Login

Redecentralize

We’ve had enough of digital monopolies and surveillance capitalism. We want an alternative world that works for everyone, just like the original intention of the web and net.

We seek a world of open platforms and protocols with real choices of applications and services for people. We care about privacy, transparency and autonomy. Our tools and organisations should fundamentally be accountable and resilient.

Home

Parent
Adam Ierymenko [LibreList] Re: [redecentralize] Thoughts on decentralization: "I want to believe." 2014-08-04 16:04:39 (4 years 11 mons 15 days 07:16:00 ago)
On Aug 3, 2014, at 1:45 AM, Jörg F. Wittenberger <Joerg.Wittenberger@softeyes.net> wrote:

> Am 02.08.2014 21:47, schrieb Adam Ierymenko:
>> On Aug 1, 2014, at 9:02 PM, Steve Phillips <steve@tryingtobeawesome.com> wrote:
>> 
>>> 3. For a year or so I've had a design for a zero-knowledge server that nonetheless implements partial search/querying functionality for anyone with the key.  Perhaps this could also play some role in the ecosystem.  I'll try to write something up.
>> I've been thinking about that too, but I think it's important to take a step back and think through the problem. I really want to push through the Little Centralization Paper (Tsitsiklis/Xu) a little more.
>> 
>> To me the key thing is this:
>> 
>> Our hypothetical "blind idiot God" must be as minimal as possible.
> 
> I'm with you.
> 
> We've been toying with such an idea for a while too.
> 
> But looking into this "little centralization paper" I'm left puzzled 
> what *function* the centralized thing should provide?

That's what I'm scratching my head about too. Their work is so theoretical it simply doesn't specify *what* it should do, just that it should be there and its presence has an effect on the dynamics of the network.

I'm toying around with some ideas, but it's still cooking.

> My over-all impression so far is, that the paper mostly concerns 
> efficiency and load balancing.  I'm not yet convinced that these are the 
> most important points.  IMHO reliability and simplicity are much more 
> important (as you mentioned in your blog post too).  I view efficiency 
> more like an economic term applicable to central service providers 
> operating services like FB.

Efficiency is really important if we want to push intelligence to the edges, which is what "decentralization" is at least partly about. Mobile makes efficiency *really* important. Anything that requires that a mobile device constantly sling packets is simply off the table, since it would kill battery life and eat up cellular data quotas. That basically eliminates every mesh protocol I know about, every DHT, etc. from consideration for mobile.

>> In Turing-completeness there are shockingly minimal systems that are universal computers: https://en.wikipedia.org/wiki/One_instruction_set_computer
> 
> I'm afraid there needs to be some compromise.  That's too simple to be 
> usable.  How about allowing some kind of hashbang syntax in the script 
> to pull the language of users choice to execute the update?

I agree... I just furnished it as an example to show that the complexity *floor* for systems like this can be pretty low. Usually the practical design is less minimal than what theory allows.

: