Jim, in the Gilder testimony of 2004 that you cited, he made a number of statements that I believe he wouldn't make today, given the events and developments of the past two, short years. He's got the calculus for the three basics right: storage, bandwidth, processing. But, I don't believe he's quite got it right when it comes to applications and the moving parts of the delivery of services. In some ways his praising of Google in 2006 demonstrates my last point, since Google's content delivery at the upper layers, using other peoples wires (and routers and switches) at the lower layers, is now regarded as the model to beat. Yet, it runs contrary, if not down right antithetical, to the pro-silo leanings he voiced back in 2004.
Likewise, his regard for Huber's Corvis, for whom I have a good deal of respect, is replete with interpretations and errors that only Gilder could make (and usually get away with), where he confuses "content" and connectivity. He also makes some statements about what an all optical network can do, which I agree with as futures, but not today.
During these last several years a little company named Infinera has demonstrated dispersion compensation capabilities that were once reserved for optics only. Also, by keeping things "electrical", or supporting OEO at a very low power scale, it is delivering many benefits still not achievable in an all optical model, especially with respect to signal handling and mediation that won't be achievable in photonics-only elements for another decade, at least.
I began posting about Infinera on the gildertech.com forum back in 2001 and it never received traction because it was always in the shadow of Corvis. Today Infinera is considered an ascendant play in those quarters, and Infinera has, with its OEO design, become the choice of Broadwing, as well. See:
"Infinera Gets Corvis (Sort Of) JULY 18, 2006 - Corvis, the all-optical company, sees its gear end up in the hands of anti-all-optical startup Infinera"
lightreading.com
The last third of the testimony he gave reads like a monthly GTR (Gilder Technology Report), dilled with pointers and details about his ascendant stocks.
In fairness, I still believe he's on the right track w.r.t. the ultimate dominance of the lightpath model, but like other aspects of his testimony, he's talking in the present tense about technologies that won't be ready for mass market adoption until sometime beyond 2010.
The remainder of your post was both inspiring and in some ways provocative, but also confusing in places. For example, what did you mean by "denigrate the per-bit life-cycle cost of fibre?", when we were discussing wireless bits. In any event, wired or wireless, assigning attributes to bits can be considered a fool's game at times, when you think about it. Bits sometimes exist in the absence of a signal (e.g., 0 vdc during a prescribed period, absence of a zero transition, pause), and can actually be encoded to mean eight or more bits during quiescent times, as well, depending on the convention used in an encoding scheme.
I've got to hand it to you that your characterizations of sloth, foot dragging and sloppiness in attention paid towards developing cogent policies were very much in line with my own thinking. I'll leave it at that, except to state: Well done! and Thanks.
FAC |
