Especially using OrBot on the phone Tor sometimes feels faster. Since I am not sure on how to really measure this I have been thinking about what the reason for this could be.
I have the following theories:
- Everything goes over a single connection and using fewer connections speeds up things, if the exit node is better in handling this.
- The exit node might have a DNS cache that already knows about the hostname I am looking up. Potentially it even has an HTTP cache, speeding things up further.
- The connection from me through the circuit and eventually to the website I am accessing is a better path than from me directly to the website. This could potentially be due to the connections between the data centers of the nodes (special lines).
- Stuff that Tor or the Tor Browser blocks or requests that don't have responses, because they block Tor.
- OrBot/The Tor Browser take fewer resources.
Has any research been done on that? Are any of those clearly impossible? Is there a good way to measure that? Just doing a small script testing download speed is probably not enough, most websites load a lot of resources, sometimes even third party. Has this probably been measured by OONI? Is there maybe a project related to making Tor faster that analyzes this?