Unpredictable Patterns

Contact

  • Re-imagining magic

    +

    +

    +

    +

    +

    +

    I am reading Chris Gosdens remarkable The History of Magic: From Alchemy to Witchcraft from the Ice Age to the Present (2020) and already in the beginning there is an amazing remark that I think is worth thinking about. Gosden retells the story of the Azande, as studied by Evans-Pritchard, and how they viewed magic. If someone happened to be crushed by a granary collapsing over them, they would readily agree that the granary had collapsed because of long neglect and the lack of any fixing, but they would as readily point out that there was a reason it collapsed at the time it did and killed the man it killed.

    This reminded me of the fact that we simplify earlier beliefs at our peril. The Azande are not dispensing with causality and science, they are just adding a layer of causality – the direct cause, the neglect, is recognised but only as an intermediate cause. The direct cause was somebody cursing or harboring bad will towards the person who died.

    The Azande belief in magic is a belief in a much more complex model of causality than ours, positing extra causes where we rely on Occam’s razor.

    But that view of causality may be much more interesting than our very simplistic view. What if there are separate causal chains that lead back to someone disliking the person, perhaps “cursing” them? Why would that be less of a cause just because it is not a direct cause? We need not believe in magic to recognise that the reality we live in is a complex causal web.

    Magic, then, is an invitation to understand the many causes that create our future and reality, and as such it may end up being quite useful.

    Why do things happen? There is no simple answer to that question and if magic is reimagined as the recognition of complex causal webs we would do well to examine its premises more closely.

    +

    +

    +

    +

    +

    +

    + ,

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • Valuing time (Mental Models X)

    +

    +

    +

    +

    +

    +

    How do you value time? There is a series of questions related to this question, but the biggest of them all is an existential one: what should I do next, how should I spend my time?

    Time is all we have.

    When we value our time the first challenge is to choose the resolution, the unit of time that we value. This will turn out to matter. What usually happens is that we value time in the units we measure it:

    • We charge a price per hour. Surprisingly many people still do this and charge by the hour. This means that you value the hours you have accessible to you and try to price them. If you are expecting to live another 10 years, the total amount of hours available to you are – roughly 16*365*10 = 58400 hours.
    • We can also value our time in days, weeks, months and years. Months is interesting, since we still accept that our salary is set in monthly installments. This works in a couple of different ways. You can calculate what you are being paid by the hour and use that as a market value of your time – if you have a monthly salary of 50 000 SEK, you could say that you are getting the equivalent of 50 000 after taxes / 160 (a “man month”) = 312 kr an hour. Now, this is a bit muddled by retirement funds, insurance etc, but you could say that at that salary you are getting that order of magnitude of remuneration for your time. I can buy 10 years from you at that monthly salary at the price of 6 MSEK. Now, interestingly many people price-adjust: they look at their salary and adjust their work input to a point where they think the equation is more fair – they don’t work as hard or bring all of their creativity to work – but that work and that creativity risks being just withheld, not utilized elsewhere.
    • Other options include valuing our time in absolute terms – that is in our life time. We could say that the average person lives 80 years and the real question is what it is worth buying a portion of that. In reality noone is buying absolute units of time after all, they are buying relative portions of your life. So what is 1/8th of your life worth to you? That is the price you should charge for 10 years. Here people often get befuddled and start thinking that they are buying their own life – they work now to be able to retire later and then spend their own life. But the reality is of course that they are selling their life at a constant price, and are hoping that this means that they can refrain from selling portions of it later.
    • Another relative measure is what you can create. Say that you can test a reasonably complex project in 6 months to see if it flies and works and if it has value — this is optimistic, but this is sort of a minimum viable time to test something. If that is the case, 10 years represents 20 projects. How should you price projects? That is what the question becomes if that is how you measure your time and charge for it. How do measure the chance to write a first draft of a novel, start a company, learn the basics of an instrument or a craft? What is the list of projects you would try if you forced yourself to measure your time in this way?

    All of these methods are provisional, since they are ignoring the fact that the value of life goes up in relation to how little of it you have left. We have all read about the people who completely revise their time plans when they survive an accident or get a diagnosis that sets a definitive limit on the time they have. Yet, we seem unable to implement the insight they had without a scare of our own. That has always seemed peculiar to me – we can simulate so many other things – why not simulate surviving a plane crash or getting a diagnosis of cancer? And how would you then value your remaining time?

