Can't the mixing only be done if enough people want to transact at the same time? If you enforce a higher minimum of mixing rounds, could that lead to waiting times or how does it work?
Hey, essentially, the mixins automatically are routed through existing wallets - they don't need to necessarily be running a transaction at the same time.
However, you are correct, enforcing a higher mixin results in longer waiting times as you need to wait for the funds to be routed through the increased number of 'proxy' wallets.
Additionally, a higher mixin size can result in higher transaction costs.
We picked a ring-size of 10 for a couple of reasons. Firstly, there is no evidence to suggest that a ring-mixin size of 12 is statistically safer than a size of 10. Some evidence was put forward to suggest that there would be no security benefit incurred above a size of 4 (which is what Monero originally used, you can read about their original decision in this Monero Research Lab paper:
https://lab.getmonero.org/pubs/MRL-0004.pdf).
However, a paper was put forward suggesting that this default ring-size was linkable (
https://arxiv.org/pdf/1704.04299.pdf). Monero labs published an official response addressing this paper.
In a nutshell, lots of discussion happened within community, and many believe that 10 is the ideal minimum for preserving security (
https://github.com/monero-project/monero/issues/1673).
We countered the downside of the increase in transaction time by reducing the mean time between blocks from 240s to 60 seconds (which is still considered safe, even though it can increase the number of orphan blocks created, causing the blockchain to become a little more bloated). This substantially increased transaction speed, even when using a mixin size of 10. Additionally, we (@SadBatman) implemented some compression on the blockchain to counter the effects of transmitting/ downloading the slight additional bloat this incurred.
Finally, some other things were done to reduce transaction costs such that the cost incurred from this ring-size was negated. I wasn't directly involved with reducing the transaction costs, so I can't speak in great detail about it. But, I believe we actually reduced the transaction costs by significantly more than just the amount required to counter the incurred ring-size cost.