You need lots of nodes to make it secure, performance is low even by VPN standards, and it’s dangerous to run a node. Unless you have some likely way to change one of those it’s going to be really hard to get volunteers – and payment is worse because most customers will expect better service.
The more nodes you have participating the more secure an onion system tends to be. Since the Tor network can carry most kinds of traffic, the motivation to avoid a fork is strong.
> The more nodes you have participating the more secure an onion system tends to be.
Tor isn't very large as it is, and (I would guess) it's the largest. If another onion routing network didn't grow the audience, you would have two even smaller networks.
> the Tor network can carry most kinds of traffic
Isn't Tor limited to routing TCP? That would rule out QUIC, for example.
Why haven't there been more onion routing projects. (Maybe there have been and I am just not aware.)
Perhaps the same reason(s) we never saw widespread adoption of remote proxies, despite their usefulness in many situations.
Although in some respects onion routing seems quite an improvement over "simple" proxies.