Re: [arch-d] [EXT] Re: IAB Technical Discussion on Fragmentation: 2023-05-03

[Brian E Carpenter <[e-mail]>]

Vittorio,

On 06-May-23 20:56, Vittorio Bertola wrote:

...

> The original idea was that we would have places where we would talk and reach rough consensus around policy compromises, such as the IGF, and then the technical community would follow. However, "the technical community never accepted to consider anyone else's views and in most cases only showed up in those venues to say that engineers know best and that the decisions they made in the 1970s were still the best possible ones" (sorry for this summary, but this is how things were described to me by a government officer some time ago).

It's many years since I last had the pleasure of discussing Internet infrastructure with government officials, but I still think this is a caricature. (Especially since the basic decisions taken in the 1970s were taken under the aegis of a US Government agency, but let's leave that aside.) That type of criticism used to come from officials who were heavily under the influence of PTT monopoly thinking; we never got it from officials who were working to break those monopolies.

Well, that was a long time ago and the issues are different now, but I still doubt the validity of the argument. What is undoubtedly true is that changing the nature of Internet infrastructure is essentially impossible at this point in history; even the trivial change of increasing the address size has taken us 30 years and is only 43% complete.

The issues discussed here (splintering and centralization) aren't really infrastructure issues anyway, so the potential IAB/IETF influence is quite small.

> 
>> The fact that the Internet is permissionless and implementation
>> precedes standardization is good: it creates a dynamic environment
>> that has tremendously benefited everyone involved, and has prevented
>> rent-seekers who would like to stop investing and innovating and
>> instead charge tolls.
> 
> You must be stuck in time at least 15 years ago. Today, the most common device for accessing the Internet is a smartphone, on which you can only pick between two operating systems. Each of their owners controls everyone else's access to offering services: you cannot distribute an app without their permission and without abiding by their unilateral requirements, and often without conceding them small perks (rents) like a 30% commission on any payment. And this is just an example: the cloud infrastructure market, as another example, is increasingly like that, with 3-4 dominant operators (mostly the same as above) progressively conquering all others.

Correct. Almost none of that is a result of the infrastructure specifications from the IETF. Ironically enough, after breaking the national PTT monopolies 20 or 30 years ago, the Internet is now the superhighway for a new set of international monopolies. As I used to say many years ago during early discussions of network security, don't blame the roads for helping the bank robbers escape.
    
> As happened before in other industries, the lack of regulation and the consolidation that follows innovation fostered mono/oligopolies, and now we need a bit of permissions and planning to rein them back and restore choice and competition. This will inevitably make the Internet less uniform, less global and less borderless, but it's a price that we need to pay.

It's the upper layers that are concerned here, not the transport infrastructure. Is the W3C having the same discussion?

    Brian