[Arnt]

It seems easier at first, in much the same way that an O(n) algorithm may seem faster than O(1) at first.

You'd have to move very approximately one tanker full of water for each person every week, for the rest of that person's life. You can think of much more effective ways to do it than with tankers, but can you think of any that are more effective than moving the person to the water?

[richk449]

I’m not sure why “effective” is the metric you are using. I’m saying it is easier to dig a canal that can provide water for tens of millions of people than it is to move tens of millions of people to a new state.

[Arnt]

This has been discussed before on HN. Quoting an actual engineer from faulty memory: Boring a tunnel through the mountains is well beyond what's currently doable. Doing it on the surface would require lifting the water a few km up in order to let it flow down on the other side, and would be the largest engineering project ever, expensive to build and equally expensive to operate. Total water use (including farming and industry) may well be well over 1000kg per person and day even if it's restricted because the water is expensive, lifting that weight several km requires a lot of energy and expense.

Sorry about the word effective, it should have been efficient. I used the word because the pumps may be compared to other means of transport in terms of water moved per joule spent.