After some summer time off and away from the country, I’m back and also apparently back to blogging. After coming home, I was reading some issues of IEEE Spectrum, which is the flagship magazine of the organization formerly known as the Institute of Electrical and Electronics Engineers. As you know, Kush, after this break I will be joining you at the IBM Watson Research Center and will be in a group looking at problems in service science. Some might argue that what I will be doing will not be electrical engineering, e.g. as defined by IEEE, but I’m not yet sure how to classify this emerging field. Sometimes I claim that electrical engineering is the liberal arts of engineering, in the sense that the education is quite broad, so hopefully I’ll be prepared to make some impact.
And of course I plan to continue to read Spectrum. Coming back to that, one of the articles I was reading was about 3D television. Let me quote from it a bit: “2010 is 1954 all over again, but this time with 3-D. … Will the early 3-D adopters get the last laugh, just like the early color TV advocates? A definitive answer would require 4-D glasses to see into the next decade.” Back at Cornell, I had taken the Human Perception: Applications to Computer Graphics, Art, and Visual Display class with David Field where I had learned about several of the possible technologies described in the article and more, so it was good to revisit that. The article goes on to say, “At the moment, unfortunately, no signaling standard exists. That means the glasses you bought for your first-generation Samsung might not work on your friend’s first-gen Sony.”
The trade-off between standardization and innovation is the rub with technological development, isn’t it? (And indeed the same with formulaic language in epic poetry or mathematics prose.) When we were discussing prize theory last year, this whole crossing the chasm and S-curve business came up, but I wonder how much more there is to it. In any case, I had previously promised to write a bit more about standardization, so let me do that.
Bowker and Star, in their book I linked to above, argue that classification is the first step in standardization and I think they are right. I think the step of coming up with dimensions of measurement, like volume or information rate, however is even more important. That is, the question of what properties to put on datasheets, e.g. in synthetic biology. Defining standardized units of measure is very much intertwined with this.
Stepping back from 3-D tv or color tv to the origins of electrical communication itself in telegraphy, let me quote from Willoughby Smith’s inaugural address to the Society of Telegraph Engineers and Electricians, which was reprinted in the Journal of the Society of Telegraph Engineers and Electricians. Smith said, “It is probably within the memory of the youngest member of this Society that electricians found it difficult to correctly convey their meaning for want of a recognised system of units. How long we should have remained in such an unsatisfactory state it is difficult to say, had not the British Association for the Advancement of Science, in 1860, appointed a committee to determine the best unit of electrical resistance. It was fortunate that this committee was constituted of eminent men who could so thoroughly realise the importance of the work entrusted to them, and who, after eight years of careful labour, produced, not only a practical unit of electrical resistance, but a coherent system of electrical measurements. It was, however, foreshadowed at the time that a future might arrive when, owing to the progress of the exactness of physical measurements and the effects of time, new determinations might become necessary. There can be no doubt that the time has long since arrived, and that the whole subject requires a careful re-investigation. At the present time we are fast returning to the position we had before the formation of the committee referred to.” One of the main issues faced at the time that he was referring to was the question of how electrical resistance was dependent on the material properties of the device and its specific environmental conditions, so a question not only of metrology but of fundamental electrical science. Incidentally, Smith is very Kuhnian in describing the crisis, no?
Moving forward from telegraphy to radio, but still in the pre-Shannon era, let me mention a recent paper by Frederik Nebeker (whom I had met in Bletchley Park). He gives a very nice and succinct summary of the advantages of standards: “The IRE standards, published in 1928, defined three fundamental properties of [radio] receivers—sensitivity, selectivity, and fidelity—and specified procedures for measuring them. These standards made it easier for engineers to design a system as a whole and to optimize overall performance while keeping the projected cost of a receiver at a particular level; they served the engineering community by making possible unambiguous communication; they were valuable to manufacturers for quality control; and they made it easier for purchasers to evaluate and compare radios.” But more interesting to me is when he recounts some words from The Principles of Electric Wave Telegraphy and Telephony, a textbook by J. A. Fleming (the inventor of the diode electron tube): “The quantitative aspect of the subject is, however, of special importance at the present time. There comes a stage in the development of the technical applications of scientific discoveries when exact measurement is the very life and soul of further achievements, and when empirical methods and personal skill have to be replaced by careful predetermination and precise measurement.”
Although I may be way off, I have some of the same feelings about service science that were expressed by Willoughby Smith and by J. A. Fleming about their respective fields, the same feelings that are also often attributed to workers in the communications field in the early 1940s. These feelings, however, hopefully also create a great research environment.
Connecting to your last post, I’m not quite sure how a photograph becomes Lena, how a worm becomes C. elegans, or how a data set becomes the Iris Data Set. I think there are some properties of the thing itself but also external social factors at work. Perhaps one does need those 4-D glasses, or at least some Bletchley Park Good-Turing skills to predict which data set the next blog post will use. (As a side note, unless you steer me differently, I’ll probably talk about going to Antarctica in my next post.)