RE: IOTA - The good, the Bad (and the Ugly)
Enforced timestamps do by no means prevent transaction order from changing. Transaction propagation is slow and you may receive an 'early' transaction way later than newer ones, in this case you will have to revert the early transaction and do them all again, this is extremely inefficient (chances you have an orphaned block in ethereum are way lower for comparison, because there are simply less blocks).
What I meant with that is that you will have to outsource your PoW to a third party/self-hosted service, at what scale these cloud services will be depends on the demand for them/how many transactions per second you need. Anyway, my point is simply that doing them locally may become more difficult over time once ASICs are introduced (which the hardware team is working on).
Yep I think the PoW structure for IOTA is far more efficient than that of other cryptos, especially because IOTA is mainly for microtransactions, we don't require a huge cost to rewrite the tangle direction.