On means & methods for predicting the timing of IPv6 adoption

Previously filled.


I’m more interested in concepts and approaches rather than the answer itself. Some background on developing the answer to this particular question follows below.

My interest here is more around how to phrase these claims such that they are meaningful over the time scales that we’re working with. One of the difficulties in working with claims over long enough periods of time is that institutions, currencies and measurement apparatus can vary sufficiently or fail outright across those time scales. I’m concerned with approaches to the “and how shall we test that?” Election results: easy; Hollywood movie sales: a procedure exists; Simon-Ehrlich or Bjorn Lomborg type wagers: controversial, but possible; and so on. What is of interest in this particular case is not whether something will happen in a binary sense but how soon something that is known to be happening is going to reach a significant enough level to matter, for some relevant definition of “matters.”

My interest is in how one goes about framing these questions (claims) about the future in a meaningful way, one that will be meaningful in the transition into the future as action against the claims will not be continued if the previous calls to action are no longer relevant. Futurism being but memoir writing if it isn’t tied to planning and thus to action. In the IPv6 case, the timing is important from an online entertainment perspective as the investment in the new networking scheme is significant. From a consumer standpoint, customer premises equipment is rarely churned unless it is force-replaced by the internet service provider. These technologies take multiple decades to roll out and be adopted so they are well within the time-span that we consider in the course (-10 years < t < +10 years).

On using wagers to “know the future”

There have been and maybe still are interesting expreriments with prediction markets against diffusion of innovation questions. Spoiler: the market (game) failed when two fine folks figured out how to run an arbitrage scheme on it; there were no countervailing forces sufficient to counteract the scheme. There exist other surveys in academic paper form of intra-corporate opinion markets. Separately filled.

Behind these is the thesis that a market can “know” things (encode knowledge) that no individual participant can “know” (understand). The aphorism often used is “In the short run, the market is a voting machine, but in the long run, it is a weighing machine.” This folk wisdom is usually attributed to Benjamin Graham and then Warren Buffet.


On the evolving IPv6 transition

In this particular case, we have (the industry has) a sufficient answer for business purposes and elements of the industry have been acting upon the answer as suits their temperament for risk and prospection. Google: yes (as shown), Comcast: yes, AT&T: no, Verizon: yes. RFC 6598 of 2012-04 were developed to support ISPs who have as-built infrastructure that will not be making the transition. Parts of Yahoo’s merchant ad systems went live 2012 (I drove that); the rest is transitioning now. The Google chart is interesting in that it did not hit 5% until two years ago. Prior to that one could reasonably say from a business perspective that there was “no business need” to consider the technology.

Some of these dates give the concept of how long these things take:

IPv6 was “in trials.”
B2B-type production availability; e.g. Solaris.
Microsoft shipped IPv6 default-enabled in Vista.
IPv4 was considered to have been exhausted in North America
“the Internet is full, please dial in at a later time.”

There are other longer technology transitions still under way.

On the continuing 64-bit transition

There is a wonderful survey article from ACM Queue that surveys the transition to 64-bit two-score year transition to 64-bit technologies. One is beginning to see the industry transition wholly to 64-bit for server-class and office-work-class gear and 32-bit and below for so called “IoT” leaf-level devices. The key signal for the future here is Intel’s announcement last month repudiating substantially all of their consumer-focused IoT SBC product lines. The supporting staff is now gone; Intel won’t be in that line of business going forward. And they aren’t a supplier in low-power “mobile” consumer gear meaningfully either. It’s an interesting question that: in 2027, “Intel Inside” means … ???


That other cultural transition, that didn’t happen.

The previous examples are all very in-trade and deeply technical. A more culturally-relevant transition that is still unfolding and is likely to several-multiple more generations to complete is the transition to the metric system in the United States, outside of narrow application domains. We tried but the U.S. population still trades against gallons of gas, quarts of milk, auto tires in pounds-per-square-inch; 55 miles per hour saves lives and gas-gallons, and thermostats shall be set to 65 degrees or lower. We are proscribed against carrying more than 3 ounces of toothpaste into airplanes nowadays.

One could imagine that what can’t (won’t) be different in any future scenario is the U.S. popular or commercial measurement system.

Originally a discussion point for PDV-91.

Comments are closed.