    In a sense these events have been dramatized to a point where we are kept back from seeing the real insights in them: the stories around survivors or those diagnoses with deadly diseases hide the fact that the equation they solve for time / value is the same equation we all face, just seen from a different angle. Maybe it is not too hard to understand, maybe it has to do with fear – as so many things in life do. Your fear of death at some time passes your fear of not being successful, rich or powerful enough.

    It requires a special kind of courage to be time honest to oneself. To honestly assess what we are doing and change if it does not respect the time we have left – not left to sit on a mountain and watch sunsets (though that is nice), but time left to build, create and explore. Time to learn. It is not that we should stop spending time, it is rather than we should spend it differently. How we can spend it so that it compounds.

    For Rilke work – the work – was the imperative, following Rodin.

    I struggle with this, more and more lately. The pandemic is one reason, it forces a re-examination of the way we live our lives (we who are so immensely privileged that we can act on it). It whispers the Rilkean admonition to us:

    Wir kannten nicht sein unerhörtes Haupt,
    darin die Augenäpfel reiften. Aber
    sein Torso glüht noch wie ein Kandelaber,
    in dem sein Schauen, nur zurückgeschraubt,

    sich hält und glänzt. Sonst könnte nicht der Bug
    der Brust dich blenden, und im leisen Drehen
    der Lenden könnte nicht ein Lächeln gehen
    zu jener Mitte, die die Zeugung trug.

    Sonst stünde dieser Stein entstellt und kurz
    unter der Schultern durchsichtigem Sturz
    und flimmerte nicht so wie Raubtierfelle;

    und bräche nicht aus allen seinen Rändern
    aus wie ein Stern: denn da ist keine Stelle,
    die dich nicht sieht. Du mußt dein Leben ändern.“

    Rilke, Archaischer Torso Apollos

    You have to change your life. And the only way I know how to do this is to change the way we spend our time.

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • Roam – some thoughts

    +

    +

    +

    +

    +

    +

    One of the things I have come to feel is increasingly important, is to organize some sort of personal knowledge management. I have a number of different tools that I have been playing around with including Notion and Bear. The challenge is that I have essentially three things I want to solve for my in my workflow – one is writing, I want a dedicated writing tool and there design has turned out to be surprisingly important. I have ended up with Bear as the best and most interesting alternative for writing texts intended for consumption by others. The second thing is storage — a place where I can store texts, ideas, presentations, information – and this is where Notion seemed very tempting and secure. But there is something about Notion that I cannot quite put my finger on that makes it a hard tool for me to work with. I can make pages and throw up documents and…but the feeling is that I have a very large attic where I can store things, but where they essentially do not compound.

    Enter Roam. A wonderfully quirky and interesting tool that allows you to build knowledge graphs. It is experimental – probably less secure – and forces you think about the knowledge graph you want to create and how to navigate within it, but as you start to learn how to do that it is surprisingly generative. It builds structures and let’s you explore them – some of the structures are not just storage, but devices of discovery as well. It is really interesting.

    And another thing – it is oriented around today. This may be the most genius move in all of the Roam conceptual framework. I open my graph, and then just start a new day, start logging things, reminding myself, thinking things through — Roam is a perfect tool for thought writing or meta-cognition – the kind of writing you do to find out what you think and to generate ideas, the kind of writing that is really a kind of dialogue with yourself.

    So currently I am thinking Notion for pure storage of things I need to have as resources, Bear for writing and Roam for project management, thinking and exploration. It is a pain not to be able to do it all in one single tool, but this is where I am now – I will write an update later and see how this goes.

    +

    +

    +

    +

    +

    +

    + ,

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • Regulate Tech #5: The Truth!

    +

    +

    +

    +

    +

    +

    New episode of Richard Allan’s podcast. All feedback welcome.

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • A weird lesson from prozac-infused fish

    +

    +

    +

    +

    +

    +

    Scientists report that the residual amounts of drugs, like Prozac, are affecting the fish that live in waste water in weird ways. Here is a report from ScienceAlert:

    To probe further, Polverino and his team ran a two-year experiment in the lab, subjecting generations of guppies (Poecilia reticulata) to targeted concentrations of fluoxetine at low levels, on par with the chemical’s pollution in aquatic environments, as well as high levels.

