Re: [dev] paste_AT_

From: hiro <>
Date: Fri, 6 Nov 2015 15:25:35 +0100

> The idea of wanting a connection to a central database is what makes
> surveillance effective and in the end will reduce your freedom to noth‐
> ing. So keeping to a more »data packet« approach of spreading informa‐
> tion is something I see as the suckless way of distributing data.

Centralization's prime purpose is not surveillance. It makes some
things more efficient and that might *include* surveillance, but the
opposite, a lack of centralization doesn't prevent surveillance. If
security is your aim then you are taking the wrong approach.

Taking out centralization on one layer (here application layer) also
doesn't mean you suddenly have sealed your universe from mankind.
Unless you're personally moving every electron or photon around on
this planet or even micromanaging every energy state of the universe
you'll at some point give control to some entity that does the same
thing for you at a bigger scale and lower cost.

Unless we get something like quantum communication going you will
depend on an underlying physical layer system.
And even if we get quantum communication you'd still need a similar
system to transfer entangled particles around the world in big

Nonetheless I can see at my university that the military is keen on
decentralized radios. It's being marketed to students as tools for
disaster and emergency management. In some it induces images of
suffering, starving women and children and the idea might be that
charitable donations would come in really fast if the facebook was
properly utilizeable in these areas. I think this is naive, but I
understand aliens will attack us and we'll all burn in world-wide
atomic blasts while engineers with webscale radios in their backpacks
will prevent extinction by seducing the last women alive with their
walkie-talkie. "This is peer-2-peer girl, never played it, huh?"

I can partly see why military wants it. In contrast to your proposal
such technology would be able to serve as the foundation of a
self-sufficient system, your's would be on top of some transport (the
internet i assume).

So yeah, I don't really buy into your Internet of bunkers (IoB).

One very good example to realize that centralization is a common
optimization used even with distributed protocols is Bittorrent. It
wouldn't be as successful if it wasn't used over the internet
(hierarchically centralized routing). Nobody would use it without
tracking sites, torrent catalogues (centralized meta information) or
DHT (this is distributed over many peers but still basically a
centralized database with only one instance. You could split it up
into sealed networks, but then it wouldn't be as useful for bittorrent
as a whole any more.)

The bittorrent experience to the user would also be much slower and
thus slightly less useful if there weren't all those seedboxes in
datacenters or computers with fast uplink (central to clusters of the
mesh/internet in terms of bandwidth*delay as a distance
representation: so with this definition of distance it's still a
centralized bulk data transfer).

Counter-example: Having a fast uplink at home might make some want to
contradict my opinion. But if you look at it's cost, it's commonness
and how much of (for example) bittorrent traffic really goes through
these nodes you might find it's not significant at all.

Trying to finish up the bittorrent topic: the main advantage of
bittorrent over something like a CDN for me today is that
low-"distance" clusters get generated inside non-payed peered networks
automatically and for virtually no cost (the bigger cost for peers
would be to get *out* of a certain ISP's routing bubble). This
basically hacks around the non-dynamic business models of ISPs that
currently result in many overloaded peering links. Watching something
on youtube over Deutsche Telekom is regularly a worse experience for
me as a user than streaming a video via bittorrent on some network
with even less and shittier peerings links. But what does that mean in
context to what I said before: Bittorrent is merely a way of making it
east to work around the *lack* of centrally optimized distribution on
a higher level.

The whole CDN and bittorrent stuff is just about optimizing
fundamental infrastructure for the masses. Of course if you just have
to suit the needs of select privileged few or just the government or
it's enforcement army you can work at higher overhead cause you can
make others pay for the higher costs.

The suckless way of distributing data should be to create less of it
in the first place, and make it so valuable and so compact that the
distribution happens virtually on it's own, like diffusion.
I wonder what, coming from you, would be more valuable: the
information system or the information...
Received on Fri Nov 06 2015 - 15:25:35 CET

This archive was generated by hypermail 2.3.0 : Fri Nov 06 2015 - 15:36:09 CET