Sequel to my previous post, which was discussed at https://hubski.com/pub?id=133996. I'd love to hear what y'all think, especially mk and b_b.
You should post more often. Love it.
- On the right most of the stimuli are internal, the process of evolution is non-linear and hard to predict. It’s doing its own thing, all you can do is try to nudge it one way or another and wait, often without any certainty about consequences.
Another way to view the spectrum is as an answer to a question: how likely were the ancients pray to it?
I gotta say, that transition had me perplexed. I definitely experienced a moment of, "Where the hell is he going to go with this?" You followed through and delivered an interesting approach to control and interaction.
- Growing kettles instead of designing them would obviate our worries about pentagons, constraints and alternatives. But how would it work? Imagine if you could easily create new random kettles in large numbers. You already have a list of desirable characteristics that you can score each of them on. If you selectively mix the characteristics of only desirable kettles you would be in effect breeding them, and repeated actions of selection over time would cultivate a wild crop of kettles into something more tame.
I don't know if you saw this post - I have been watching this genetic car algorithm for an hour now.. It was my first encounter with genetic algorithms in a way I could really understand. The randomly generated cars take off down a track and the one who makes it furthest (the "elite clone") continues over to the next round. New cars are mutated and the race is repeated. Like your idea with kettles, and using mutations instead breeding, this little game demonstrates how we can arrive at the "perfect" design for a specific circumstance given enough time.
You should also check out TNG's recent post regarding micromanagement. It's a bit more of a candid look at micromanagement in chaotic, real life situations. As much as we can analyze and explore control, management, and interaction of individuals and objects, throw a real life boss in the mix and everything gets much more complicated. :)
Thanks insomniasexx. I had two different people tell me in drafts to drop that paragraph, but I just couldn't kill that particular darling :) There was also a version that put all of it in parentheses. Perhaps I should have stayed with that..
This is likely a more erudite audience than I was envisioning when I wrote it, and so the genetic algorithms stuff probably seemed overly belabored. What do you make of my attempt to apply its lessons to human groups, especially the "utopian participatory panopticon" thought experiment at the end?
- This is likely a more erudite audience than I was envisioning when I wrote it...
As opposed to the morons who read Ribbon Farm? Kidding, obviously :)
More seriously, neither top down nor bottom up can lead to optimal solutions, but for vastly different reasons. Top down control we can reject out of hand for all but the simplest of systems, because it is impossible to define a set of regulations and rules that can account for all complexities and contingencies that are inherent to a society. Your three strikes example is illustrative in that it's ostensibly a simple rule made to have simple results: That is, rational actors will not continue to commit crime in light of the cost benefit trade off of marginal personal gains vs. a long prison sentence. It utterly fails to recognize that humans aren't Homo economicus or even something resembling such. Thus the effect of such laws is prison overcrowding, sentences that are draconian compared to the crime in many cases, and drug addicts being sent to prison for essentially harming themselves. The consequences are bleak and apparent, though no politician has the fortitude to do anything about it, due to the system stability that is our voting method (winner take all elections that serve to embolden the extreme and harm free expression).
Bottom up has a very different problem, which is that it is necessarily heavily dependent on historic contingency and initial conditions. For example, suppose a bird is born with a mutation that allows it to fly higher, faster, farther while expending less energy. This bird, according to the theory of survival of the fittest will pass on her genes to her progeny and so on. However, what if this bird were born in a year with a sufficiently harsh winter that her mother dies, and she starves before she can realize her potential? Her gains shall never be realized. Alternately, and perhaps more illustratively, she is the second born in a clutch of two, and the species has already evolved to only feed the first offspring, the second being a kind of biologic insurance policy (and this is a real thing in some bird species). The stability that is built into the system will kill her chances (literally) of effecting change.
We see the same thing in society. When the energy well is sufficiently deep, it is impossible to effect change in an optimal direction, because doing so would either be too disruptive to the current system, or the change isn't part of what Stuart Kauffman calls the adjacent possible, which is to say systems can only move from one state (I mean a thermodynamic state, to be clear) to a state that is very close to it (one can jump from two to three, or two back to one, but not from two to five, say). I suppose my point is that even when starting a system from scratch, we are still limited by the initial possibilities (which are finite), as well as by the initial/boundary conditions into which the system is born.
As an interesting aside, I think Obamacare may be a good example of a governmental attempt at growing a system, and how that can be completely derailed by entrenched special interests. The theory behind the ACA is actually a small government, conservative idea. The idea is that there are three "pillars" that must be intact to support the system. One, that coverage is available to everyone. Two, that coverage must be bought by everyone. And three, that subsidies are available on a means tested basis so that poor people aren't unnecessarily burdened. These three rules of the game are supposed to set up a competitive marketplace whose outcomes will be competition and the elimination of the free rider effect. All of the law could be written concisely and clearly. Instead, what we have is a multi-hundred page behemoth of a law that for many is a stand in for government waste, bloat, and cronyism, all extras ostensibly for 'cost control', while unsurprisingly benefiting the very folks who "helped" craft the law (insurance companies!). I have no idea if the ACA would have worked on its original premises alone, but forgetting whether it works or not, it's a great illustrative example of a push-pull between bottom up growth and top down control.
So in the end, although neither is optimal, I'll take bottom up the majority of the time, given that it allows for human ingenuity much moreso than does bureaucracy. mk is fond of saying that he made Hubski just to see what would happen. Give people the rules of the game, and then keep your hands off. I think it's working so far, but I try to remind him and others about that every time changes are being contrived: will this change limit or expand possibility? If it is limiting, I'm usually against it, maybe for no other reason than to be contrary.
Yeah, exactly. Bottom-up allows humans to see what is possible. But bottom-up also allows some humans to be top-down towards others :)
I'm not too concerned about the limited degrees of freedom. We're human and we have our limitations, no point worrying about that. In programming terms, I just want us to avoid regressions. It's wasteful of human lives to spend time changing states to one that we already knew long ago had problems. Avoiding repeating history and all that.
- As opposed to the morons who read Ribbon Farm? Kidding, obviously :)
Oh, I was totally serious. I wasn't sure how much to assume regarding familiarity with genetic algorithms and so on. At least the people I know at hubski wouldn't need too much belaboring of those properties.
(BTW, do we have some way to do nested quotes? Has that come up?)
Also, nested quotes haven't come up as far as I'm aware. forwardslash did a lot of work to redo the markup, but I don't think that was part of it. Maybe he'll jump in an enlighten us.
- I'm not too concerned about the limited degrees of freedom.
Me neither. I only brought it up because there's a pervasive cultural myth that evolution leads to optimal solutions. It doesn't. It takes what is there and coaxes it to extract resources more efficiently. Therefore, there is a case to be made that bottom up with strong regulation (top down) may be advantageous. That is, when presented with a choice between two extremes, it's often the middle way that should carry the day.
I was explicitly trying not to be ruled by our preconceived notions of what it means to own. Given that ownership of software isn't quite how we used to do it, are there benefits to changing the rules around a little bit elsewhere as well?
Basically, our planet is getting crowded enough that giving people so many rights bundled up in 'ownership' is starting to crack at the seams. In my opinion :)
I think a price tax would be incredibly useful in the domain of software, and IP generally. A home is different from any other investment in that we live in our homes, so they're so much more than just a credit and debit sheet. However, a tax on derelict IP would be amazing. Patent trolls would be doomed, and rightfully so. Perhaps the tax could be deferred for a period of time for businesses that are actively developing a revenue stream off their idea, because I could see it being burdensome for a startup with limited resources. However, it could do wonders for innovation.