    The results were stark: the fish on antidepressants seemed to lose their capability for individuality as a result of their exposure, with variations in behaviour between separate animals diminishing as the dose got stronger.

    Basically, they all started acting the same.

    So this is interesting in a number of ways – it reminds us that second order effects of human actions (or third order) are affecting the world in a myriad of mostly unexplored ways – and argument for increased uncertainty. But it also forces us to ask the question of what happens with a group that starts behaving similarly – and this is where the really interesting results come out:

    “For fish populations to thrive in the face of environmental change, members of a group need to behave differently from each other,” Polverino says.

    “Unfortunately, we found that such behavioural diversity is eroded in fish populations exposed to fluoxetine, and might place large groups of fish at an increased risk of perishing in a changing and increasingly polluted world.”

    Lack of diversity in behavioral patterns creates a brittle group, vulnerable to change in the environment – an environment that is changing rapidly. Now, if we follow this thought to its logical conclusion we could say that any system that loses in diversity as the diversity – uncertainty – in its environment is growing is doubly increasing its risk of catastrophic decline or collapse.

    This brings us back to a more general lesson: that diversity – in behavior and composition – is key to searching complex and fluctuating fitness landscapes for solutions. So we should benefit from finding out what processes in an organization that reduce behavioral diversity. This brings us to an over-interpretation of these results that amounts to a real heresy in today’s management work: that a strong culture may be a force that increases the brittleness of an organization.

    Fish need behavioral diversity to thrive.

    Modern organisations pride themselves on having and building strong cultures – and these cultures are routinely hailed as the engine of success for companies like Google, Netflix, Apple and smaller start ups as well. But what if there is a point at which a culture becomes too strong and results in a reduction of diversity in behavior?

    A point where the company culture makes us behave as fish on prozac?

    Maybe the real insight here is this: a company should ensure that it has a culture for all of the reasons we have heard again and again – it strengthens the organization, creates purpose and encourages innovation – but companies should also encourage counter-cultures, real differences in culture between different parts of the company, like geographical regions or different functions, to ensure behavioral diversity?

    Thinking back at Google you could see this emerging: EMEA had a different culture, an organizational dialect, and YouTube, again, had its own culture within the company. Over time that may turn out to be an interesting strength.

    +

    +

    +

    +

    +

    +

    + ,

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • Practicing predictions (Mental models IX)

    +

    +

    +

    +

    +

    +

    Predicting is hard, especially the future, as Yogi Berra supposedly pointed out. But it is interesting – but not necessarily for the reasons we originally think. Predicting the future is interesting not because you want to find out if you are right, but you want to use the predictions you make to tease out the narratives that are organizing your understanding of the world.

    I have started to run a predictions journal for myself and the way that I have structured it is that I am looking at a prediction, assessing the probability and then writing out the narratives that I use to fix on that probability. It is really interesting in that it forces me to not just blurt out a percentage probability, but really ground it. And in doing so I also need to examine my own thinking.

    Take a simple example: one of the predictions I wanted to assess was the likelihood that the stock market drops 30% before end of May this year. I assess that as a 15% probability, but I started out assessing it as 2-3% probability, because I felt it very improbable. Then I wrote out the narratives, and they started to feel jarring to me.

    I increasingly felt, as I was writing, that I underestimated the risk based on one single narrative – that there are no alternatives to the stock market right now. On the other hand I strongly believe that the stock market is over-heating and that the support packages provided during the pandemic have artificially inflated share prices to at least some degree. It is just hard to see how, and a 30% drop feels a lot.

    My prediction template in Roam and the beginning of the narratives I am writing out.

    So I stuck to my view that it is unlikely, but corrected – when I read my own narratives, the probability from 3 to 15%, which is a huge leap. I do think the poverty of my narrative made me overreact, and if I come back to and update this prediction I think I will probably be adjusting it down.

    There is an interesting misunderstanding around narratives – that they are just stories. That writing out the narrative that underpins your belief is just an exercise in justification and that you will not find anything in those narratives — but that is, I think, very misguided. Writing out a narrative is not making up, it is exploring and discovering.

