Cone of foresight: an expanded veil of ignorance for our future

Counterfactual consequentialism in our complex world & living with uncertainty

We strive for the best we can attain within the scope the world allows.

What's common amongst all these?

• Bob gets drunk when he's out at his friends and decides to drive home. He's a great driver despite the alcohol so drives pretty well. He's pulled over and charged with a DUI regardless.

• Alice lies to get her job as an investment banker. She has a stellar career and gets found out after 7 years when she's a Partner. She's unceremoniously fired.

• Jack has been meaning to fix the leak in the bathroom for weeks, despite being asked by Jill. Finally he said he did it. Three weeks later the sink breaks, bathroom floods and they have to redo the floors.

• Sam buys stock of Guns & Ammo Inc, purely by looking at its financial performance. It grows over the next two years, but because they've been lax in their background tracking program. Two teens end up buying the guns from them and shooting up a mall.

• Chris invests the money in his company’s Treasury into Bitcoin, thinking he will make an extra return for his investors. He’s confident that the value won’t be too volatile. But the overall market crashes and he loses the company half its Treasury balance. a16z is not pleased.

You do something. That action has consequences. Some of the consequences are known and some unknown. At what point does your fallibility for lying give way to make you not liable?

For instance, Bob clearly knew what he was doing is against the law. He still might have better ability to drive compared to the average driver, but it's still against the law we've all decided to set up. So he gets pulled over. And since law isn't individualised he gets charged. Is this foreseeable? Absolutely. Fine.

Alice on the other hand might have done well to consider the repercussions of her actions. Maybe she'd have been found out at the first challenge and not gotten in. And she was willing to take that risk.

But once past the gate it becomes an internal game that she showed herself adept in. It was a clear flaw in the bank's hiring policies that she exploited. Here the harm done is tenuous (maybe she pushed a candidate who would also have been good out of the way, maybe she would’ve been worse counterfactually, maybe she could’ve encouraged many more to pursue this treacherous path). But once she's been there a few years and is rightly considered integral (she made Partner), shed be let go. Still her fault of course, but could she have foreseen this? Unlikely.

There is a Cone of Foresight. I see it as an updated version of Rawls’ veil of ignorance, extended for uncertainty in the future, and helpful when considering making plans.

Among the essential features of this situation is that no one knows his place in society, his class position or social status, nor does any one know his fortune in the distribution of natural assets and abilities, his intelligence, strength, and the like. I shall even assume that the parties do not know their conceptions of the good or their special psychological propensities. The principles of justice are chosen behind a veil of ignorance.

Not just a minmax principle, making those least well off have dignified lives (since you could be them), but making decisions that leaves humanity better off in the future; both that it exists at all and that it plays out like you’d imagine.

Like if the veil of ignorance asked you to make decisions about the society you’d design if you didn’t know who you’d be in the society, the cone of foresight asks you to act such that the results of your actions can be blamed on you only if you could’ve legitimately foreseen those actions. I used to call this counterfactual consequentialism in my youth (22.k.iii.4 here), though it turned out only I found this exciting.

So even though no harm was done in this instance, you could probably blame Bob for his blackhearted decision to drive drunk. And maybe blame Alice a little less, at least after the first few years of her subterfuge. And Sam and Jack probably even less, because their cones would’ve been much fuzzier when looked further out.

Chris though, Chris is another matter. There are actual companies who did this, and did it successfully. Successful here means using the fact of having done it to get their company more pride, glory and customers. So maybe he should’ve been more prudent? But at the time of making the decision it might not have been too bad?

If this still makes you feel its his fault maybe replace Bitcoin with taking on debt. At the time it was cheap! And when the market went all topsy turvy and the sales fell of a cliff, the debtholders say “hey, nice company you got there, we’ll take it from here”. And Chris is sad.

But what if Tim, a silent observer so far, spent his efforts at trying to do environmentally friendly activities like protecting the Amazon, and only ended up bringing more attention to the lucrative timber industry there?

