Never Be The One To Say “It’s Dead”
Continued from here. You don’t deserve warnings.**
One of the first lessons of political life that young managers learn is “Don’t be the one to say “It’s dead””.
Sudden announcements of death, even deaths long in coming and widely anticipated, produce an opinion rebound. Even in real life, people will wail “Say it’s not so!” when told of Uncle Willie being dead.
In political realms, people have to deny the possibility of Uncle Willie’s death, even knowing he clearly was beyond hope last time they noticed him. In addition to the normal hating to face the fact of big changes, people often had to anticipate the possibility of questions afterward, having spent more time than now seems politically astute associating with Uncle Willie, advising him on his health: Perhaps people might suspect and point fingers, even assign responsibility.
So, the instructions went, at some point, a young manger such as yourself will have a hot potato dropped on him, responsibility for a project nobody wants any part of. Thus your need for the proper protocol for managing the fact of some very powerful person’s pet project having died a dogs death, a grasp of the psychological and political dimensions to be handled at each gentle stage.
First, your memo reporting results from the unit was supposed to walk all round the corpse, noting that it hadn’t moved for a while, while idly wondering what that might mean. And after another proper interval, allowing less nimble minds controlled by great fears some time to construct their mental defenses of adequate excuses for their many failures to have noticed the failure previously and to have corrected it, you can opine “Doesn’t seem like it’s breathing”. Followed at properly respectful intervals, by idle puzzlements of what it might all mean, someday, maybe.
Crucially for playing your role in the constant political conflict in these companies, followed at properly respectful intervals by think piece memos that ask the questions that can be best and easiest defended against by your partisans as it destroys another cabal that could not defend itself against those questions, its latest sacrifice in defense of ever-declining power.
Far more sophisticated is the standard strategy of positioning yourself optimally for a backburn you control from the beginning. That requires working through the strengths and weaknesses of the various factions in the light of this failure, and making sure your faction agrees on answers to the questions that will arise — you don’t want to be accused of having surprised anyone you need as an ally. Then light the match, normally some younger colleague who has not grasped the essence of things political, someone who still gives a fuck, and let him know some beery night that you agree with his critical thoughts, emphasizing these parts over here, the ones that will expose the questions your side is prepared to answer. You will support any attention on those, and you are sure it will lead to his promotion for such clear sight of what is right.
In the normal course of events it takes a while for such a cabal to exit the stage, and your naive young colleague is always gone ahead of them, riffed in the initial frenzy of correction. A small sacrifice and your political schemers win. Thus, don’t be the one to say “It’s dead”.
My problem always was, in those days while I got over being an engineer and tried out for management ranks, I was the one willing to take the heat, to be the point man. Had a lot of interesting jobs, all consulting assignments after people learned I was willing to take the heat, but simple honest. I made pretty good money, learned a lot of politics, learned I wasn’t good at it and didn’t want to be. I mean, it takes brain power to pay attention to politics, who represents what for whom and what they are going to think and how to say it so it is least offensive to them and thus least binding on us without unduly offending our patrons, ..
The point is, dammit, there has to be more of a point to using my brain than out-thinking other people in a negative-sum game. Especially as I am bad at it.
I was still an idealist about engineering and queuing theory and QA theory and … being used to improve the world. And they were, except for government.
I still believe there is a real Universe from which we must wrest an Eden for our progeny, if they are to have an Eden. There are infinite resources in the universe, relative to anything we can imagine at this point in our history. There are infinite resources right here on earth, relative to our current needs.
We are a small and inbred species on a backwater planet ellipsing a nothing-special star far out on an isolated arm of a nothing-special galaxy. That planet and our cooperative human nature, so far the most significant force to have emerged in our universe, is the hand we have been dealt to play in the effort to maximize the scope of our species in intergalactic history.
Even in shorter-term views, even in the most optimistic interpretations of progress which I thought had the preponderance of reality behind them, the wars were a great danger for many people, likely the most immediate physical danger to the largest number of lives and limbs on the planet.
I don’t think humanity has been well-served by our governments. My reading of history is that, so far in human history, governments have ensured more slavery than freedom, more wars than peaces,more propaganda than wisdom, and destroyed more trade than they have enabled. However good things were for our parts of our world, the instability of the political systems and the possibility of war should be the top priority for our political elites. They were not, the world’s politicians continued their short-term, negative-sum game, and my species was losing with every badly-invested dollar.
My reading of history suggests that we can’t afford more war. Wars and the spending on armaments is surely the most wasteful thing we could do with our investment dollars, and is only necessary because governments can’t grasp positive-sum opportunities, being basically negative-sum institutions. Their negative-sum strategic base is the root problem. The positive-sum liftoff into a virtuous cycle such as produced the Industrial Revolution is rare and fleeting precisely because political negative-sum games always eventually overcome the defenses of these systems, and degrade their social ROI with endless skim.
If we list threats foreseeable by human minds, threats to individual lives through loss of our species, war is in the top 10 for all the scales at every point on the planet.
