Re: Reviewing the Solid protocol

po 10. 4. 2023 v 19:54 odesílatel Gavin Nicol <gtn@rbii.com> napsal:

> My note about 'pod' is a more generic problem with external storage
> (especially semi-structured, or structured)  that is not directly under the
> control of the user - SOLID being one case. Minimally these systems expose
> the user to risk of denial of service, and creates pressure for
> centralization. Saying that users a free to host their own server is a poor
> answer because historically the people that have actually done that are on
> the fringe... how many people choose to host a webdav server instead of
> using google drive or dropbox?
>
> Another good example of a similar nature is medical records - everyone
> decries the rough monopoly that EPIC holds, and there are a myriad
> proposals (many of which use blockchain as a meme to imply security and
> decentralization) that boil down to "EPIC is an evil centralized monopoly
> that is monetizing your data! We must be free of this - all you have to do
> is move your data into our system, and we can break the chains!". The
> outcome of that is obvious given the business incentives.
>
> It's not clear that path forward is to have a server-based system for
> sovereign data at all. Given the storage capacity of modern mobile devices,
> broadly available connectivity, and modern cryptography, there are other
> approaches. Would exploring those fall under the charter?
>

I believe there are (at least) three distinct aspects to consider when
examining the centralization/decentralization spectrum:

   1.

   Specifications: These guidelines determine what can and cannot be done.
   A centralized specification offers relatively few choices or paths, whereas
   a decentralized one provides numerous options. Solid likely falls somewhere
   in the middle in this regard.
   2.

   Data ownership: Centralized websites represent one end of the spectrum,
   as users are tied to the website. Federated systems offer a slight
   improvement; however, an inherent asymmetry exists, where site owners can
   cancel users, but users cannot cancel site owners. Solid is a federated
   system and shares this limitation. Nevertheless, allowing migration from
   one site to another could be a potential solution. While Solid doesn't
   currently support this feature, the use of relative URIs may lay the
   groundwork for it.
   3.

   Data expression: Central websites typically provide limited options for
   user-generated content creation. APIs offer some improvement, but they
   often conceal data. A universal API permits decentralized data storage,
   akin to writing on paper, but with the added benefit of hyperlinks to form
   a graph. Solid excels in this area, enabling high flexibility through
   Turtle and JSON-LD.

So, there are various forms of decentralization, and each element of that
is part of evaluating an overall system.

I think it will be a quite good innovation.  My slight concern is that html
is nowhere mentioned in the spec.  So it might be underspecified how
developers interact with solid, for example, in the browser.

>
> On Sat, Apr 8, 2023, at 1:25 AM, Melvin Carvalho wrote:
>
>
>
> st 29. 3. 2023 v 3:21 odesílatel Gavin Nicol <gtn@rbii.com> napsal:
>
>
> There are quite a few competitors to Solid out there in various forms:
> self-sovereign identity, self-sovereign data etc. are all being/have been
> worked on for quite some time now (at least 10 years) and it still feels
> premature to standardize on anything. There are lots of technical and
> social issues to work out...  in the case of Solid (nominally
> decentralized), it's somewhat questionable why a Pod is necessary, for
> example.
>
>
> Hi Gavin, the term pod has been removed from the charter
>
>
> https://services.w3.org/htmldiff?doc1=https%3A%2F%2Fraw.githubusercontent.com%2Fsolid%2Fsolid-wg-charter%2Ffix%2Fcharter-drafts-template%2Fcharter%2Findex.html&doc2=https%3A%2F%2Fraw.githubusercontent.com%2Fsolid%2Fsolid-wg-charter%2Ffix%2Fuse-technical-terms-storage%2Fcharter%2Findex.html
>
> And also occurs nowhere in the spec.
>
> Instead the term "storage" is used, which can be thought of as an enhanced
> WebDAV with access control
>
> If you had examples of competitors, it may be possible to do a comparison
>
>
>
>
> On Tue, Mar 14, 2023, at 11:30 PM, Danny Ayers wrote:
>
> It is in scope for TAG, the W3C and whatever. But the idea your version is
> the one. If you want to keep the W3C relatively independent, that doesn't
> work. If Microsoft and Apple have to drop their version of scripts, so
> should you.
>
> On Sun, 24 Jul 2022 at 15:18, Tim Berners-Lee <timbl@w3.org> wrote:
>
> Solid is a growing protocol/movement, and the tech parts of it — the Solid
> Project — are basically a W3C Community group.
>
> Solid adds things which the web needed but hadn’t yet standardized,
> including global single sign-in, standard access control, and a fast API
> for data read-write between an app and a store (a Solid Pod).  By making
> the API to the store universal, it means you don’t have to change the store
> when you make a new app, which completely changes the architecture and
> markets and business models which are possible. It also leaves individuals
> empowered rather than exploited.
>
> Would it be reasonable for the TAG to review the architecture at a high
> level, or review the protocol?  It would be useful to get a knowledge of
> the Solid stack in neighboring parts of the technology.
>
> (A separate future question are the client-client interop specs which are
> needed for interop between apps, such as contacts, chat, etc.)
>
> See https://solidproject.org/. https://solidproject.org/TR is where the
> specs end up after their github-based proces.
>
> Best wishes
>
> Tim BL
>
>
>
>
>
>
>
> --
> ----
>
> https://hyperdata.it <http://hyperdata.it/danja>
>
>
>
>