Betting the Spread on Inexorables
|Venkatesh Rao||May 5, 2017|
It’s been quite a week. In the US, the healthcare bill that ‘experts’ had written off just a few weeks ago managed to squeak through the House of Representatives after all. In France, Marine Le Pen has made it to the runoffs, and is currently polling at 38% versus Macron’s 62%. I guess we’ll know if the pollsters have any clue what they’re talking about in a few days. On another front, the ether crypotocurrency crossed $100 for the first time in what appears to be a broad trend buoying up the cryptocurrency markets (old man bitcoin hit $1700, and is hovering around $1500 as I write this, while ether seems to be hanging on precariously around $90). Back in August 2014, during the ether genesis sale, when I bought a small amount (sadly not enough to be a life-changer), it seemed like yet another joke. Most of us in the group that was chatting about it at the time and bought small amounts were treating it primarily as curiosity about the technology rather than a reasoned speculative investment. I’ll give myself zero credit for having made a ‘wise’ investment. If I come out on top on ether, it’ll be largely dumb luck combined with being sufficiently herd-minded around more plugged-in people. It was also around the same time that the alt-right hit my radar, and as with ether, I took it seriously, but not seriously enough. I lost a bet with a friend about if/when/how strongly the alt-right would influence mainstream politics.
It’s a good week to talk about predictions, and why conventional wisdom is wrong about an important matter: "long-term thinking" (whatever the hell that means) is not better than short-term. Even those who seem to be prescient long-term thinkers and win on that claimed basis misrepresent what they actually do. Here's the thing they don't tell you: the way to long-term prescience is through aggressively short-term commitments around inexorable trends. Trying to insulate yourself from short-term pressures and "planning" long-term is the absolute worst thing you can do. Working to quarterly pressures is terrible. You should be working to weekly pressures.
Your future planning canvas
1/ Like almost everybody who watches macro events, I saw both the reactionary wave of nationalism and the cryptocurrency boom coming a few years ahead of non-watchers.
2/ Like almost everybody who saw it coming, I got the timings, magnitudes, spreads, and expected volatility profiles wildly wrong. That’s why I’m not writing this from my own private island.
3/ It's not enough to be right to win cred and $ for long-term prescience. You have to be right enough, often enough, over several years.
4/ A fascinating disconnect in discourses about planning and anticipation is that people insist long-term thinking is better than short-term thinking even while acknowledging that we suck at prediction.
5/ Investors in financial markets at least have a very rich variety of ways in which to bet on the future given all kinds of combinations of uncertainties and ambiguities. Most of us don’t.
6/ For most of us the short-term vs. long-term question concerns much more real, everyday matters: what to major in in college, whom to marry, where to live, what kinds of projects to prioritize.
7/ Most of us are shut out of early-stage risk opportunities (barring the occasional cryptocurrency genesis sales or startup employee options). It's S&P index-tracking for us proles.
8/ I’ll make a radical claim: even those who say they work to long-term plans, and win, by having longer horizons than others, are unwittingly misrepresenting their stories.
9/ Just because you claim, after a big win, that you plan with a 5+ year horizon where everybody else is on a quarterly horizon doesn’t mean that’s what you’re actually doing.
10/ Here’s the basic truth about planning: you cannot plan better than you can predict. A plan is a largely deterministic sequence (modulo a handful of contingency paths) of actions on a pre-determined schedule.
11/ What people actually do when they claim they’re “planning long-term” or even just "thinking" long term, is betting on very powerful secular trends with the potential to swamp out a large class of uncertainties.
12/ When you look at their actual activities you realize that it is the opposite of planned. It’s organized around the shortest-term-possible thinking.
13/ What is “long term” about their thinking is not the planning, but the fact that all the wild short-term experimentation and play they’re doing shares key assumptions
14/ Assumptions like “bet compute power will be 2x cheaper in 2 years.” Not enough for detailed planning, but enough to bias random actions away from an informational zero mean.
15/ Let’s call this approach betting the spread on inexorables. You don’t plan, but you also don’t experiment/play entirely chaotically. You try multiple things around individual inexorable forces.
16/ Recent historical examples are mainly from computing: Moore’s Law for CPUs, falling cost of bandwidth, falling cost of memory and storage.
17/ But other domains have examples too: political reactionary energy may not be governed by a clear law like “white anger doubles every 18 months”, but is also an inexorable (if more volatile) force.
18/ Controversies about the future tend to blow up or not based not on importance or science, but the extent to which they involve inexorables.
19/ Climate change is an example: whether you are a believer or a skeptic, the argument is about an inexorable force: rising temperatures/atmospheric energy content.
20/ Betting on inexorables is the equivalent of dollar cost averaging in experiment-design space. You spread bets across things you can't predict.
21/ You can’t usually predict the precise timing, magnitude, or volatility of how an inexorable trend will evolve, but you can still ride it this way.
22/ So if you want to bet today on “machine learning” (a very good bet), and don’t understand the nuances of whether deep learning methods or GOFAI (“good old-fashioned AI”) methods are better, what do you do?
23/ You bet on as broad or narrow a spread of experiments as possible, that share the “machine learning will be big” assumption, given your knowledge and risk tolerance levels.
24/ What if you don’t have the right skills to make such bets? You get as close to the action as possible given your skills. If all you can get are the cheap seats, get them.
25/ Now to the key question: horizons. Once you’ve identified the inexorables you want to bet on via experiment spreads, what should be your planning horizon?
26/ The answer, paradoxically, is as short as possible. You think CEOs responding to quarterly pressures are being unwise? My recommendation is to actively seek out and subject yourself to weekly/daily pressures.
