Complex Systems, Multiscale Information and Strange Loops (by Silly Rabbit)

Complex systems

Neat and accessible primer on complex systems, multiscale information theory and universality by Yaneer Bar-Yan, and a related paper on the conceptual applications of the same topic: From Big Data To Important Information (suggest start reading from section VII if you read the primer, and from sub-section D on page 13 if you just want the markets application).

Machine learning software creates machine learning software

Lots of buzz about Google’s AutoML announcement at the Google Annual developer conference I/O 2017 last week. AutoML is machine learning software which takes over some of the work of creating machine learning software and, in some cases, came up with designs that rivals or beats the best work of human machine learning experts. MIT Technology Review article on AutoML.

One-shot imitation

Also lots of buzz around one-shot imitation using two neural nets, as demonstrated by OpenAI. Personally, one-shot imitation is the one AI-type concept which gives me the fear. But if Elon’s supporting it then it must be OK… right? One-shot imitation paper here but, more to the point, watch this video and tell me you are not at least a little bit afraid.

The power of the platform

And to the practical applications of technology, I really like the language of this recent press release by Two Sigma CEO, Nobel Gulati, and particularly the paragraph:

Moving forward, durable advantages will to accrue to those building a substantial platform based on massive amounts of data, along with the technology and institutional expertise to use it. Building such a platform requires significant and ongoing investment in R&D, and a fundamentally different culture and mindset to apply a scientific approach to the data-rich world of today.

Personally, I believe that the 2020s will be more defined by big compute than big data but this is, nonetheless, a powerful statement and language, and there’s a key implicit point buried in here on the cultural balance of ‘researchers’ (math and physics natural genii) and ‘production engineers’ (coders who, by nurture, have seen and solved many practical problems). Specifically, how the majority of quant funds have to-date been culturally focused too heavily on the math genius research folks to the detriment of hiring and rewarding the more workmanlike practical folks who can build and maintain a substantial platform which, I agree, is the new durable advantage.

去吧

I was reminded last week by China’s censorship of Google’s latest AlphaGo win against Ke Jie just how substantial a stance it was when Google shut down its Mainland search engine in 2010 and why these kind of bold moves (bets) are essential to developing a truly winning technology company (and also why I don’t live in China anymore!). As Rusty Guinn has written about: A man must have code.

Strange loops

Finally, to bring us back up to the level of self and consciousness, I finally got ‘round to reading Douglas R. Hofstadter’s 2007 book I am a Strange Loop. A long, winding and compelling book summarized by the quote “In the end, we are self-perceiving, self-inventing, locked-in mirages that are little miracles of self-reference.” If you dip in and only read one section, read the section on simmballs in Chapter 3, which loops us back to where we started this column on multiscale information.

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/16045/

She Screams, He Kidnaps (by Silly Rabbit)

Proximity of verbs to gender

Sometimes biases embedded within language are subtle, counter-intuitive things which you have to tease out with many layered neural nets. Other times, they are just bluntly and painfully predictable: Data scientist David Robinson tracked the proximity of verbs to gender across 100,000 stories. She screams, cries and rejects. He kidnaps, rescues and beats.

Wiki-memory

Previously I shared some research on how recollections of successive events physically entangle each other when brain cells store them. As a fascinating and different approach to studying memory, in this paper a group of European researchers used Wikipedia page views of aircraft crashes to study memory.

Fool me once, fool me twice

Sooner or later, someone is probably going to put a visually compelling 2D ‘map’ of data reduced from hundreds or thousands of dimensions via t-SNE in front of you and make some bold assertions about it. This beautiful and interactive paper provides a handy guide on what to watch out for.

A veritable zoo of machine learning techniques

A couple of months old, but still useful: Two Sigma researchers Vinod Valsalam and Firdaus Janoos write up the notable advances in machine learning presented at NIPS (Neural Information Processing Systems Foundation) 2016. Headline: The dominating theme at NIPS 2016 was deep learning, sometimes combined with other machine learning methods such as reinforcement learning and Bayesian techniques.

The NIPS conference has, improbably, found itself at the center of the universe as it is the most important event for people sharing cutting edge machine learning work. It’s in LA this year in December and promises to be very interesting, although quite technical: https://nips.cc/

Silicon Valley: a reality check

And, finally, this one is a little inside baseball but, if you can push through that, there is a very useful and accurate parsing of the types of technology companies being started and funded in the Valley and the simultaneous parallel dimensions that exist here. (You can skip the Valley defense bit and jump to the smart parsing bit by hitting ‘CRTL + F’ , typing ‘Y Combinator’ and start reading from there).

