Goals and carrots [WIP]

On reflection, it currently appears to me, that the source of my motivation is beauty which comes quite close to something like meaning. In mathematics, beauty has long been identified as a lighthouse guiding search in high-dimensional darkness. Then meaning is something that beauty can acquire by being applied in the real world for its own sake.

It is straightforward to identify a number major challenges that all of us humans are facing on a global level, with all of us in it together in a joint future. These challenges are of ecological, social, and economic nature and are characterized throughout by complexity, a thing which we are just able to identify, and just beginning to understand and take serious, although with a long history of ideas attached (Refs: Gell-Mann: The Quark and the Jaguar, ). This understanding should be available on a collective level to inform policy making. Policies are recipes for action put down in a table. Each situation that can be encountered should have a line in the table and the corresponding entry in that line tells you what to do in this situation. Sound knowledge of policy making should be available at all levels of human activity, starting from individuals, out through families, organizations, communities, regions, countries, up to planetary scope.

Ignoring complexity issues usually leads to any of a wide array of possible fails with quite variable impact (Refs: Clausewitz: the essential unpredictability of war, the fog of war, “friction” – the disparity between the ideal performance of units, organisation or systems and their actual performance in real world scenarios (Book I, Chapter VII), Dörner, 1997: The Logic Of Failure: Recognizing And Avoiding Error In Complex Situations; Taleb, 2008: The Impact of the Highly Improbable; Trivers, 2011, Deceit and self-deception: Fooling yourself the better to fool others; tbc). Classic examples include the release of rabbits in australia, the Cobra effect, and so on. Apart from particular technical approaches to immediate concerns, e.g. automated driving, education, elderly care, environmental cleaning jobs, and so on, AI surely has cooperative potential for augmenting human capacity and provide systematic contribution to policy making at all relevant levels of small, medium, and large, e.g. the individual, communal, regional and global levels.

We would like to contribute to peeling out the cooperative and augmenting aspects. As a team we are facing several complex problems, for example the self-management of a geographically distributed team. Since this is a really big one, it cannot be expected to be “solved” anywhere soon, so it’s necessary to default to effective heuristics and some glue to build approximate solution. I will start by listing and briefly describing some heuristics that appear promising to me, touching on the following items: open culture, transparency, empowerment, local autonomy, self responsibility, attention, credibility, past / future horizons, variability, local minima, optimality in non-stationary environments.

The first one is openness. In my experience, this is of fundamental importance and I have become a proponent of radical openness, but I am aware that the appeal is not always intuitive.

The basic argument is about the cost of closedness (see also and expand: hidden cost, true cost). This cost is just not considered in closed approaches, which clearly is negligent and careless. We have to be aware that we are working with knowledge, and the beginning of knowledge are ideas. Ideas are not isolated, standalone entities but have to be considered on an ecological level. They are born and live in an ecosystem of knowledge and ideas, and critically depend on each other. The classical quote about “standing on the shoulders of giants” 1 is just about that. This means, we depend on access to ideas of others, and fostering the ideas of others depends on sharing your own ideas. The actual success of any particular enterprise hinges on a few very small detailed modifications on knowledge widely available to everyone else (FIXME: history of innovation with examples).

Don’t worry about people stealing your ideas. If your ideas are any good, you’ll have to ram them down people’s throats. – Howard Aiken 2

This is fully coherent with my own experience, allowing the vanity of having had at least one good idea at some point. Keeping things closed and proprietary thus incurs large cost. There is the immediate cost of securing access to code and documents, the cost of complicating publishing and deployment, and the longer term cost of impeding evaluation and weeding out of bad ideas through discourse with others. In the end, “closing” things in a complex world is impossible, and basically it amounts to “obscuring” things. Then we know, security by obscurity does not work in the long run.

Security experts have rejected this view as far back as 1851, and advise that obscurity should never be the only security mechanism. – Wikipedia 3

Reinterpreting security as a kind of robustness to “attack”, a good heuristic can be gleaned from biology.

Work-in-progress [WIP]

Distributed repesentation / biology: not one organ, not one function (Monod?), Empowerment: empower others to elevate yourself in their wake?, Transparency, local autonomy, responsibility, Team of teams, Uncertain future. Horizons. Discount factor. n-step. predictive information. local minima. variability, discount factor 4, Attention / awareness G. Franck, Credibility (Trouble: F***‘in with a real one, no quests)

More refs:

  • Cobra effect
  • The myth of the objective, usefulness of useless knowledge
  • Team of teams
  • Economy of attention
  • The quark and the jaguar
  • Die Logik des Misslingens
  • Black swan
  • Self-deceit



1 Newton?

2 https://en.wikiquote.org/wiki/Howard_H._Aiken

3 https://en.wikipedia.org/wiki/Security_through_obscurity

4 https://stats.stackexchange.com/questions/221402/understanding-the-role-of-the-discount-factor-in-reinforcement-learning

tag cloud

robotics music research sound books jcl startups intelligence feed ethology computational audio AI psychology models jetpack organization motivation micro management funding dsp brain algorithms testing test synthesis sonfication smp scope risk principles musician motion mapping gt generative exploration digital decision datadriven computing computer complexity aesthetics wyafinding visualization tools theory temporal sustainability stuff sonic-art sonic-ambience society signal-processing sequence self score robots robot-learning python pxp priors predictive policies philosophy perception organization-of-behavior open-world open-culture neuroscience networking network navigation minecraft midi measures math locomotion linux learning kpi internet init health hacker growth grounding-fail grounding graphical gaming games explanation event-representation epistemology embedding economy discrete development definitions cyberspace culture creativity computer-music compmus cognition business birds biology bio-inspiration android agents action