...and why the fuck do their predictions matter in the least?
As always, variety is good, and it is obvious that HTML5/js will be a good fit for many games. Many others will still require *at least* flash, silverlight (silverlight 5 will integrate XNA and may work on OS X, very interesting for a game dev), or even C/C++.
The last question is why does a new technology always seems to imply that alternatives will automatically shrink? The world is not a zero sum game, and we constantly *expand* our horizons...