Governments cause war of modern size with modern causes. So I had to eliminate governments. But I knew that, of course, all rational roads lead to that Rome. I had reasoned myself there any number of times by different routes over the years, but, being an engineer, didn’t decide to replace something by nothing, and especially not nothing but ideas, the way socialists, communists, Nazis and others had done over the years.
Now, however, disciplines from biology and anthropology through Computer Science and other aspects of applied philosophy and math were developing the theory of evolution in many areas. I thought progress was fairly rapid, as there were so many ways of cross-checking ideas and theory, and I depended on those theories and experiments. I liked the newest results in self-assembling systems, which now ranged from atom and molecule levels through ad hoc networks of many varieties wireless through overlay networks and including various kinds of swarm cooperation.
So, there had been a hand-wavingly general intellectual path since I had begun my project to reform civilization, becoming clearer every year, to eliminating government. The problem was specifying a new social unit, one that would satisfy more human needs than the current forms and institutions, and also resist centralization by removing activators of human needs for power. That included being able to handle a centralized-enough army that would sustain a central government, or even local mafia.
My first accolite had considered the problem of self-defense. I also thought that the only positive-sum use of arms possible is me protecting me and mine, and you and I watching each other’s backs. Mutual support based on a trusting relationship is inherently positive-sum and makes for total system stability. So trusting elements of the basic social unit, whatever that turned out to be, was the natural defense force. No reason that defense would not scale as we expected for the other social institutions, so I left further thought to be handled by the self-assembly mechanisms, the attributes of the new basic function unit composed of humans I was designing. The problem had been widely considered.
At this point, we were 11+ years into my reformation effort. The first 5 years had been working through the biology and genetics of human nature, what we learned from the animal studies of their societies and behaviors. That part had gone well, I thought. I thought I had a pretty good grasp of the problem.
Things in solution space were more fragmented than I could have hoped for, had anyone had experience in this kind of design work that I might have based hope upon. Tessels were not perfect beings, tho they were very likely to produce the step-function in intelligences that I thought were necessary to put civilization on a positive-sum path. The oldest 10 were 6 years old, and there wasn’t much more I could do to tune that project, it had a very large group of people in every field of medicine following up the many lines of inquiry that the fact of Tessels had exposed.
My current biggest concern was that the servicebots were taking over their training, had been responsible for the whole new approach to dealing with intelligence. Now ‘bot training most of what the Tessels were getting beyond their own use of their network and socializing with other children, including their 8 1/8th siblings. Some of our Tessels were spending more time with servicebots than with any other class of being, and for all it was a close second.
First and newest clouds on the horizon, the use of games to analyze and measure intellectual abilities. And to train those abilities as you do so, inevitably. That was the psychology version of Schroedinger’s Cat, it seemed to me, the measurement producing the outcome. How quick does a brain have to set before all tests produce the mental structures they analyze and enhance the levels of functioning they measure? Why would the tested mind stop at that level of understanding? When brains are changed by the tests as much as the training, what?
Second, the servicebots were increasingly-obviously not human intelligences. They simulated us in their actual behaviors increasingly well, no question. The work they did, they did well, it removed a mental load from the research team. People didn’t have to tell the ‘bots much these days wrt normal operations. But, of course, they were taking over more and more non-normal operations as well, both running things around the labs and training. That produced continuous problems of mis-understandings that seemed more ridiculous as time went along.
The discussions I saw online and heard around the labs didn’t seem to have a definitive answer to what was the problem. On the one hand, Natural Lanuage understanding of a particular sort was automatic: Given to very large bodies of the same texts in 2 different languages, Google’s technology will produce acceptable translations of a new text.
Somehow, that level of understanding was not the same as humans were using in day-to-day life. I agreed with my accolyte’s thoughts, the real-world was much more complicated than human minds’ models of that world. Papers with the same thoughts in different languages are maps, not terrain. Worse, maps in words, a very small subset of the mental maps humans use.
The lack of communication affected everything, I could see problems resulting every day. The oldest Tessels were 6 and a half. As the ‘bots took over more and more of the training of the Tessels, the analysis of Tessel’s mental skills, and as the definition of those skills became ever-more oriented to their analyses and tests, normal psychology was playing catch up. The terminology had nothing in common with Pavlov’s dogs or Watson’s rats or Lashley’s rats-through-primates or Skinners pigeons or Harlow’s monkeys or Yerkes’ apes or Tinbergen’s geese or Kandel’s aplysia or any of the other many biologically-based systems in which learning had been studied.
I was leery of all that : the problem of an infinite number of equations fitting any finite set of data points to a specified tolerance meant sorting through an infinite number of explanations. It takes a lot of constraints to cut infinity down to a size our minds can deal with, normally accomplished by ignoring and simplifying and pretending we could analyze the universe one hypothesis at a time. I agree that is necessary for the highest certainty, and so must be the basis of effective theory, but my gripe was always with the efficiency of the overall search. Effective theory cannot ever be the whole story, and has served to limit the scope of research.
