[ih] History of the Economics of the Internet?

[Brian E Carpenter]

On 30-Jul-26 07:22, Craig Partridge via Internet-history wrote:
> Adding perspective on two points in this excellent note.
> 
> 
> On Wed, Jul 29, 2026 at 7:34 AM John Romkey via Internet-history <
> internet-history at elists.isoc.org> wrote:
> 
>>
>>> On Jul 28, 2026, at 5:35 PM, Brian E Carpenter via Internet-history <
>> internet-history at elists.isoc.org> wrote:
>>>
>>> I think there are some really interesting economic questions, such as:
>>>
>>> - why did advertising, and charging for domain names, become major
>> factors in Internet economics?
>>
>> For domain names, scarcity. This was unintentionally encoded in the system
>> by having only a handful of TLDs at the start, with .com the most
>> recognized and therefore most valuable. Even if businesses hadn’t come
>> online “smith.com” or “sex.com” would have had a lot of competition for
>> them and therefore value.
>>
>>
> Which was ironic, as one of the major motivations for domain names was to
> solve the problems of a flat namespace. In a flat space, there could only
> be one "frodo" and domain names meant every sub-domain could have its own
> "frodo".
> 
> Dredging my memory from discussions of the period, I would say the
> expectation was that everyone was thinking in terms of universities (so
> subdomains hanging off example.edu) and not enough about corporate names.
> DNS also came well before the Web.
> 
> 
>>> - how has the design of BGP4 affected carrier economics (competition vs
>> collaboration)?
>>> - why did flat-rate charging become the dominant model for ISPs?
>>
>> I remember John Curran predicting this in the early 90s (I think it was
>> the early 90s anyway), basically it can cost more to meter than it’s worth.
>> At some point the cost per bit becomes noise.
>>
>>
> While I'm quite sure my buddy John said this, there were several prophets
> of flat rate charging at the time.  I think all the NSFNET regional
> networks used it.  Steve Deering strongly pushed back against people who
> suggested usage charging led to efficiency and would point out the effort
> they put into metering wasn't worth it (and it shaped some of his IPv6
> thinking).

Somewhat related (and also related to my question about BGP4 and carrier
economics), I posted "Metrics for Internet settlements" [1] in 1996, and was
taken out for lunch at the next IETF by a couple of ISP heavy-hitters, one
of whom was John Curran, to explain to me in no uncertain terms that if I
ever brought this draft up in an IETF session, they would conspicuously leave
the room on account of the Sherman Act. I thought they were over-reacting,
but this was clearly an aspect they were worried about.

Apart from the complexity and overhead of detailed accounting, I suspect
that avoiding anti-trust pitfalls also helped to lock in flat rate charging,
as well as the BGP4 peering model.

[1] https://datatracker.ietf.org/doc/html/draft-carpenter-metrics-00

    Brian