Re: [icnrg] This view on web3has some interesting ties to ICN

[Dirk Kutscher <[e-mail]>]

> …at least in my opinion. It’s not clear if you follow this person’s reasoning whether some of the similar arguments about NDN/CCNx’s allegedly better distributed protocol design matters in practice.

Thanks for sharing and for re-kindling this discussion. I had seen the article, too.

My view on this:

* the hilarious difference between "crypto" claims and reality it well observed, in my experience. I could imagine though that most of the contradictions (actual centralized mode of operation) will probably be ascribed to initial deployment factors by web3 proponents.
* in any case, both the scientific quality (peer-review, let alone meritocratic processes as we have in the IETF) as well as the level of operational considerations (that would be required for Internet-scale operation, and that is typically very important in Internet protocol/system design) leaves room for improvement.

On the analogy with ICN on distributed protocol design and decentralization:

* While some web3 proposal make broader claims, it is IMO obvious that these systems generally try to establish an alternative platform for content (i.e., file) access – but **not** for an Internet-level infrastructure. Even for web services, these systems fall short of expectations, because the web today has so many more applications, services and protocols – take WebRTC for example. Totally unclear, how systems such as IPFS would cater to that. Obviously, ICN operates on a different level.
* The discussion and my analysis on centralization so far has many people, including me, to think that the main driver is really economic concentration and correlated phenomenons such as exploiting personal information on centralized platforms, i.e., it could be seen as a failure to apply antitrust and similar actions. There is probably little hope that ICN would be able to induce any measurable difference in the same socio-economic environment.

Still, IMO it's worthwhile zooming a bit into the technologies for a more nuanced comparison: Most web3 proposals that I have seen assumed some kind of overlay, P2P-inspired system, using DHTs for name resolution etc. – plus a good portion of cryptocurrency fluff of course. These systems are not amenable to decentralized operation in a strict sense, because you need access to some form of logically centralized resolution system etc.

ICN has two main tenants that are relevant to mention here: access to named data as first-order principle (i.e., really not requiring resolution conceptually at all) and the built-in security framework, enabling building local trust relationships for communication without dependencies on external trust anchors (in the blockchain or web PKI).

Unlike most web3 systems, you can really build autonomous application in local networks, i.e., operate in a decentralized way. Moreover you can build systems accessing named data without location notion at all, not only for files or web objects, but for any kind of data and service.

Some of complexities in todays networks stem from issues such as:

* complexities in setting up and managing secure communication endpoints, servers etc.
* limitations in TCP/IP or QUIC/UDP with respect to steering traffic to copies or alternative loci of data/computation

The latter is a real issue in CDN, edge computing today, and ICN has a potential opportunity here in making content replication, local provisioning, location-independent access to computation results less complex without compromising on security.

Now, this alone will not change value chains, business models and general capital concentration – this would take political actions etc. However, legislation and regulation also need a solution space to navigate, e.g., when demanding anti-monopolistic measures such as user data and service portability, there needs to be technical alternative in the first place.

All of this a very much theoretic deliberation of course, unless more systems get built to demonstrate some of this.

Looking forward to your comments.

Dirk