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/16051/

And They Did Live by Watchfires: Things that Don’t Matter #4

Oliver Bird

There are two kinds of stories we tell our children. The first kind: once upon a time, there was a fuzzy little rabbit named Frizzy-Top who went on a quantum, fun adventure only to face a big setback, which he overcame through perseverance and by being adorable. This kind of story teaches empathy. Put yourself in Frizzy-Top’s shoes, in other words.

The other kind: Oliver Anthony Bird, if you get too close to that ocean, you’ll be sucked into the sea and drowned! This kind of story teaches them fear. And for the rest of their lives, these two stories compete: empathy and fear.

Oliver Bird, Legion, Chapter 4 (2017)

Skinner: What are you doing in here?
Linguini: I’m just familiarizing myself with, you know, the vegetables and such.
Skinner: Get out. One can get too familiar with vegetables, you know!
— Ratatouille (2007)

It’s four in the morning, and he finds himself drawn to a hotel and casino that has been out of style for thirty years, still running until tomorrow or six months from now when they’ll implode it and knock it down and build a pleasure palace where it was, and forget it forever. Nobody knows him, nobody remembers him, but the lobby bar is tacky and quiet, and the air is blue with old cigarette smoke and someone’s about to drop several million dollars on a poker game in a private room upstairs. The man in the charcoal suit settles himself in the bar several floors below the game, and is ignored by a waitress. A Muzak version of “Why Can’t He Be You” is playing, almost subliminally. Five Elvis Presley impersonators, each man wearing a different colored jumpsuit, watch a late-night rerun of a football game on the bar TV.

A big man in a light gray suit sits at the man in the charcoal suit’s table, and, noticing him even if she does not notice the man in the charcoal suit, the waitress, who is too thin to be pretty, too obviously anorectic to work Luxor or the Tropicana, and who is counting the minutes until she gets off work, comes straight over and smiles. He grins widely at her, “You’re looking a treat tonight, m’dear, a fine sight for these poor old eyes,” he says, and, scenting a large tip, she smiles broadly at him. The man in the light gray suit orders a Jack Daniel’s for himself and a Laphroaig and water for the man in the charcoal suit sitting beside him.

“You know,” says, the man in the light gray suit, when his drink arrives, “the finest line of poetry ever uttered in the history of this whole damn country was said by Canada Bill Jones in 1853, in Baton Rouge, while he was being robbed blind in a crooked game of faro. George Devol, who was, like Canada Bill, not a man who was averse to fleecing the odd sucker, drew Bill aside and asked him if he couldn’t see that the game was crooked. And Canada Bill sighed, and shrugged his shoulders, and said. “I know. But it’s the only game in town.” And he went back to the game.

— Neil Gaiman, American Gods (2001)

I had a dream, which was not all a dream.
The bright sun was extinguish’d, and the stars
Did wander darkling in the eternal space,
Rayless, and pathless, and the icy earth
Swung blind and blackening in the moonless air;
Morn came and went—and came, and brought no day,
And men forgot their passions in the dread
Of this their desolation; and all hearts
Were chill’d into a selfish prayer for light:
And they did live by watchfires…
— George Gordon, Lord Byron, Darkness (1816)

A Year Without Summer

In 1816 — 200 years ago — much of the world was experiencing a “Year without Summer.” We now know this was a result of the 1815 eruption of Mount Tambora, a volcano on the sparsely populated Indonesian island of Sumbawa, an eruption which sent some 38 cubic miles of rock, ash, dust and other ejecta into the atmosphere. For reference, that’s roughly 200 times the volume of material ejected in the eruption of Mount St. Helens, but only a tenth or so the size of the Lake Toba explosion off Sumatra that some researchers believe caused one of the most perilous bottlenecks in human genetic history.

At the time in 1816, the world didn’t know the cause. Well, except for maybe the people living on Sumbawa. The effects, on the other hand, couldn’t be missed.

In New England, the clouds of ash that blocked the sun led to remarkable drops and extraordinary variations in temperature and precipitation. In the Berkshires, there was a deep freeze in May. It snowed in Boston as late as June 7. Cornfields in New Hampshire were ruined by frost on August 14. The Dartmouth College campus was blanketed by snow as late (early?) as September. It caused as near a true famine as the U.S. has ever experienced. Hardy crops — some strains of wheat, potatoes and the like — got most of the nation through the year, as did a culling of wild game that likely came as a bit of an unpleasant surprise to the squirrels, hogs and possums that were usually spared a place on the American table.