    The reason is that when you write out a narrative you split yourself into two: the writing self and the scrutinizing self. This is why most people who want to become writers never do – they give too much space to the critical self, and give up, judging their own first drafts too harshly (first drafts are, as Hemingway reminds us, always sh*t so don’t do that – write away). That judging is at play also in writing out non-fiction narratives, and you will not let yourself get away with easy solutions. Between the writing and the judging self you can explore the set of beliefs and narratives that you exist in.

    Narrative is an instrument, a tool that uncovers cognitive structures governing you.

    So practicing predictions through writing out the narratives that underpin them is a way to discover the world you live in.

    +

    +

    +

    +

    +

    +

    + , ,

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • Balancing our intelligence investment portfolio

    +

    +

    +

    +

    +

    +

    It is no secret that AI is currently a hot subject – in politics, law and, of course, in investing. The idea here is right: this new technology may well revolutionize business after business and not paying attention to it would be folly. But there is an interesting question sitting at the edges of the current fervor with which we pursue artificial intelligence, and that is how we are structuring our investments in intelligence in general.

    Computer science and engineering has brought us enormous value, and as artificial intelligence technologies start influencing the way that companies are organized and the business models – it takes time for technology to do so – it will become a massive competitive advantage, not least by enabling a higher pace of learning. At the same time we know that there are other kinds of intelligence that we are not investing as much in: human and natural intelligences.

    Andrew Lo, respected economist and researcher at, among other places, MIT has suggested in a recent paper that we need a Brain Capital Grand Strategy. His observation is this:

    Our current economy is indeed a Brain Economy—one where most new jobs demand cognitive, emotional, and social, not manual, skills, and where innovation is a tangible “deliverable” of employee productivity. With increased automation, our global economy increasingly places a premium on cerebral, brain-based skills that make us human, such as self-control, emotional intelligence, creativity, compassion, altruism, systems thinking, collective intelligence, and cognitive flexibility [1]. Investments in brain health and brain skills are critical for post-COVID economic renewal, reimagination, and long-term economic resilience.

    Smith, E et al “A Brain Capital Grand Strategy: toward economic reimagination” Nature 26.10 2020

    This may seem obvious, but it does raise the issue of how we, as a society, have balanced our investments in intelligence. That artificial intelligence is promising is not a reason to abandon ways of investing in natural, human intelligence – and hence, brains.

    A balanced public investment intelligence portfolio will allow us to make progress faster.

    Especially since human intelligence is not a lower grade version of artificial intelligence, but rather entirely different. It has occurred to me, and many others, that we would be much better off if we categorized intelligence not as human and artificial, but as human, dolphin, bat – and looked at the different strengths – perhaps seeking to build artificial dolphin intelligence is much more interesting than replicating human intelligence? And in human intelligence we still know very little about the brain, and how it works and how it can be utilized. Another alternative woudl be to speak of biological and computational intelligences.

    In a sense, the AI-project is dependent on this research as well, since the organizing metaphor for that work has now become neurological.

    So seeking opportunities to invest in biological intelligence and brain research may not be such a bad idea.

    +

    +

    +

    +

    +

    +

    +
    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • Essay on utopias (in Swedish)

    +

    +

    +

    +

    +

    +

    Available over at Axess’ wonderful website here.

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • The importance of red team storytellers

    +

    +

    +

    +

    +

    +

    In an Edge-seminar held by Tetlock recently there was an interesting back and forth about the value of predictions. One view – held by Danny Hillis – could be simplified as: people do not think in predictions, they think in stories and so when people are wrong about the future they are in fact not wrong about predictions, but about what story we are in. This is profound in two ways: if it is true, probabilistic predictions are not a good way to become better at dealing with the future, unless you use them as a tool to craft an alternative story.

    Hard at work with red team story telling.

    This suggest a really simple, but probably impactful, intervention for anyone serious considering the future of an issue, a company or a particular system: uncover the main storyline, and then appoint a small team to be “red team storytellers” who tell stories that are compatible with the evidence but arrange it differently. The idea flashes across in this dialogue between a few of the seminar’s participants:

    Hillis:   There’s another dimension to it, which is that the very act of taking evidence and using that to adjust your prediction demands a framework of explanation. You have an operative story that you think, “Okay, well, this is how Hillary is going to get elected.” If you look at the failure of the intelligence community in the case of WMD, it was a failure of storytelling, not so much a failure of evidence interpretation. Retrospectively, you could tell a story of Saddam deliberately pretending like he had weapons of mass destruction, there’s a story about his generals lying to him; none of those stories were told beforehand.                 