27/ You want to actively create the kind of churning, noisy environment that makes planning beyond a couple of steps ahead almost impossible. You don’t want to insulate against short-term pressures.
28/ This idea is of course not original to me. It's in the zeitgeist: Taleb's written extensively about it, agile development/DevOps is big in tech. Big CEOs talk this talk.
29/ The only area where planning longer is a good idea is where you have to contain the downside: things like health insurance or safety or ability to make rent or avoid defaulting on loans.
30/ Everywhere else, drive up the noise levels, fog of action and VUCA (volatility, uncertainty, complexity, and ambiguity) until you can see no more than 1-2 steps ahead.
31/ If you can see more than 2 steps ahead, you aren't taking enough risks. You gain on competitors when you can see fewer steps ahead than they can.
32/ Why this paradox? Other things (intelligence, baseline information, resources) being equal, the person who sees more steps ahead is likely hanging further back out of risk aversion.
33/ Sound scary? But that’s what it means to bet on an inexorable trend. If it is actually inexorable, it will prevail over the noise. The noise will ensure only the best candidates in your experimental portfolio survive.
34/ But don’t take my word for it. Two of the finest examples of global optimization and long-term survivability/value creation we have around us -- natural evolution and free market -- are 0/1-step look-ahead systems.
35/ So saying “be as short-term as possible” is equivalent to saying “be as Darwinian/market-like as possible” in your activities.
36/ Ironically, given we're talking about looking ahead in time, the best watchmaker we know of is a blind one. Risks are most likely to be introduced into such systems by trying to add "long-term thinking."
37/ If you want a pseudo-technical "explanation", you're trading off predictability in time for coverage in experimental space via a sort of biased ergodicity effect.
38/ If you have a budget of 10 "actions" you can take in a week, do you go deep (10 actions on 1 front) or broad? (1 action on each of 10 fronts?). False dichotomy.
39/ This is only a simple depth versus breadth tradeoff IF there is no correlation across the 10 fronts. If they're connected by a shared assumption, the game changes.
40/ The difference between unfocussed dabbling and betting the spread on inexorables is that assumption. That’s the signal in the noise. It’s what can convert a random walk into a drift towards compounding value.
41/ In hedge fund investing, hidden correlations usually represent risks, as in the famous case of LTCM in the 90s. In innovation, they represent serendipity potential.
42/ The reason 10 experiments around an inexorable trend is NOT spreading yourself too thin is that you're creating potential for serendipitous focus.
43/ I define serendipitous focus as _unplanned concentrations in resource allocation creating long-term compounding coherence out of short-term volatility. _
44/ This makes planning not just a neutral idea, but an actively BAD idea. You want to be open to serendipitous spillover from 1 experiment to another in your portfolio. Don't kill the goose that lays the compound-interest eggs.
45/ Planning introduces a certain rigidity into how you're going to get something done. It dampens correlations created by inexorable trends, creates reallocation friction.
46/ It also encourages debt-driven thinking, where everything you haven't done on a plan is debt hanging over your head, to be paid off through "execution" (heh!).
47/ But around an inexorable trend, the goal is to not to blindly pay off commitment debts simply because you think it's "execution discipline" but to constantly question the reality of the debt.
48/ Say you decide to write a serialized novel on your blog and have a 20-chapter outline laid out. But a comment on Chapter 1 suggests a different/better plotline.
49/ If you insist on sticking to writing the novel to the outline you laid out, or finishing novels clearly nobody wants to read, you've basically lost the value of serializing on a blog.
50/ The "right" strategy in this novel writing example is in fact to go the other way: parallelize the novel by writing 3-4 first chapters that share a narrative theme you want to bet on.
51/ Want to write a novel about a brave new crytptocurrency future? Don't write an outline of 10 chapters. Write 3 stealth first chapters instead. Bet the spread, not the outline. Keep switching parallel/serial bets as things change.
52/ Unlike debt owed to others for hard, liquid forms of wealth, with assets backing them, intention-debt is ephemeral. It's only as real as long as your intentions make sense.
53/ So here's an ultra-radical thought for you: try going a month with all planning abandoned. Simply identify a bunch of 1-step-look-ahead "frontiers" around a key shared assumption.
54/ Your only rule for the month will be that you'll choose what to play with/improvise on/experiment with from that bucket. Otherwise you'll just do whatever the hell you feel like.
55/ This is the anarchist null hypothesis of productivity. If just doing whatever the hell you want around 1 key assumption creates compounding movement/momentum, why do more?
56/ Sure, you'll create a mess. But having to clean up a messy but valuable condition beats executing a clean and perfect but meaningless plan because you're "not a quitter."
58/ The more I think/write/read about it, the more it seems true to me that the most interesting lives emerge not from planning but from putting yourself where you can plan the least, but ONE big good thing is happening amidst chaos.
59/ Not to get too meta about it, but a bunch of my creative efforts right now (fiction/non-fiction writing, consulting projects I'm accepting) are an experimental spread bet around "short-termism is actually good, long-termism is actually bad."
60/ It's a small idea that's been turning into an inexorable intellectual trend over 30 years. And if your work involves working with ideas, this is one you want to bet on any way you can.
_Feel free to forward this newsletter on email and share it via the social media buttons below. You can check out the archives here. First-timers can subscribe to the newsletter here. You can set up a phone call with me via my Clarity.fm profile page. _
Check out the 20 Breaking Smart Season 1 essays for the deeper context behind this newsletter. If you're interested in bringing the Season 1 workshop to your organization, get in touch. You can follow me on Twitter @vgr
Copyright © 2017 Ribbonfarm Consulting, LLC, All rights reserved.