Maybe I should tone down my criticism of Chipmixer for this, and start more actively pushing users to upgrade so they can use Segwit. Sadly, I still see far too many people using 1xxx addressesand thus I infer, non-Segwit wallets (in many cases, stupid exchange wallets). Chipmixer may damage their anonymity set if they exclude users who have not upgraded. Whereas per the above, the anonymity set is all-important for such a service. I think they seem sufficiently clueful to do Segwitif they could, when they can without hurting the service in other ways. Unlike some other parties who have spent the past two-plus years deliberately dragging their feet on this, Chipmixer may have a valid reason to wait.
Any thoughts on how this problem could be solved?
I've always assumed it was due to the slow network adoption of Segwit. Until a few months ago, Segwit transactions comprised significantly less than 50% of transactions on the network. Moving all of Chipmixer's activity to Segwit would have therefore compromised its anonymity set. Best to use the most common form of Bitcoin address, right?
Now that Segwit adoption is hovering in the 50-60% range, the transition is more justifiable. (To be fair, I'm not sure about the proportion of bech32 vs. wrapped P2SH usage, though.)