There will eventually be a bunch of bins of the Sky Lake quad core CPU. That's how Intel handles it every generation. See, for example, what they did with Haswell quad cores:http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&IsNodeId=1&N=100007671 4814 600436886 600030236
And that's just desktops. Among servers, there's this:http://ark.intel.com/products/family/78581/Intel-Xeon-Processor-E3-v3-Family#@All
And a whole bunch of laptop bins, too. All of those except for the Xeon E3 1220L are the same chip, possibly up to a respin or two.
For Sky Lake, Intel started with some high bin desktop quad cores, with the Core i5-6600K and the Core i7-6700K. But while the former has been plentiful since launch, the latter is basically nowhere to be found. That presumably means that very, very few of the dies can meet the specs for the latter.
Sometimes you paper launch a product because you know that your competitor is way ahead and you've got something that will narrow the gap considerably or perhaps even take the lead. See, for example, the Radeon R9 Fury X, or to go back a little further, the GeForce GTX 680. You do that because you want people to wait and buy your product rather than buying a competitor because you're not remotely competitive. That's obviously not Intel's situation today, where AMD is so uncompetitive that for their last two generations, they haven't even bothered releasing server/high-end desktop parts at all.
So why a paper launch? My theory is that Intel is trying to maintain the perception that their CPUs are still improving and so it's still worth upgrading. Sandy Bridge was a great CPU in its day, so much so that it could be clocked far away from the chip's real limits and still handily beat older generations. Ivy Bridge was a little better but not much, but Intel was able to maintain the perception of progress by setting a higher stock clock speed so that stock Ivy Bridge beat stock Sandy Bridge by a decent margin, even if overclocking both basically eliminated that gap. Going from Ivy Bridge to Haswell was similar.
But then 14 nm got delayed. I mean really delayed, more so than 22 nm. Rather than having nothing new, Intel came out with Devil's Canyon, a refresh of Haswell. Basically, take the same old Haswell chips and factory overclock them. Thus, we got the Core i5-4690K and the Core i7-4790K. With the latter especially, instead of having to overclock yourself and hope you were lucky with the die you got, Intel would essentially do it for you with a fairly conservative overclock. Even though they had nothing new to offer, higher clock speeds maintained the appearance of progress.
Then came Broadwell, which was unambiguously slower than Haswell, on top of being extremely delayed. But with a large Crystallwell cache, it was a laptop chip from the start and not terribly interesting in desktops. But Sky Lake could very easily have also been slower than Devil's Canyon, and that would have been hugely embarrassing to Intel.
The solution? Release a top bin of Sky Lake that is clocked just high enough to barely edge out a Core i7-4790K. Hardly any dies can meet that bin? Oh well, you don't sell many of them. But reviewers for launch-day reviews don't know that, so they say hey, Sky Lake is great. And then hopefully people go to buy one and if the flagship is out of stock, buy some lower bin instead.
Launch day shenanigans to try to bias reviews from gullible reviewers have been around for quite a while, of course. One of the more notorious was the EVGA GeForce GTX 460 FTW. Recall that Fermi was a disaster, so AMD won by miles in whatever efficiency metrics you care about and could set their lineup however they wanted. When the Radeon HD 6870 was about to launch, Nvidia picked out a handful of the best dies for a GTX 460, overclocked them to ridiculous margins, gave them to EVGA to make a super overclocked version of the card, and sent out samples to reviewers with the hint that they ought to compare a stock Radeon HD 6870 to an extremely overclocked GTX 460.
Some less than reputable reviewers bit and basically recommended that people buy the EVGA GeForce GTX 460 FTW rather than a Radeon HD 6870. Never mind that the former was a publicity stunt that sold out within days and was very quickly discontinued, that more than a few of them weren't stable at their factory overclocked "stock" speed, or that you could have overclocked a 6870, too.
Even so, what does it say about Intel that they're so far ahead of AMD and still feel the need to pull such deceptive review shenanigans? Perhaps that their marketing people feel a need to do something to justify their existence?