    Had those stories been told beforehand, the evidence could have been interpreted differently. The same evidence could have been interpreted differently. In fact, that’s part of what I was saying about the job of the intelligence community in storytelling is providing the frameworks in which you can interpret evidence and make predictions.        

    Tetlock:   Right, right.        

    Brand:    So you want red team storytellers.        

    Hillis:   Well, yes.        

    Tetlock:   Part of the official operating procedure of the CIA and officials is to have that. In the case of Iraq WMD, that process broke down. You had the situation where the Director of the CIA did say to the President of the United States, “It’s a slam-dunk, Mr. President.”        

    Stewart Brand’s notion of red team storytellers captures something that has been worrying me about prediction markets and probabilistic reasoning: it assumes that the propositions or predictions make sense in themselves, when, in fact, we need to integrate them into a narrative to understand them.

    This in no way means predictions are useless. They can be used to challenge the main narrative that you are working with. What predictions would have to be true for this to be the case?

    Important ground work then would be laying out the main narrative in the business you are in, the most naive and most common story about why you are where you are. Then start to look for cracks.

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

  • A question of power?

    +

    +

    +

    +

    +

    +

    One of the core criticisms of platforms is that they have become far too powerful. While this is more of a throw-away comment, it is interesting to examine it up close to see if it can be turned into a hypothesis and tested – and what I am most interested in testing is not if platforms have become too powerful, since that will be a subjective evaluation, but rather if they are more powerful now than 10 years ago.

    In order to test this we need an operational definition of power. One of the simplest and best definitions of power is found in Max Weber’s work – he defines power as the probability that you get your way in given the sum total resistance from other parties. This probability then is the measure of power – if 100% you are all-powerful in a social system, and 0% you are utterly powerless.

    If you can complete your projects with a high probability, even in the face of resistance from other parties, you have a lot of power.

    Weber’s definition of power is probabilistic and operational

    Armed with this definition we can now do a counter-factual thought experiment – we can look at projects that platforms and tech companies executed on historically and ask if they would have a higher or lower probability of success today.

    • Take the Google Books-project. Is the probability higher or lower that Google would be able to complete a project to scan all the world’s books today than ten years ago?
    • Or StreetView – if Google announced today that it would travel all the world’s streets and record imagery to enable better navigation, would the probability that they were able to complete that project be higher or lower than ten years ago?
    • Or Libra – is the probability that Facebook succeeds with an alternative crypto currency launch of the same ambitions it had in the first launch lower or higher?
    • Would Facebook be more likely or less likely to be able to acquire Instagram or Whatsapp today than they were when they acquired them?

    It seems obvious, then, that tech companies actually have less power today than they did 2, 3 or 10 years ago – at least by the design of this experiment.

    Now, you can criticize the design of the experiment, but by this definition of power applied to these counter-factuals, well, it seems as if Big Tech has less – not more – power.

    But there is a difference, you may argue, between having more or less power and having too much power. I readily concede that this may be the case, and if we can find an operational way to think about power that shows it I would be interested in what that is.

    Power has fascinated social scientists for a long time, and as a result there is a plethora of different definitions of power, but a few interesting forms of power that would reward further exploration, I think, include:

    • Innovation power – this ability is divided into two, and we can use the myth that Plato has Socrates tell about Toth and Thamus to illustrate it. In the myth Toth is the genius inventor that develops new technologies and ideas, and Thamus is the king and ruler that decides if these inventions should be made publicly available. These two forms of power are closely related – Toth’s power to invent and Thamus power to disseminate. We probably undervalue what it means that a lot of both kinds of power reside firmly in the tech sector, and is found in large tech companies. They invent and disseminate the future.
    • Recruiting power – the ability to find the finest and most creative minds and attract them to an organization is a transformative power that also leads to a generative advantage – it strengthens the innovation power referenced above. This, too, I think is severely undervalued.

    There is much more to say about this, but any discussion of power would benefit from a closer examination of what we mean by using the term and how we can test our assumptions.

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +

    +