For instance after Titanic sank the US Senate mandated every ship have enough lifeboats for every passenger. Sensibly so. But later the ship SS Eastland, ungainly beast with lifeboats crammed on her top deck, turned right over like a kebab in open water, killing 800 people. Should the US Senate be held responsible for this? The owners should’ve made a better ship, sure, but the Senate should’ve realised idiot owners exist who wouldn’t have made a better ship¹. Or the discovery of nuclear fission. Or the Treaty of Versailles².

If you believe in this, as is kind of inevitable with a certain form of utilitarian argument, then you can shrug your shoulders and go ‘huh, you reap what you sow’. If you’re of a different philosophical bent, then the failure to predict the future beyond what you can credibly do is key to almost all of human progress.

Beyond the cone of foresight, once the spiralling edges get fuzzier we'll have to leave utilitarian calculus and make some deontological strides. We have to abandon careful analyses of what we should do to maximise a certain value and aim to create a philosophy that tries to define certain action sets.

For instance, if we were to try and ensure the ultimate goal of colonising the galaxy, is it useful to work on SpaceX or Blue Origin or generally try to push our spacefaring along? Maybe a little, and maybe directionally, but that's unlikely to be the key bottleneck.

A while ago an economist by the extraordinary name Frank Knight said there were things you don't know and things you don't know you don't know. He stopped there, thankfully, but from this we get the idea of Knightian uncertainty. Basically the uncertainty that can't be quantified or measured or in general worked with.

That’s also what creates and maintains the cone. Whatever’s visible within your cone is things you can influence, and whatever’s at the fuzzy edges are beyond the ability to predict or foresee³.

Alex Muresianu @ahardtospell

the problem with the whole "dur, growth can't be infinite because in billions of years the sun will swallow us and later the universe will explode" thing is that it assumes we won't be able to develop technology to prevent the heat death of the universe

Unlike risk, which in his parlance is measurable problems that lie ahead. So if someone said to you get, wanna take a gamble on this Monte Carlo slot machine? You'd say no thank you sir, the risk is too high.

But if someone said hey you should be an AI trainer at Goldman Sachs, that's less quantifiable. You don't know what the job is, if it will exist, or your qualifications or lack thereof. Put another way, the probability distribution to understand this is unknown.

So, back to our galaxy colonisation dreams, which is much closer to the latter than former. In fact the bottleneck is likely to be something you don't know today. That's kind of what Knightian uncertainty is.

The key issue is that there exist a large number of unknown unknowns in attaining this goal. We have no idea how to even get to a point of tackling them. In the absence it might be best to just maximise economic growth and technological development, since that's more likely to help us have the necessary tools. And once we have the necessary tools, when the time comes, and we're roughly aligned with our moralities of today, we'll be able to attain the goals⁴.

When I wrote about thinkers and doers as two (fuzzy) globs of people, this is what I was talking about. When Elon promises the moon repeatedly and delivers it occasionally, this is the difference. “How to get full self driving” is a wicked problem of Knightian uncertainty. “Get the rocket to land back on a funnily named floating barge” is not. Who knew.

We didn't land on the moon by the Victorians building bigger barges. We did so by inventing steel and discovering ballistics and building computers. These ended up tracing a connected path in hindsight, though they were nowhere in the cone of foresight looking ahead.

This isn't just a philosophical debate about apportioning blame when it comes to people making decisions that impact themselves or others badly. It's also the cornerstone of how we think about making decisions whose outcomes we won't know until much later. Planting the trees the shades of which our younger generations rest under, this is not.

But hey, you say, as you type away at this wondrous machine that was designed built conceptualized and engineered by humans, surely some of us see further than others, and if some of us see it then the ability must exist, and if the ability must exist, we can make it.

And I will answer sure, if you are able to find those oracles, and ensure that they will be treated as Delphi instead of Cassandra, and the record will stand in testimony against the vagaries of time, then we should do it. But I am not holding my breath.

