Late reply:
Game calculation is also made via email. Players receive calculation tasks they have to solve and send back. When everything is calculated the game tick ends.
That would be a vilanious scheme of Toady to popularise math.
interesting. What happens when there is an error?
Crowdfortressing. Multiple people are sent the same calculations (e.g. whether cell [34,76,23] changes temperature and, if so, by how much) and their answers are compared[1]. Outliers are rejected (personal ratings downgrade) and concensus - if any - is reached (raising other personal ratings), or else further requests are made of more and more subscribers until one is. It would be distracting and inclined to generate bias to pass along the contexts, of course, so any value identities are randomised/normalised to entirely nonrelevant labels.
You have to the in the top 20% of those so rated to be the
player in the game, to try to ensure at least five people are supporting (or trying to wreck) every tick of every actual game in progress. Though as no-one bothered to crowdcalculate that estimate it was not actually discovered to be a very very wrong assumption regarding those supply and demand quantities.
Can't remember if I/someone else suggested this before: Random reshuffling of Z-levels at various intervals. And a relatively high chance of total reversal of vertical order.
[1] This function is also crowdsourced, with further checked and balanced.