Is Poal the underdog? | Poal: Say what you want.

(post is archived)

You are currently inside a comment thread.

Click here to see all the comments (57).

[–] • 0 pt

That's a fair criticism, and I can see where you're coming from.

My thinking has been that if it's ridiculously easy to bypass (like hiding downvotes via CSS in reddit as an example) it simply disadvantages and potentially deceives novice users.

But additionally, to even get anywhere close to a traditional site's alt management I'd have to compromise user privacy and track ip addresses to some degree which I currently don't do.

If it was only tracking for the purposes of establishing a vaguely unique userbase it wouldn't be so bad, but it still means storing some information that I could potentially be compelled to give up to a government and I see that sort of thing as bad.

But really the main reason I've not explored this at all is because of the plan for notabug to not just be federated, but radically p2p to a point where I won't even see individual traffic myself.

It's not just a matter of it being simple to compromise any approach at doing this; it's also a pretty difficult to do at all without compromising privacy when you assume the p2p model.

Maybe some sort of highly salted hash of signup ip address attached to account could be a first step at giving non-reversible, deterministic ip based identifiers to attach to accounts. (but this is kinda scary too, if the salt leaks reversing the hashes isn't so hard given how small the ip address space is)

That would get you to a roughly voat level of protection (1 vote per IP) but where alts are publicly identifiable (this is a requirement if you want to identify alts at all in a distributed way even if just for voting purposes)

This would still require account signup going through a trusted entity. Without that it's impossible to do anything like this at all in a distributed environment without everyone knowing everyone's IPs associated with account.

If you stop at centralized trusted account verification then votes have to be publicly attributed to individual users to be able to do anything useful with that info.

So you can do a similar trusted authoritative party to accept private votes and provide public scores like a traditional site.

That's what's necessary to get anywhere close to the level of protection a traditional reddit clone provides.

I think building something like this out would be a good thing, and I aim to support it if I don't build it out myself.

The general approach of notabug is to build a decentralized content environment that you can centralize services on top of.

But this approach means that the features requiring central services will likely be the last to be built out.

The thing that I'm actually blocked on right now is waiting on GUN to fix an issue with crypto logins in node.js

I will be using this to authoritatively count votes for users to save bandwidth, signed by a key representing notabug.io

This approach allows for the decentralized distribution of centralized services (like what will be necessary to get closer to traditional clones for voting) and this will allow for models near identical to voat, reddit, poal etc... to reach wide distribution without huge costs for any single party.

That's a fair criticism, and I can see where you're coming from. My thinking has been that if it's ridiculously easy to bypass (like hiding downvotes via CSS in reddit as an example) it simply disadvantages and potentially deceives novice users. But additionally, to even get anywhere close to a traditional site's alt management I'd have to compromise user privacy and track ip addresses to some degree which I currently don't do. If it was only tracking for the purposes of establishing a vaguely unique userbase it wouldn't be so bad, but it still means storing some information that I could potentially be compelled to give up to a government and I see that sort of thing as bad. But really the main reason I've not explored this at all is because of the plan for notabug to not just be federated, but radically p2p to a point where I won't even see individual traffic myself. It's not just a matter of it being simple to compromise any approach at doing this; it's also a pretty difficult to do at all without compromising privacy when you assume the p2p model. Maybe some sort of highly salted hash of signup ip address attached to account could be a first step at giving non-reversible, deterministic ip based identifiers to attach to accounts. (but this is kinda scary too, if the salt leaks reversing the hashes isn't so hard given how small the ip address space is) That would get you to a roughly voat level of protection (1 vote per IP) but where alts are publicly identifiable (this is a requirement if you want to identify alts at all in a distributed way even if just for voting purposes) This would still require account signup going through a trusted entity. Without that it's impossible to do anything like this at all in a distributed environment without everyone knowing everyone's IPs associated with account. If you stop at centralized trusted account verification then votes have to be publicly attributed to individual users to be able to do anything useful with that info. So you can do a similar trusted authoritative party to accept private votes and provide public scores like a traditional site. That's what's necessary to get anywhere close to the level of protection a traditional reddit clone provides. I think building something like this out would be a good thing, and I aim to support it if I don't build it out myself. The general approach of notabug is to build a decentralized content environment that you can centralize services on top of. But this approach means that the features requiring central services will likely be the last to be built out. The thing that I'm actually blocked on right now is waiting on GUN to fix an issue with crypto logins in node.js I will be using this to authoritatively count votes for users to save bandwidth, signed by a key representing notabug.io This approach allows for the decentralized distribution of centralized services (like what will be necessary to get closer to traditional clones for voting) and this will allow for models near identical to voat, reddit, poal etc... to reach wide distribution without huge costs for any single party.

parent
link