Re: One Ring To Rule Them All

On 2 June 2013 23:25, Melvin Carvalho <melvincarvalho@gmail.com> wrote:

>
>
>
> On 2 June 2013 17:27, Evan Prodromou <evan@e14n.com> wrote:
>
>>  I think it's unlikely that we're all going stop hacking, sit down around
>> a big table, hash out the perfect social networking protocol, and then rush
>> off to implement it.
>>
>> First, because if you think about it too little, you come up with an
>> insufficiently powerful protocol to do what people need done.
>>
>> Second, if you think about it too much, you'll go down so many ratholes
>> that you'll never actually publish a protocol.
>>
>> I think that by their nature, FSW technologies require internetworking
>> protocols for instance-to-instance communication.
>>
>> I think that developers will implement those protocols that make sense
>> for their users, or for acquiring new users. I don't think they'll pick a
>> protocol because it looks great or because it's easy; they'll do it because
>> they have to. Because there are lots of users on that other internetwork.
>>
>> There will probably be some components that we'll see making up most of
>> the internetworking and client interfaces from here:
>>
>>    - domain-based IDs (HTTP URLs and/or Webfinger)
>>
>> HTTP URLs, yes.  Webfinger is promising, bit it is not yet a standard.  I
> know you follow the IETF standardization of WF, but it's not ready yet.
> There's good reasons for that.  But I do see progress.  When WF can
> interoperate with other serializations it will be first class.  I actually
> think the WG is doing a great job but they inherited a huge mess based on
> XML and XRD.  Moving to JSON has been a big plus, imho
>
>>
>>    - RESTful APIs
>>
>> Definitely RESTful APIs are extremely powerful, perhaps more powerful
> than the web itself.  Tho you'll find almost all APIs break the rules of
> REST one way or another, but this still is OK, in most cases.
>
>
>>
>>    -
>>    - JSON
>>
>> JSON is very advantageous in that once you fetch the object, it's all in
> memory.
>
>
>>
>>    - OAuth
>>
>> I dont think OAuth is the one auth system to rule them all.  It has it's
> place as part of a trusted third party paradigm, but it's only one way.
> Auth comes down to sharing enough entropy such that your attacker has
> little incentive to try and attack.
>
>
>>
>>    - HTTPS for on-the-wire security
>>
>> A nice goal, and I agree.  But we are still some ways from "HTTP
> everywhere".  Facebook actually got to 100m users with HTTPS.
>

Sorry typo mixed up the two, that should read "HTTPS everywhere" :)


>
>>
>>    -
>>
>> Finally: I think federation can be well-served by a monoculture of Free
>> and Open Source servers. There are network effects between users, but there
>> are also network effects between sysadmins, developers, documentation
>> writers, translators, third-party developers, and so on. There are hazards
>> of stagnation, and choke points, but more people working on the same
>> codebase is better than lots of people working on different codebases.
>>
>
> FLOSS certainly is a big plus, but it's not a magic bullet.  We need to
> demonstrate interoperability that will grow the network effect.  We need to
> allow friending from one system to another, or establish why it's not
> practical.
>
> Thanks for the comments, enjoyed reading, and thought provoking!
>
>
>>
>> -Evan
>>
>
>

Received on Sunday, 2 June 2013 21:28:29 UTC