#bitchat

To solve the spam problem I think we need a combo of PoW + WoT.

To send a note, you need to do some proof of work.

Once you have your note out there, if people favorite you, you are added to a web of trust tied to the geo hash. If you teleport, you start over under the new geo hash.

More simply:

- PoW limits spam, but still allows you to post.

- WoT lowers the PoW needed from a Pubkey.

Cc: nostr:npub12rv5lskctqxxs2c8rf2zlzc7xx3qpvzs3w4etgemauy9thegr43sf485vg

Reply to this note

Please Login to reply.

Discussion

disagree. relays that rate limit and ban ips on abuse is all you need for a public chat app.

True. But that’s a coordinated effort. Meaning bitchat will have to rely on some SLA from the relays they are connected to.

Which is fine, but it’s outside of the scope of bitchat itself.

As an individual, starting over with the PoW won’t be an issue. Even with ephemeral keys.

For an attacker it gets difficult.

A VPS can out pow a phone.

100% but it will be costly to run it to spam every geohash

We are racing to solve blocks here, we are creating up hill friction.

Lol, no, its not. I have an FPGA here in my drawer that will smoke the living daylights out of whatever threshold you can come up with that is even beyond what is reasonable to expect from a phone, on whatever novel hashing algo you can come up with.

I don't think you appreciatie the order of magnitude differences at play here. You are bringing a twig to a cannon fight.

As nostr:nprofile1qqsr9cvzwc652r4m83d86ykplrnm9dg5gwdvzzn8ameanlvut35wy3gpz3mhxw309aex2mrp0yhx5c34x5hxxmmd9uqsuamnwvaz7tmwdaejumr0dshszythwden5te0dehhxarj9ekxzmny9u0ljp2l mentioned, you are better off doing traditional rate limiting and hope for the best

Clearly shows my lack of knowledge.

Relay rate limiting is a good solution, but then bitchat has to deal with finding rate limiting ips.

Or bitchat can run their own relay.

Tried to find some nice source with some numbers, but ended up asking GPT-5 (i'd rather not but im past my bed time as it is). It confirmed the numbers i had in my head.

Generally, assuming the same amount of transistors and power consumption performing sha256 hashing (i.e. 'all else being equal').

If a CPU is the baseline, a gpu is about 100x better, an fpga 1000x and an asic somewhere between 100.000x to 1.000.000x.

Roughly, but you get the point.

i do. not a good approach. i understand now.