IQNets: user classes, network locality, quid pro quo, torrents at core, on gaming reputation

[Zenaan Harkness]

On Thu, Oct 24, 2019 at 04:46:44PM -0300, Punk - Stasi 2.0 wrote:
> On Thu, 24 Oct 2019 22:49:48 +1100
> Zenaan Harkness <zen at freedbms.net> wrote:
> 
> 
> > 
> > Classes of nodes (from a node's perspective):
> > 
> >   - direct friend nodes with a meat space asserted trust level, which
> >     may be anywhere on the trust axis - both above and below zero
> 
> 	you can have the option to mark nodes as 'trusted' or 'friends'
> 	but I don't think the network can or should be built assuming
> 	that there are 'trusted' nodes.  So, no not a 'friend to friend'
> 	network. 

Ack.

You don't have to hop via friends, either initially, or at all.

Config.


> > Storage metrics, just as with network metrics, are also maintained in
> > order to establish fairness - quid pro quo.
> 
> 	storage of what

Something that may or may not be encrypted, as in, I hand out a bunch
of say 100KiB "storage undertaking" tokens - if they're on my mobile
phone, compared with a permanently connected computer, that's a
connectivity metric, and of course permanently connected tokens are
more valuable.

This is similar to Freenet, with a strong emphasis on quid pro quo,
and "anonymous" or at least intended to be.

If I want to upload a new 'tube, and I want it to be accessible at
high speed in this overlay net, I need it split into suitably small
chunks, and I need each chunk suitably replicated (so that if I
replicate each chuck say 3 times, they're not all replicated on
mobile phones normally only on at the same time - thus the need to
store storage metrics.

Desperate to upload my 'tube, I have a few options

  - find a free hosting service

  - host it myself

  - ask friends to host it

  - trade my own storage anonymized storage undertakings with others,
    and then make use of some randomized combination


Having 'friends' is simply a way to use something that exists (actual
friends, if I have any), to increase the difficulty of people gaming
the metrics e.g. of a utilized storage undertaking.  "Crowdsourced"
metrics are something bittorrent does a bit of in this way...


> > The internet as it was conceived has core design principles which are
> > desirable for robustness and censorship resistance, such as peer to
> > peer communications and flexible routing.
> 
> 	the internet is a global spying machine with close ties to the
> 	worst scum on the planet, the US military and accomplices. So I
> 	highly doubt that 'censorship resistance' has fuck to do with any
> 	'core desing principle'
> 
> 	Anyway, that's off topic. I strongly suggest you focus on the
> 	main issue at hand : how to use cover traffic to make something
> 	that isn't as shitty as tor. 
> 
> 
> > In other words, the goal is high speed and relatively efficient
> > network utilization 
> 
> 	The goal is a safe space for criminals, both imaginary(govt
> 	designated) and real criminals. 
> 
> 	And in order to make something that's better than 'low latency'
> 	networks speed and efficiency proably have to be sacrificed or
> 	traded off.

Such things are a trade off.


> > I.e. "large file distribution" is a first class citizen in the IQNets
> > network.  With a shout out to cypherpunk Juan for suggesting the idea.
> 	
> 	the idea is hardly mine at all - and while I think that yes file
> 	sharing as cover traffic makes sense, it won't necessariyl be
> 	high speed file sharing for fucktards who want to download the
> 	latest disney movie in ultra super highest quality. 

Such must fall back to "zero hops, dl direct from Disneyflix". Having
hops go first via friends may tend to incentivize such sane "fall
backs".