Most people eat large quantities of food all at once. Stop half way and wait 20 minutes and you might not be hungry anymore. If you eat too quickly then you don't know at what point you'd eaten enough to satisfy your hunger.
How is it possible to "generalize" from one data point? The normal method of generalizing is finding a similarity shared across many data points, and proposing it applies more widely.
Perhaps one could generalize badly with 2 data points, but isn't it just impossible with 1, and therefore he must actually be doing something else?
The way he "compared" them was to say the list of criterion applies to both of them, and that it clearly shouldn't apply to the other things. What's wrong with that?
If it has certain merits, and certain demerits, then as long as you carefully select which things to read normally, and which to read with it, it will help you.
If you want to skim or go back, don't use it. But there are sometimes things you read where you don't need that.
Guy is amazed you can get 28th on google in no time.
You can get first on Google for obscure terms just by commenting on YC. The site you are on matters, the issue is only whether knol gets special treatment or matters in the normal way.
And you should read more carefully. I did not suggest the thing you say is a bad idea.