early questions are not worth more than the rest

Yes indeed, it is a myth. All the questions have equal importance. More careless mistakes in the beginning will not help you assess your true potential, so the point is never lose a question which you could have got right. It doesn't mean that you should be spending extra time for those first few problems, instead improve on reducing the careless mistakes and spend an average of 2 mins for a single problem. The less careless mistakes, the sooner you reach your maximum level questions. Don't try to assess how hard the question is, since it is an adaptive algorithm you might not know what level questions are being thrown to you.

So it's actually about how soon you can reach (say) 700+ level questions and how many of them can you answer correctly, instead of the test trying to zero in on your level?

I took the Knewton courses, and in one of their earlier lectures they talk about this. The founder said that in the PAST, the algorithm used placed a lot of weight on the earlier questions, and a lot of people figured this out. Some students would increase their score +100 pts by taking advantage of this. The makers later on changed the algorithm so that test takers would be severely penalized if they did not finish the exam. So the suggested strategy was to spend an extra few seconds on the first 10 questions if need be.

Last year's Princeton Review's "Cracking the GMAT" said the same thing, and it is undeniable that it seems to be common knowledge that your score swings more during the first 10/15 questions of every section.

Now, is that still true? is the new algorithm so different?

I'm actually confused right now.

I don't quite understand what you are trying to say. But don't waste time gauging the question difficulty, its a complete waste of time. A 600 level scorer will take more time to reach that 700 solve because of difficulty and careless mistakes, while 700 level test taker will be able to move quickly to that level and also stay on that level for a long time. All the GMAT questions can be solved, but what you do in stipulated time determines your score. If you spend more time to increase your actual capability, then once you reach that level it takes half the time to reach the level where you started because you cannot keep on spending more time for every problem and for every additional minute or seconds you spent initially to improve your level, you have much less time for each of the latter problems. So the key is to get good at content knowledge, know your strengths and weakness, guess where you need to and where don't need to and reduce careless mistakes.

I'm actually confused right now.[/quote]

And so am I. Any experts there to clarify this ?