Re: Webmention writeup in The Register

On 18 February 2016 at 23:47, Ben Werdmüller <ben@withknown.com> wrote:

>
> Question I don't know the answer to: is it ever acceptable to attach to a
> centralized system (eg Akismet) to provide services like spam prevention?
> Or should our goal always be complete decentralization? If the latter is
> the case, can complete decentralization ever capture network-wide discovery?
>

Complete decentralization is impossible.  Specs themselves are a form or
centralized out of band knowledge sharing.  Decentralization vs
centralization is a continuum, where you lean one way or another.  What
probably defines where the needle is, is how much freedom the participants
have to make choices.

Taking webmention as an example (based on the last time I reviewed it) it
has a great decentralized feature, namely, that it promotes cross origin
communication.  This gives you are really free way to communicate on the
web in a relatively decentralized way.  Well done!

But it also has built in degrees of centralization, such as having to
support form encoded parameters (while the charter of this group is
supposed to be solution in JSON for example) and dependencies on
microformats etc.

So when adding a service like askimet, the question to ask is: how well
does the formulation allow implementers to be free to do what they want,
and hence grow the network effect by allowing different solutions to work
together.  If you say askimet is the ONLY spam prevention technique, that
might be slightly too restrictive (or it might not) but if however you say
spam is a real problem and I'd like to be able to slot in a set of
solutions to help, askimet being one of them, that is a much more
decentralized formation.

In short, some centralization is OK, e.g. using ascii or unicode (or even
binary!) are things that can get broad acceptance because it allows you to
do pretty much everything you need.  Some forms of centralization favour
one solution over another are rarely acceptable unless you can show
ubiquitous adoption.


>
> ben
>
> On Thu, Feb 18, 2016 at 1:41 PM, hhalpin <hhalpin@w3.org> wrote:
>
>> On 2016-02-18 14:03, Ben Werdmüller wrote:
>>
>>>
>>>
>> I'm super-happy to see WebMention get some good press. However, there is
>> a point that the article makes rather correctly re spam, and it applies not
>> only to WebMention but any federated system.
>>
>> While unforunately W3C/ERCIM didn't renew my D-CENT contract, we're
>> booting up a new research project at INRIA and University College London to
>> look at this problem called NEXTLEAP. The same researchers behind it are
>> the folks who discovered the TLS attacks and statistical disclosure attacks
>> on Tor, so expect some good work in this space to be fed to the W3C
>> shortly. I'm sure its a solvable problem.
>>
>> --
>>
>> Harry Halpin (W3C/MIT) harry@w3.org
>>
>
>
>
> --
> *Ben Werdmuller*
> CEO & co-founder, Known
> withknown.com | werd.io
> <http://goog_1933028737>
> +1 (312) 488-9373
>
> Known, Inc | 421 Bryant St | San Francisco, CA 94107
>