wrong later and it didn't.
Had RMS been the shot caller Linux would be a curiosity today. People like him, while well intentioned, can't help but strangle babies in cradles in the name of their agenda. The kernel would be on GPL4 or 5 by now and about the only thing you might be able to use it for is a non-profit operation. The RMS mentality would have precluded set-tops, portables (binary blobs, DRM, etc.) the cloud and many other use cases. The best case would have been "for-profit" forks and then decline.
Also, Linus doesn't suffer fools. Over the years there have been contributors that, while possessing some talent, were destructive to the process. Linus has reliably kicked them to the curb and kept them from ruining Linux development. It's a simple, unfortunate truth; some people don't play well with others and if they get a foothold in something they ruin it.
These two aspects of Linus, good and firm judgement about policy and people, have ultimately been the most important because failure of either would have killed Linux long ago regardless of the early enthusiasm. That one person embodies the drive, talent and judgement to take Linux this far while protecting it from the bad ideas and fools that prevail is a small miracle.