Platt Perspective on Business and Technology

Moore’s law, software design lock-in, and the constraints faced when evolving artificial intelligence 2

This is my second posting to a short series on the growth potential and constraints inherent in innovation, as realized as a practical matter (see Part 1.)

My primary goal in Part 1 was to at least briefly lay out the organizing details of a technology development dynamic, that is both ubiquitously present and certainly in rapidly changing technologies, and all but ubiquitously overlooked and even by technologists.

• One half of this dynamic represents what can be seen as the basic underlying vision of futurologists, and certainly when they essentially axiomatically assume a clear and even inevitable technology advancement forward. I add that this same basic assumption can be found in the thoughts and writings of many of the more cautionary and even pessimistic of that breed too. The basic point that both of those schools of thought tend to build from is that whether the emergence and establishment of ongoing New is for the good or bad or for some combination in between, the only real limits to what can be achieved through that progression are going to be set by the ultimate physical limits imposed upon us by nature, as steered towards by some combination of chance and more intentional planning. And chance and planning in that, primarily just shape the details of what is achieved, and do not in and of themselves impose limits on the technological progression itself.
• The second half of this dynamic, as briefly outlined in Part 1, can be represented at least in part by the phenomenon of technology development lock-out, where the cumulative impact of mostly small, early stage development decisions in how new technologies are first implemented, can become locked in and standardized as default norms.

I cited a simple but irksome example of this second point and its issues in Part 1, that happens to be one of the banes of existence for professional musicians such as Jaron Lanier, who chaff at how its limitations challenge any real attempt to digitally record and represent live music with all of its nuances. I refer here to the Musical Instrument Digital Interface (MIDI) coding protocol for digitally representing musical notes, that was first developed and introduced as an easy-for-then, way to deal with and resolve what was at the time a more peripheral and minor-seeming challenge: the detail-level digital encoding of single notes of music as a data type, where the technologists working on this problem were more concerned with developing the overall software packages that MIDI would be used in. The problem was that this encoding scheme did not allow for all that much flexibility or nuance on the part of the performer in how they shaped those musical notes, leaving the resulting music recordings more crudely, stereotypically formed then the original had been, and lacking in the true character of the music to be recorded.

One of the hallmarks of technological lock-ins is that they arise when no one is looking, and usually as quick and at least easier solutions to what at the time seem to be peripheral problems. But with time they become so entrenched and in so many ways, as the once-early technology they were first devised for grows, that they become de facto standards, and in ways that make them very hard to move beyond or even just significantly change. And such entrenched “solutions,” to cite a second defining detail of this technology development constraint, are never very scalable. The chaffing constraints that they create make them lock-ins because of this and certainly as the technologies that they are embedded in, in effect outgrow them. The way that they become entrenched in the more developed technologies that form around them, leaves them rigidly inflexible in their cores.

Human technology is new in the universe, so I decided while writing Part 1 to turn to consider biological evolution and at least one example drawn from that, to illustrate how lock-in can develop and be sustained over long periods of time: here on the order of over one billion years, and in a manner that has impacted upon essentially all multi-cellular organisms that have arisen and lived on this planet, as well as all single cell organisms that follow the eukaryotic cell pattern that is found in all multi-cellular organisms. My point in raising and at least briefly discussing this example is to illustrate the universality of the basic principle that I discuss here.

The example that I cited at the end of Part 1 that I turn to here, is a basic building block piece of the standard core metabolic pathway system that is found in life on Earth: the pentose shunt, or pentose phosphate pathway as it is also called.

• What is the pentose shunt? It is a short metabolic pathway that plays a role in producing pentoses, or 5-carbon sugars. Pentose sugars are in turn used in the synthesis of nucleotides: basic building blocks of the DNA and RNA that carry and convey our genetic information. So this pathway, while short and simple in organizational structure, is centrally important. And any mutations in any of the genes that code for any of the enzyme proteins that participate in this pathway are 100% fatal, every time and essentially immediately so.
• Think of this as one of biochemistry’s MIDI’s. If a change were made in the MIDI protocol that prevented a complete set of digitized notes from being expressed, any software incorporating that mutation would fail to work, and would constitute a source of fatal errors as far as any users are concerned. Any change: any mutation in the pentose shunt that limited its ability to produce the necessary range of metabolic products that it is tasked with producing, would be fatal too.

Does this description of the pentose shunt suggest that it is the best of all possible tools for producing those necessary building block ingredients of life? No, it does not, any more than any current centrality of need for MIDI and its particular standard in music software as that has been developed, indicates that MIDI must be the best of all possible solutions for the technology challenge that it addresses. All you can say and in both cases is that life for one, and music software for the other, have evolved and adapted around these early, early designs and their capabilities and limitations, as they have become locked-in and standardized for use.

Turning back to biology as a source of inspiration in this, and to the anatomy of the human body with its design trade-offs and compromises, and with its MIDI-like design details, I cite a book that many would find of interest: and particularly if they have conditions such as arthritis or allergies, or know anyone who does, or if they are simply curious as to why we are built the way we are:

• Lents, N.H. (2018) Human Errors: a panorama of our glitches, from pointless bones to broken genes. Houghton Mifflin Harcourt Publishing Co.

This book discusses a relatively lengthy though still far from complete listing of what can be considered poor design-based ailments and weaknesses: poor design features that arose early, and that have become locked-in for all of us. So it discusses how poor our wrist and knee designs are from a biomechanical perspective, how humans catch colds some 200 times more often than our dogs do from how our upper respiratory tract is designed and from immune system limitations that we have built into us, and more. Why for example do people have a vermiform appendix still as an evolutionary hold-over, when the risk of and consequences of acute appendicitis so outweigh any purported benefits from our still having one? Remember that surgery is still a very, very recent innovation, so until very recently acute appendicitis and a resulting ruptured appendix was all but certain to lead to fatal consequences. And this book for all of its narrative detail just touches upon a few primarily anthropocentric examples of a much larger list of them that could be raised there, all of which serve as biological evolutionary systems examples of design lock-in as discussed here.

Looking at this same basic phenomenon more broadly, why do cetaceans (whales, dolphins, etc), for example, all develop olfactory lobes in their brains during embryonic development, just to reabsorb them before birth? None of these animals have, or need a sense of smell from birth on but they all evolved from land animal ancestors who had that sense and needed it. See for example: Early Development of the Olfactory and Terminalis Systems in Baleen Whales for a reference to that point of observation.

I am going to continue this narrative in a next series installment, where I will introduce and briefly discuss a point of biological evolutionary understanding that I would argue, is crucially important in understanding the development of technology in general, and of more specific capabilities such as artificial intelligence in particular: the concepts of fitness landscapes as a way to visualize systems of natural selection, and adaptive peaks as they arise in these landscapes. In anticipation of that line of discussion to come, I add that I began to at least briefly make note of the relationships between steady evolutionary change and disruptively new-direction change, and the occurrence and stability of lock-ins in Part 1. I will return to that set of issues and discuss it more fully in light of adaptive peaks and the contexts that they arise in. Then after developing this foundational narrative for purposes of this series, I will turn very specifically to consider artificial intelligence and its development – which I admit here to be a goal that I have been building to in this progression of postings.

Meanwhile, you can find this and related material at Ubiquitous Computing and Communications – everywhere all the time 3 and also see Page 1 and Page 2 of that directory. And I also include this in my Reexamining the Fundamentals 2 directory as topics Section VIII. And also see its Page 1.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: