Side Project stuff

[JR Conlin]

I've been meaning to send this list out for a while.

One of the projects I putter around with from time to time is the idea
of a Peer to Peer short messaging system (like twitter). There are a few
that exist, mostly based off of bittorrent sync
http://labs.bittorrent.com/experiments/bittorrent-chat.html
http://missiv.es/
etc.

The problem with these is that the underlying protocol isn't open,
therefore it's unaudited, and so it's not really trustworthy. Also, most
of these systems try to prevent public disclosure, which may not always
be a user's goal (Twitter is generally a public platform and succeeds
mostly as such.) My intent was to create a trusted, peer-to-peer system
where posts are signed to show that they did come from the intended
author, but could source from any machine that "follows" them. Machines
could announce lists they follow so that new folks could be discoverable
(likewise, those lists could be posted like graffiti to various cloud
storage services like torrent seeds). There are a few other thoughts as
well, but that's the crux of it.

I've been collecting links to interesting distributed tech for a while
now. Possibly, some of these links might be useful here as well:

https://pinboard.in/u:jrconlin/t:sideproject/

[Martin Honermeyer]

(Shamefully brought this discussion off-list by error. Please excuse
me. See our discussion included. You might like to read from the
bottom, starting with "Did you hear about Twister[1] yet".)

Answer to jrconlin's last mail:

Please do not call it "bitcoin generation", because it does not
generate any bitcoins. Twister has its own blockchain to allow users to
store their username <=> public key information in. So this is more
like a registry.

Mining blocks is necessary so there will always be blocks available to
store new user registrations in. As with Bitcoin, the CPU power needed
to mine blocks will get higher as time succeeds. But I think you can't
compare this to Bitcoin, where _every transaction_ gets stored in the
blockchain, compared to Twister, where only _new user registrations_
are stored. So the number of records will be _much_ lower.

The ulterior motive of those "promoted messages" was that sometime
companies will be motivated to spent some processing power to get their
message to people, which will assure new blocks will always be mined by
someone.

I was also skeptical and would like a registration mechanism which does
not depend on mining. But thinking about it, a blockchain is a superior
mechanism for storing such data decentrally but totally trustable. A
blockchain, because it is a chain, can prove the _order_ in which
registrations happened. That is ultimately important when registering
usernames.


Am Fr, 25. Jul, 2014 um 10:57 schrieb JR Conlin <jrco...@gmail.com>:
> Huh, that confuses me a bit. I'd suppose that the system should
> survive
> without the need for bitcoin generation, since I'm effectively
> providing
> resources by hosting a node. Essentially, the system should be self
> sustaining from that point. (Granted, if the system becomes very
> popular, it means that I may need to take on more traffic, but again,
> I'd wager that there would be commercial providers willing to step in
> (like hosting providers and VPNs).
>
> Granted, I'm EXCEPTIONALLY paranoid (ask Tarek or Alexis sometime),
> so i
> tend to dig in and be suspicious of all sorts of things.
>
> Also happy if you want to post this back to the list for further
> discussion.
>
> On 2014/7/25 1:49 PM, Martin Honermeyer wrote:
>> No no, it's not using Bitcoin wallets. It has nothing to do with the
>> cryptocurrency. It just uses the blockchain mechanism from Bitcoin.
>>
>> That means people have to mine blocks for the system to thrive. They
>> are rewarded with being able to send so called "promoted messages"
>> to
>> all users. This is a kind of spam. But it is rate-limited so you
>> won't
>> get more than 1 message per 8 hours. A totally acceptable drawback
>> in
>> my opinion.
>>
>> Also, sorry for sending this off-list. That was not my intention.
>>
>> Am Fr, 25. Jul, 2014 um 10:44 schrieb JR Conlin
>> <jrco...@gmail.com>:
>>> Ah! No, I hadn't seen that one yet.
>>>
>>> Interesting idea using bitcoin wallets as the storage mechanism.
>>> May
>>> take a bit of time to go over the code to make sure that I'm not
>>> also
>>> spending my electricity to make someone else profit. ;)
>>>
>>> Thanks for that.
>>>
>>> On 2014/7/25 1:13 PM, Martin Honermeyer wrote:
>>>> Did you hear about Twister[1] yet? It uses fully open protocols:
>>>> A
>>>> Bitcoin-like blockchain for username / key registration and
>>>> Bittorrent
>>>> for sharing (signed) posts and metadata. So you can trust
>>>> identities
>>>> and be sure of the origin of each post. Apart from the technical
>>>> facts, it works almost exactly like Twitter, with public
>>>> hashtags and
>>>> a private conversation system built in.
>>>>
>>>> I am already using it for about a month. It works quite well.
>>>> I'd like
>>>> the (Twitter-like) 140 character limit (and some other
>>>> limitations) to
>>>> be broken up someday though. Also, it has not been audited yet.
>>>>
>>>> [1]: http://twister.net.co/
>>>>
>>>> Am Fr, 25. Jul, 2014 um 6:19 schrieb JR Conlin
>>>> <jrco...@gmail.com>:

>>>>> _______________________________________________
>>>>> Dev-decentralization mailing list
>>>>> Dev-decent...@lists.mozilla.org
>>>>> https://lists.mozilla.org/listinfo/dev-decentralization
>>>>
>>>
>>
>

[jr conlin]

True, his white paper describes how he's using the coin generation
mechanism much better: http://arxiv.org/abs/1312.7152

There are ways to do the sort of identification without using a block
chain, but they're (by definition) not self contained, or in the case of
things like x500, a tad unweildy for completely different reasons. There
might be ways to make the block generation a bit more transparent as
well, which might help waylay some concerns.

I'm still going to dig into this a bit more, but I do agree that it does
sound very promising.