Wednesday 31 August 2016

What are collective action problems?

Who's a what now?

Two criminals are caught by the police and interrogated in separate rooms. Each woman can either confess or stay silent. If both stay silent, there isn't enough evidence to charge them and they are indicted for a lesser offence. If one confesses, she can cut a deal but her partner goes to jail for a far longer time. If both confess, neither gets the deal and both are locked up for a long time.




The best course of action for either woman is to confess because no matter what her partner does, confessing leaves her better off. The problem is that if both women confess, they are far worse off than if they had kept their mouths shut. In essence, cooperation gives the best results overall but defection is optimal for the individual. 

Collective action problems are essentially scenarios where individually optimal actions lead to a worse overall outcome. You can conceptualise of them as multi-actor prisoners dilemmas, situations where the best course of action for everyone requires individuals to take actions which are not best for them. i.e: Fishermen have an interest in maintaining fish stocks which requires that they limit themselves to catching a certain number of fish per year, yet as there are so many fisherman the actions of a single boat will not affect the overall fish stocks to any significant extent. Hence fisherman have an incentive to and do engage in extensive over-fishing despite the fact that doing so destroys their industry.

Why do I care?

Collective action problems are interesting. They're a decent argument against Libertarianism or unfettered free markets generally. After all, the reasoning behind markets is that disparate actors maximising their own outcomes tend to produce socially optimal results. Not the case when it comes to CAP's (or any externalities for that matter, an externality being a transaction which effects third parties). They're also a prism through which to view certain social and political processes, especially dissent, ideological centripetalism and generally the way ideologies, especially totalitarian ones, tend to spread,

Monday 15 August 2016

Eternity & Recurrence

Assuming time goes on forever, what shape does the world come to take?

Do all possible things come to be?

Does the world reach a dead end eventually, a static place where change ceases to occur and the present becomes eternal?

Does the world begin to loop? If yes, is this really eternity? Is the same patch of video repeated over and over again an infinitely long video?

The impossibility of immortality

What is life?

A person is a process. As long as the process runs, as long as it receives data and changes itself and/or its environment in response, it is alive. If that process stops running, either temporarily or permanently, it is not alive. A permanent stop is what we would term death.

Here's an example. I upload your mind to a computer. I run you in a simulated world. You exist, you feel and think in the same manner as you do now. You are alive. If I freeze the simulation, put it in sleep mode where the data and current state of all things, including you and your thoughts, is preserved. Are you alive? You don't think. You don't feel. Hence I say no. The fact that a backup of you exists which, if run, could produce a live you does not change the fact that right now, there is no active version of you running and you are not alive.

(N.B: No you're not unAlive when you're asleep. You're still running, albeit in a different mode.)

What is immortality?

Living forever. Not for a long time. Not for a very long time. Not longer than the universe is likely to exist. Forever. Infinitely. Without any end.

Trivial Impossibility: The speed of light as a constraint on computational complexity

The speed at which causality propagates is limited. This limitation is commonly known as the speed of light, although that is a far too specific name for a more general law. (I think). One implication of this is that there is a limit to the strength of computers we can create, even assuming we had access to unlimited resources. Why? Simple. A computer must transmit data from one place to another. From memory to processor or from processor to processor. Why is this true? Simple yet again. Even assuming we can overcome current limits which require us to separate memory and processors etc..., as long as processors or components have some mass it will be impossible to pack too many in one location lest their combined gravitational pull destroys them or causes a black hole to form, Back to the need to transmit data. As your computer gets larger, the lag/latency arising from communication gets more and more extreme. A Dyson sphere around our sun has a transit time of more than 1.6 second for data to reach from one edge to the polar opposite point. A computer the size of a solar system has even greater latency. Of course you can design around this to some extent with distributed systems, but there's only so much you can do. Past a certain point, increasing the size of your computer will cease to bring increases in strength as the limiting factor becomes not processing power but latency/bandwidth. The law of diminishing marginal returns applies.

1:Past a certain point, you can't make your computer any faster.

The second problem is that older life is more complex life. A living process/person builds up memories of past events. The longer their life, the more memories. The more memories, the more work the computer the person is running on has to do to store those memories and access/integrate them into present decisions/processes in a timely fashion.

(N.B: This assumes processes which do not ignore old memories aka don't have a cut-off point. i.e: possible that alien life integrates memories into basic algorithms determining decision making processes before discarding the rather than keeping them for reference as we do. For a number of reasons, I believe this memory-discarding type of mind is highly unlikely to exist or be comparable to us. If you disagree, assume this article is specific to human minds.)

2: The longer you live, the more computational strength it takes to run you

Hence the conclusion. Eternal life is impossible, at least if the laws of physics hold and my argument is sound.

Serious impossibility: Life as change

Life is change. Life is a process which takes in information and changes in response to it. All your experiences change you, change who you are.  When you gain knowledge, that is change. When you begin to think differently, when you become wiser, that is change. When you love someone and let them take your heart, that is change. Living is accepting change. Without change, there can be no life. This is true in the narrow sense as even forming new memories requires a change in the data you have access to. It is true in the broad sense in that experiences change how you think.

The longer you exist, the more you experience. Every experience changes you and eventually, after enough has passed, you are changed so much that you are no longer anything like your original self. The pattern that now exists traces it's history to you, but it is too different from you to be considered you.


Tuesday 2 August 2016

Shallow vs Deep Persuasion

Part of the persuasion sequence

***Bullshit alert: This may well be bullshit. Expect unempirical distinctions, ungrounded assumptions and fuzziness all round.***

Shallow or ordinary persuasion changes opinions by tackling the surface of a persons psyche. You convince then of facts. You get them to grudgingly accept your point of view or fully accept it for a while. The problem with this is that people are not rational creatures. This is true of all of us, and more so for most of us. People want to believe certain things, things which fit in with their world view or that of their tribe. Reds want to believe that crime is a product of degenerate sub-cultures, that their own culture is superior. Blues want to believe that crime is a product of oppression and that all cultures are equal. If you challenge facts, if you challenge individual beliefs without challenging the worldview into which they fit, it is very difficult to affect lasting change. Inevitably, people slide back towards the easy path. Most minds are too small to hold two horizons. When ideas conflict, the lack the ability or understanding to accept the paradox and walk down both paths. Rather, they choose one and it is usually the one which they have walked down further already.

To really change a persons mind, you have to change the person. You have to change their entire belief system and in doing so change their conception of self. This is deep persuasion. Deep persuasion is more stable. People don't revert to their prior beliefs. They don't need to. The tension between natural beliefs and the new beliefs which don't really fit, the tension which ultimately undermines so many attempts at shallow persuasion, doesn't exist.
















Afterthought: A fear of deep persuasion.
I wrote this piece, but I didn't want to. What I wanted  to write was an explanation of why good enough persuasion is truly terrifying. Why people think that, even if an AI persuaded us wrongly, we could realize what lies its persuasion was based on. You wouldn't. Good persuasion changes your beliefs, your basic axiomatic moral beliefs. Even if your factual beliefs don't change, even if you realize you were persuaded, the new you dosen't want to revert to the old, immoral/stupid you and has to stay where they are, even if it's with a biter taste in their mouths.

The only solution I can see to this is establishing clear Schelling fences, but that sticking to them requires an inordinate amount of willpower.