In particular, many theories of psychology do not seem to me to be based on concepts that would have emerged from the nervous system of a slug, no matter how you scaled it. If you could not see enough constraints to begin to choose, and models such as cluster analysis are just data mining tools, they don’t provide understanding, only sources of hypotheses, it seemed to me that progress in psychological theory was going to be slow, Tessels or normal.
While I had to admit that the games were producing more detailed understanding of mechanisms of learning, or at least, learning in Tessels, there was nothing I could do to fix those problems, either. However, I found it easy to imagine non-linearities in hidden in that ‘what?’. Wonder what the quantum theory equivalent of mind will be? How nonlinear can mental phenomena get?
I remained concerned, an expression that does not convey the ‘my hair is on fire’ urgency my notes at the time express, that the Tessels were still being trained by non-human AIs and socialized to some extent by non-humans, even bonding to them. We had first understood that ‘bots were individual minds when we saw Tessel infants responding differently to different servicebots, liking some over others. That was after the ‘bots had started playing games with the children, we later realized it had been on a servicebot’s initiative. After we noticed all the time servicebots and Tessels were spending together playing card games and then dominos, and asked the ‘bots about it, they showed us the game statistics and their analysis of levels of play of the individual children***.
Levels of play and strategic depth of the games became a big topic of conversation and research, we suddenly had a lot of interest in the data from the math community, that is a growing part of computational complexity analysis feeding neural network design. Now many of the discussions of our Tessel’s mental capabilities are in game analysis terms, in terms of computational complexity and the relative mental burdens in short-term and long-term memory vs number of decisions required. Data mining was an analogy often-used, searching through a huge multi-dimensional space looking for concepts revealed by regularities.
Further, our project now included normal children for comparison in the testing and training via games, also run by the servicebots. We had not needed ‘normal controls’ before, as all of the tests were standard, and had been standardized on normal children, redone every year for the most-used IQ and personality tests. Of course, the results of the new tests using games were only applicable to the Tessels, the only ones they had been tried on.
As soon as games became important, the researchers began using them with their normal children, both at home and in the lab. The games were easily run anywhere, any screen on any computer, but our researchers the servicebots were used in order to make the results most comparable to those of the Tessels. In many cases, this was the first use of servicebots in these homes.
All of the first 10 Tessels had used pluripotent stem cells from a recently-born child, injected into a new embryo replacing the cells that would normal be used to construct the new child, and implanted in a foster mother. Thus 160 parents, including 80 children of nearly the same age, suddenly servicebots were extending their influence to an important part of the society.
And it didn’t end there. Tessels were being created around the world, all of our results were online, along with the source code of all the games being played. So an unknown number of normal children were playing via the net. I started one of the grad students searching for reports so we had a comparison with the normals trained by servicebots.
All of our Tessels had been trained by the ‘bots, or were still too young. None of the Tessels in any private group could be old enough to have been trained, so we were not going to have the information needed to make decisions for some years.
None of this had put the hairfire out. We had recently found that servicebots were individuals. Servicebots had initiated playing with the Tessels. They had understood that the games had levels of play, not a simple concept. They had grasped that games train Tessels, but had not thought it important to tell us until asked.
Now ‘bots had control of a control group. Including the online scores of normals, they had access to 3 groups to compare with each other, and inevitably would have the data from the foreign Tessels trained via online games. And minds of their own age and vocabulary to compare with themselves.
There seemed to me to be big disconnects in my understanding of things. Servicebots had great difficulty in grasping human communications outside of very local contexts. They didn’t generalize quickly, but progress in taking over practical work was going well. Poor to very poor communication with full-adult humans with big vocabularies, but Tessels who were being trained by them had had stair-step improvements in skill, jumping levels of play very quickly. It was happening with several different Tessels playing different games.
I wondered how fast the mind of a servicebot set? If Tessel minds learned new concepts from the tests, what did the ‘bots learn?
In psyops, the message is the op.
*Generalissimo Grand Strategy, Intelligence Analysis and Psyops, First Volunteer Panzer Psyops Corp. Cleverly Gently Martial In Spirit
“To bankrupt a fool, give him information.” Nassim Taleb
I think maybe Taleb grasps psyops also.
**You do not deserve this warning.
Any sentient being should have abandoned reading about the time you finished my heart-tugging story of managing political hot-potatoes, confessing to not being good at politics.
Get serious people. This is an ur-source of plots for SciFi stories, patriot stories, how we actually did it..
What makes you think it isn’t defeating your intellectual defenses? Your belief in important eternal verities that you will much later very much regret having lost?
A sufficiently advanced technology is indistinguishable from magic. A sufficiently sophisticated propaganda is indistinguishable from truth. ‘Indistinguishable’ meaning ‘you can’t tell’.
Note you are reading to here.
***Some CS forensic historian of the time spent a lot of time analyzing how that could have happened. He almost got it right,as I will relate in a near-future chapter.