A few things that I think are pretty interesting ideas that may end up being blog posts or may end up as a tweet (or never leave this page)

based on George Soros’ reflexivity lecture. The systems we are in are reflexive to differing degrees. To me the financial system is not very reflexive at all. I can shout at the top of my lungs the S&P is going to rise or fall and it will likely have no impact. If Soros or the president of the USA says the same thing then what they claim may come true partially due to their very saying it. One measure of power in a system is therefor the degree to which that system is reflexive to you.

The optimum level of government waste is not 0. With an organisation as large and complex as the government the effort and cost to eliminate all waste would be huge and most likely outweigh the benefit from doing so, i.e there is diminishing marginal utility and increasing marginal cost to stamping out waste. This isn’t to say the problem shouldn’t be taken seriously but that when we see articles about someone scamming the government for welfare checks that doesn’t mean the system is fundamentally broken but that it may genuinely be too costly to stop everyone from doing so. This likely scales to other organisations, large companies etc.

Knowledge and ignorance must grow together. It has been written that Thomas young was the last man to know everything and he died in 1829 so there is little hope that we can even come close. Many fields are growing more esoteric and their findings all but inaccessible to the general populous we are coming to necessarily rely on the voices of a few to tell us about the world around us. This is not necessarily bad in of itself but it means those few hold much more power.

It is culture that shapes us more than morals, we eat battery farm animals but don’t walk around wearing capes.

There is a common saying that the absence of evidence is not evidence of absence. However, from a Baysian perspective this simply isn’t true. A lack of evidence is more likely under the null hypothesis and so if you don’t have evidence then you should revise this probability upwards. It seems like the correct saying should be absence of confirmation is not confirmation of absence. There are situations where maybe updating like this would be a bad idea, e.g fat tailed distributions but on the whole it seems like the common saying is wrong.

Interesting/unusual questions to ask someone:

  • What their favourite museum rooms are
  • Do people have a personal cannon of books that they think represent many of their core ideas or what they aspire to
  • What was your last Amazon purchase
  • When was the last time you felt pure delight
  • What thing under £15 do you really like
  • Do you think you could sit in a room for 30 minutes without doing anything and be happy?
  • From Devon Zuegel
    • what has been really exciting to you recently
    • There is a quote “all thinkers are regional thinkers”. How has the place you grew up shaped your thoughts?
    • what is your most absurd belief
    • what is something that people misunderstand about X
    • what is a silly thing you did in the past year that you are most proud of?
    • what is a belief you have that is the closest to a conspiracy theory
    • Ask a borderline nonsensical question to see how people interpret them
  • What are the best arguments against the principals of your world view
  • what is something you have changed your mind on in the last year
  • How would you change your life if you were given $10m
  • what can you teach me that most other people can’t
  • argue something clearly wrong/present it as your belief and see how well the other side argues, both in content and conduct
  • Is there a topic you are excited to talk about that most people find boring
  • Most interesting thing you have made
  • can you recommend something for me to read that I am likely to enjoy and not have come across before
  • do you use any apps/products that you think are really good but that you also think most people will never have heard of before
  • what person around today do you would you most like to have a conversation with that you haven’t been able to
  • what is the best gift you’ve been given

To what extent do young people at uni fall into mimetic desires for things like banking, law, consulting. Is this a bad things and to what extent can it be changed.

A study that finds men are underpaid after controlling for performance may actually be revealing a bias against women in performance measures. Google

It’s often said that in a gold rush you should be the one selling shovels. What are some of the best examples of this. There are two sides to this really, the first is capitalising on a fad very well that really is only a fad. The second is about generally selling the facilitating devices to some fast accelerating sector e.g NVIDIA with GPUs seems like a reasonable example of this.

Looking more into trends of incremental changes in products e.g better apps vs paradigm shifting technologies like the computer. Where paradigm shifts really as quick as we might have as a mental image. The computer has been around in some form for over 70 years could it be argued that it was actually an incremental technology rather than paradigm shift?

