1. Could you explain in layman's terms, why capping the amount of work per transaction makes double-spend attacks less likely to succeed? It doesn't sound intuitive.
Consider two situations:
1. You need to generate 1 block with 10 zeros in front,
2. You need to generate 1024 blocks with 1 zero.
Let T be the time you need in the 1st case, R is the time you need in the 2nd. T, R are random variables, of course. Now, it is true that T and R have the same expectation, but it is *not* true that their distributions are the same. In particular, the variance of T will be much bigger. What is even more important, is the difference in large deviation probabilities.
Assume that you need to complete you task within time (expected time)/10. What would you choose, 1 or 2, to maximize your chances? Well, better choose 1. That's quite intuitive. What is not intuitive, is how different these chances are. In situation 1, you will succeed with probability around 10% or so. However, in situation 2, it will be *very* low. Don't want to calculate, but it will be smth like 0.0000000000001... anyhow, practically zero. That's why, if you want to beat the rest of the network, it's much better to bet on "heavy" tx's, and so we avoid this kind of attacks by putting an upper limit on the own weight.