[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [GNUnet-developers] Re: amortizable hashcash paper

From: Tom Barnes-Lawrence
Subject: Re: [GNUnet-developers] Re: amortizable hashcash paper
Date: Fri, 9 May 2003 01:40:01 +0100
User-agent: Mutt/1.3.28i

On Thu, May 08, 2003 at 04:29:56PM +0300, Igor Wronsky wrote:
> Perhaps I'm missing something, but the whole discussion below
> reminds me of giving in to the tyranny of democracy, i.e.
  I too am a bit unsure what to make of the voting idea. I was
catching people talking about it, but as I've been drowning in email
(I was on another mailing list that recently went into overdrive.
I've now unsubscribed from that one) I hadn't been up to speed on

  I tracked down Christian's initial proposal for the scheme, and
what first sprang to mind was basically "Why?"

-If we're talking about votes applying to _signed_ content,
then why bother? I thought that one of the main ideas behind
the namespace system was that people could put some/all of their
content into a hierarchical namespace of their own that others
could easily navigate, and that they could (by whatever means
is finally decided on) hyperlink to other people's namespaces
and content.
 So you find this that and the other, and you say to yourself,
"person A is great, very reliable, person B has pretty good links,
person C seems to be a bit of a mixed bag, person D is a crapflooder,
person E is a spammer, and person F seems to just not be too good."

 And based on your opinions, you would trust or avoid different
people's stuff, and would probably think twice about trusting the
content of someone you've not encountered before (unless someone you
know of vouches for them in whatever way).

-Next: counting the votes of those people who have done _similar_
searches. How exactly do we define similarity? If we're talking
about similar == exactly_the_same_really  then I'd think that'd
be easy enough, but would it be useful? I can imagine a mechanism
for determining whether 2 searches are similar, but requiring
reams of statistics for _all_ searches that have been performed
and who did them. I feel that would be quite accurate but painful.

-Finally: The proposal emphasises that the votes be used to *sort* the
search results. On the one hand, this is quite a good thing that it's
not talking about censorship as such, *but* think about it: Everybody
searches for foo. First person checks the first 10 results, finds them
OK, and votes for them. Next person is more likely to get those
results sooner, and vote for them. And the next person. The voting
system could quite easily polarise the popularity of search results
such that certain results would usually only appear nearer the far
end of the lists and stay there.

 Even if someone had the patience to check the whole of a list
with hundreds of results, their vote would be pretty insignificant
next to the people who didn't. And of course, if few people even
get to see a certain search response, it wouldn't be too surprising
if it never got downloaded, and was discarded from the network,
never to be seen again.

> Of course, if you are trying to do app-level spam prevention by
> global voting, that might be fine,
 Actually, yes, that does sound pretty good, if the client was the one
doing the sorting. You could choose to trust certain specific people's
votes more, or switch off the thing altogether, and of course it would
hopefully reduce the chances of that polarisation effect. I think.

> content. But even on the app-level some kind of locality or nonanonymous
> voting and pseudonym ranking might be required, to thwart hostile,
> possibly bigger user groups from taking over.
 Sounds reasonable. I suppose the groups that would develop would really
be defined by people who trust each other's votes.

>And for just content
> sharing, if the namespace scheme is implemented, the need for
> spam prevention lessens, as people naturally learn to trust
> pseudonyms they like, and the pseudonyms control their own "turf",
 Pretty much the way I feel. As long as people can link to others'
namespaces, I can't see voting being very necessary.

> disabling spamming altogether. And finally, I didn't sleep
> much last night. ;)
 Well the night before last I didn't sleep at all (not advisable),
so there!


reply via email to

[Prev in Thread] Current Thread [Next in Thread]