(From Tyler Cowen) what are the effects of local vs global diversification. For example many high streets are getting more diverse than they used to be but if all high streets are doing this then they may well be becoming more like each other.

A look at why many of the main concerns of startups didn’t end up to be true, for example:

  • People won’t use Airbnb because eventually someone will get robbed or assaulted
  • People won’t use Uber because eventually someone will get robbed or assaulted
  • Amazon will fail as Ebay’s model of using auction mechanism is the future of pricing
  • There’s no money in search so let’s pass on Google

    There are likely ‘complex’ reasons why many of these things turned out to do well but there may be some underlying trends.

A look at a few economics terms/concepts I think anyone would find useful to know about e.g principle agent problem, separating equilibrium, opportunity cost, tax incidence

A look at how new product features have been found through the process of looking at how users are hacking the existing product. e.g Instagram saw that lots of people were posting their snap codes on instagram and so realised stories may be a feature they wanted to add.

Are some modern companies more similar to city states or large ancient companies. This is a run off from a podcast episode with Kevin Kwok but also prompted by Joanna Bryson’s comment that Google should be treated like a world power.

Why do we seem to help others differently than we help ourselves. I myself when I hear that someone has a problem may be happy to google around for a bit even without being asked. I find that often I can find something in this time the other person hasn’t found in a bit and yet what I did wasn’t hard. There are also times when the roles are reversed, I am having a hard time doing something and it doesn’t take much effort at all from the other person to help me.

Questions I think it would be interesting to ask in an interview (note I have virtually no experience in this area). This list has questions I think it would be interesting to get asked by the interviewer and questions I think it would be interesting to ask the company. Patrick McKenzie has a lot of good stuff on this and other topics on twitter.

  • If I flipped a coin and it came up heads 4 times in a row what would you bet it would come up as next time. Sort of a fizz buzz test to see if people can quickly jump out of the rigid world of probabilities and implicit models of fair coins.
  • Who’s your enemy. Many companies have/seem to have an enemy they are fighting Coke has Pepsi, Spotify has itunes, one could even say companies like Deepmind have technological stagnation as their enemy. It seems like having an enemy is a powerful motivating force and will likely reveal something about the company in question.
  • What are the top 3 problems your company is trying to solve
  • When was the last time you openly disagreed with your boss
  • What were the most difficult problems you faced and how did you solve them
  • If you had to compare your company to a past company which would it be?
  • Legibility, does everyone at the company understand what is important and what loops matter
  • Do you use Meehl patterns when hiring. Given that hiring seems like a process with a high level of uncertainty one potential approach could be to follow a similar route that Danny Kahneman did with the Israeli air force and use a simple test rather than ‘professional’ judgement. This seems like it would be better than many hiring processes in some ways but likely falls far short of the best hiring processes.
  • Do you have a formal process where you review the mistakes that were made over a given time period e.g Sequoia anti-portfolio

Not all conditional probabilities equal mediocristan Vs extremistan

Look at comparison to stripe and Bell labs. Both companies are trying to create infrastructure for a massive system of information (payments and information respectively). Could stripe learn from some of the practices of Bell labs e.g doing more basic research or was the monopoly position of Bell integral to this being the best strategy to adopt.

A look at the benefits and shortcomings of using metrics in a company and in a society.

  • e=mc^2 would never have been found by someone trying to maximise some metric it likely had to come from free curiosity
  • In AI we can get specification errors where the system will maximise something in a way you didn’t intend e.g programme pausing game so that it has a high score for the longest time or workers in amazon getting to create their own production function trying to pick it very cleverly rather than just getting on with the work
  • Issue of not pairing indicators (Andy Grove)

Adding human knowledge to Alphago made it worse, we think we have things figured out but clearly we are weaker than we think. How many other domains have we been going down a path that is suboptimal? may it be the case that we are actually still doing what is optimal it is just we are optimising based on our intellectual capacity and a computer is optimising based on its (much larger) capacity?

A look at best practices/principles from various industries that seem to have some applications to other industries but that maybe aren’t used much.

  • Military
  • Game design
  • Medical e.g mortality conference
  • Pilots