Saturday, December 17, 2016

WOW books

Books that put me at awe. That made me feel "Wow"

will be updated. work in progress

Fooled By randomness, Nassim Taleb
The 20/80 principle, Richard Koch
Private Truths, Public lies, Timur Kuran

Tuesday, September 27, 2016

Tradeoffs in negotiations don't work. "Loss aversion" doubles the costs

In a perfect world, tradeoffs are straightforward.
I give up on X, you give up on return on some Y valued similarly. Simple. Rational. Easy.

Loss aversion destroys this logic.

Loss aversion - Kahneman and Tversky 197x - shows that people value losses vastly more than their natural value.

I'll do vastly more try preserve my existing girlfriend, than to chase her has she been single. I'll ask vastly more for what I have than how much I would've paid to buy it.

Back to negotiations.
Suppose the UK asks the EU to give up on X (day freedom of movement), in return for whatever. Let's assume that the deal makes rational sense. (We ignore the strategic issues, have theoretic math etc)

For the EU, giving up on something is a loss. Losses are valued? Double their nominal value.
Hence, EU will ask the UK double the value in return.

Let's assume the required thing is money - for simplicity.

Say the EU says. Ok. No freedom of movement, but give £10bn yearly to the EU budget.

Now, for the UK this extra £10bn is perceived as a loss. Valued at double at £20bn

This explains why negotiating tradeoffs is so tough and usually just doesn't happen.

I am told that this idea has already been developed in a chapter by Kahneman and Tversky about conflict resolution, but I can't find the reference right now

Thursday, September 15, 2016

Two routes to rationality

Rationality can be arrived into by rationality. This is right, and this is why I'm doing it.

Some people reach the jacket of rationality for completely different reasons:

1) bad intuition / irrational nature.
Most people don't practice strict rationality. But there end up pretty much rational in multiple ways. Social learning / imitation. Or listening to some teachers. Or using life experience and emotional reaction to teach pretty much rational directions.

If someone is failing in some of the above, he might have to use aggressive "rational" rationality to get his life in order.

2) a tendency to be rational. Sometimes an obsession, sometimes loving it, etc.

NOTE route / outcome confusion.

I've noted elsewhere that we intuitively mix-up the way how people get somewhere, with its desirability.

This is a huge mistake.

However one ended up with the rationality habit, it's outcome is what matters, not why and how he got there.

(Outcome, of course includes experienced effects age utilitarian results!!)

Tuesday, August 23, 2016

Agency 1 - Stupidity alone is not the problem

My thesis is that the only reason stupidity can destroy the world is it being coupled with the agnecy problem.

Agency problem: when others have an interest different from us, and they decide for us.

We hear all the time about stupidity. And stupidity is rampant. But stupidity cannot explain all dumbness of the world.

Do stupid people not go to certified heart Doctors? they do.

Do stupid people buy smart phones that do not work? (or Windows phones that have no apps on offer?) of course not.

Society and markets have ways of filtering information.

The various phone sellers in order to keep their reputation, are not going to cheat too deeply on customers. The cellular companies know they will not get away with selling snake oil phones (they can come close).

Sometimes, the self interests of the agents is aligned with ours.
A captain on a ship will drown with the ship. A politician starting an atomic war might get obliterated in it.

But a money manager can reap his fee while the customer loses.
A politician elected in the primary method can appeal to extremists to become the party chief. Even if this will mean to destroy the party (It happens on the right and on the left).

Agency *is* the illness. Not stupidity alone.

IF the agency problem is solved, stupidity alone cannot kill us.

---- IMPLICATIONS ------

Too many. Those will be a subject of later blog posts. Many of them. You were warned

Monday, August 22, 2016

Is caring causing people to avoid efficient solutions?

People are packages of attitudes and opinions. And some are related. even if logically they are not.

Take caring about human pain and suffering vs. rational logical solutions (e.g. utility maximization, caring about side effects of solutions, unintended consequences, costs for others etc. )

Many people opposing parts of welfare things have as good a heart as those supporting them. Many times there are legitimate issues with the "just help the poor" attitude. 

Sadly, this all gets bundled up.

Is this an accident? or there is something inherent in caring that conflicts with rational utilitarian solutions?

I think there is an inherent bundling here. Not intentional. Many even deny it on both sides of the argument. But in fact, this bundling is central.

Bundling is when people hold bundles of opinions and attitudes, rather than discrete and seperate many opinions.

It will mean that a caring person will care rather than summarize things.
Then when someone offers - for example - to cut benefits for the slightly disabled to allow a decent living for paraplegics, the caring folks will always object. They are caring people, not utility calculating machines.

Moreover, when economists and policy people will point out the obvious calculation (the severely disabled are the main sufferers, etc.) they will be dismissed as cynics.

While irrational, they caring folks have a truth in their dismissal. An irrelevant truth, but true nevertheless.

You must be detached slightly to notice the calculation. The same way that an emotionally involved surgeon should not be allowed to operate. Being involved emotionally means it is hard to do the math. or stay cool while cutting someone brain.

This bundling of opinions, helps explain other wonders. Why are so many environmentalists opposed to nuclear power - arguably the most realistic way to stop CO2 emissions? Simple. If you care about the environment you hate  pollution and nuclear power is a pollutant. Calculations are irrelevant. 

Tuesday, August 9, 2016

Should science funding be random?

Some studies looking at the current system of science funding find it quite useless.

Instead of having endless people writing grants, reviewing grants and the whole torture, we will fare better by simply allocating the grants randomly based on super simplified and effortless criteria.

This kind of insight - while intuitively offensive to many - is a recurring robust finding in many fields.

Money management: Index funds outperform managed funds. Very robust long term data. Most mutual funds just cheat you by taking the fees for useless "active managing"

Interviews and student assessments: Simple numerical models outperform most personal thinking on how a student will succeed henceforth. Those time wasting procedures not only are wasteful, they actually lead to worse results than a simple formula.

The list goes on. ....

BUT.... one thought occurred to me.
At the tails, it is possible that the current system is superior.

When an idea is quite clear to have very good chances to succeed, the assessment system might almost guarantee it funding. Which a random system will not.

So, for some aspects a managed system might help

Of course, we can adjust for it. Allocating 75% of the funding randomly, and 25% to "must do" projects. 

Monday, August 8, 2016

rating online comments. And cross sites

Comments online is a hugely missed opportunity.

Many times adding a comment can elucidate a point, or add a different view.

but... lots of junk.

Two features can fix this. and on the way add lots of value for all.

1) rate commentators.
I want to know that commentator X is a PhD in physics, or is generally smart etc.

2) sort the comments by value

ReddIt does comment rating and sorting. But it is only by direct voting.

I want global rating. Very similar to the way Google rates sites. A recursive function will also take into account the rating of the raters etc.

The added value is enormous:

A, comments being useful to read. rather than a huge pile of chunk. The top rated comments will usually be more useful.

B. Dynamically this will lead to both more comments and more comment engagements.

C. Of course this will also reward high value commentators to comment more.

Lets hope someone picks this one up.