The perspective of eternity is not a perspective from a certain place beyond the world, nor the point of view of a transcendent being; rather it is a certain form of thought and feeling that rational persons can adopt within the world. And having done so, they can, whatever their generation, bring together into one scheme all individual perspectives and arrive together at regulative principles that can be affirmed by everyone as he lives by them, each from his own standpoint. Purity of heart, if one could attain it, would be to see clearly and to act with grace and self-command from this point of view.

We have always had difficulty dealing with the unknown. How we act under this umbrella of uncertainty, without being paralysed?

There are, of course, prophets. Nostradamus might be a favourite example, but Tocqueville is far more prescient, and precise, in his saying. He famously predicted the 1848 revolutions that sparked Europe in his speech to the French aristocracy, saying how they were becoming similarly unworthy as they had been in 1789, saying:

public morality is being degraded, and that the degradation of public morality will shortly, very shortly perhaps, bring down upon you new revolutions

We could, perhaps, deal with the impossibility of determining the future course precisely by throwing up our hands and becoming pure hedonists.

Hume argued that the belief in causality itself isn’t provable by experience nor reason, skeptical we can predict the future at all. Confucius asked us to study the past to define the future. Voltaire called the present pregnant with the future. William Gibson said it’s already here, just unequally distributed. Taleb thinks the world would have been predictable were it Mediocristan, where black swans were suitably rare.

Whether it’s in companies, or the sciences, or life itself, it’s rarely extreme foresight that leads one to success. Rather it’s a combination of nimbleness and willing to fill the niche that opened up, than the niche you really really wanted, that makes for success. The future is built one brick at a time, even though we don’t know the shape of the building when we lay the bricks.

We have been historically obsessed with defining the shape of it, even in the form of multiple stories we could credibly believe. As a way to not be scared of staring into the abyss of the unknown. But the cone of foresight tells us that there is no easy answer to this. We might be able to define the boundaries of the cone, perhaps, in not wanting to face existential risk, or seek harm, or run away from pleasure, but these are not a roadmap to it, but a set of principles about who we are.

As Nietzsche said, all we can do is to live well, act honourably.

Calibrate our actions in such a way as to make their consequences bearable, livable with, in a hypothetical perpetuity

To understand the present better, to see it in new ways as we can, to try to ever expand the cone of foresight, that is the way to live a good life. And to be okay with the fact that you’ll not be able to predict what comes next. If you’re lucky in science, they’ll call it a lucky guess. If you’re lucky to be in the right places at the right time, they’ll call you a visionary, with the right instincts⁵. If you’re lucky in business, they’ll call it a successful pivot, and that’s the only recipe for eventual success we have.

1. Since it’s the US Senate, I’d say yeah why not, like catching Capone with his tax dodge.

2. If we can’t easily tell the long term consequences of our inventions, like nuclear fission, how do we as a society make sure we don’t accidentally destroy ourselves? Clearly having a God’s eye view of our lives and efforts, a panopticon, is one possibility. But panopticons manned by flawed humans are as susceptible to corruption as anything else.

We adopt rules and norms and heuristics to tie off the worst outcomes of the things we can accidentally do. And for the rest of it, we believe in our inner sense of goodness and our ingenuity at course correction to get us back on track.

3. So if the world gets fuzzier as we look forward in time, how do we plan? Turns out we do a few things.

1. Like Bezos popularised, we can look at things that don't change. Like customers always wanting faster deliveries and cheaper goods.

2. We could do ‘’scenario planning’ and essentially try to guess at a few different futures, and try to do things that won't go too far awry if they come to fruition.

3. We could try assess various combinations of futures, put probability estimates on them, and maximise expected value.

4. Which brings us to a truism, and also my point of view about Will McCaskill's wonderful book what we owe our future. In a world where we could credibly predict the outcomes of our actions in the future, the book is incredibly valuable, indeed. It is prescient. In a world where there is a half life on the predictable outcomes of your actions, it falls short. There's a longer review that I'm trying to figure out how to right, but I'll leave this here for now.

5. Like Stewart Brand, or Zelig.