The situation was not much better in Western Europe, where average temperatures fell as much as 3-4°C. On the British Isles, failed wheat and potato crops meant famine for much of Ireland, Wales and Scotland. Germany had food riots. And in Switzerland, where Lord Byron was in residence with Shelley, the constant rain and cold led each to create a great deal of poetry, which, depending on your opinion of early romanticism, was either more or less catastrophic than the torrential rains that accompanied it. Under the circumstances, it is not surprising that Byron was inspired to write about the heat death of the universe some 35 years before Lord Kelvin proposed it rather more formally (and perhaps less melodramatically).

Byron’s poem, Darkness, envisions a world in which the sun has been extinguished, in which morning never comes. In this time of desperation, the world is literally tearing itself apart. Palaces are ripped to pieces for firewood, forests are set alight and people gather “round their blazing homes” just so they can see their own hands, and the faces of their family and friends. Everything the world has built is pulled apart piece by piece in search of a solution to the problem of darkness.

The stories we tell about such times of desperation tend to fall into the two archetypes Jemaine Clement’s character describes in Legion: stories of fear and stories of empathy. Byron gave us a story of fear. Empathy stories, on the other hand, follow the usual trope of necessity as the mother of invention. But even this is often just a fear story with a different outcome, not uncommonly summoning a sort of deus ex machina. Luke listening to Obi Wan’s disembodied voice instead of the computer as he aims his last shot at the Death Star. Gollum showing up to bite off Frodo’s ring finger and take a dive into Mt. Doom, saving the hobbit from the now too-strong temptation to wear the ring and return it to Sauron. And maybe there’s a story where Byron’s humanity finds a real solution to the coming darkness instead of tearing their homes and businesses apart looking for something else to burn.

The investment environment we face is not so dire as all this, but it does feel a bit grim, doesn’t it? Market returns have continued to defy the odds, but the data, our consultants, our advisors, our home offices and our instincts are telling us that the combination of demographic slowing, stagnant productivity, limited debt capacity, low rates and high valuations isn’t going to end well. Or at a minimum, we remain optimistic but confused. I’m sure we’ve all asked or heard clients and constituents asking us, “What the hell do we invest in when everything is expensive and nothing is growing?” In this call to action, are we successfully turning this into an empathy story? Or are we just ripping apart our homes for tinder so that it looks like we are doing something?

When it’s hard to see what’s two feet ahead of our own noses, when the game feels rigged, sometimes it feels like we have no choice but to stay at the table and play. After all, it’s the only game in town. And so instead of walking away and taking what the market gives us, we tweak, we tilt, we “take chips off the table,” we “go all in” and we hack, hack, hack at the beams and joists of our own homes for the great bonfire.

This bias to action is a road to ruin. That’s why the endless tweaking, trading and rebalancing of our portfolios takes spot #4 on our list of Things that Don’t Matter.

Of Priuses and Passive Investors

In 2011 a group of researchers at Berkeley examined an age-old question: are rich people driving expensive cars the asshats we all think they are? The findings? Yes, indeed they are! The study found that drivers of expensive cars were three times more likely than drivers of inexpensive cars to fail to yield to pedestrians at crosswalks requiring it, and four times more likely to go out of turn at a stop sign. The team performed similar tests in other non-traffic areas (e.g., cheating at games of chance, etc.) that arrived at the same conclusion, and furthermore identified that simply making people believe that they were part of the 2% Club made them behave more rudely.

My favorite discovery from the research was the odd outlier they discovered: the moderately priced Toyota Prius. Fully one third of Prius drivers blew by intrepid Berkeley grad students (taking a night off from throwing trash cans through the windows of some poor Wells Fargo branch, perhaps) who stepped into a busy crosswalk for science. This put it very near the top of the tables for rudeness. Most of you will recognize this as our old friend moral licensing: the subconscious tendency to feel empowered/entitled to do something bad, immoral or indulgent after having done something to elevate our estimation of our own value. The Prius owner has earned the right to drive like a jerk, since he’s saving the world by driving a hybrid car, after all. Alberto Villar of Amerindo Investment Advisors, who was the largest opera donor since Marie Antoinette, could easily justify stealing his clients’ money to make good on charitable pledges. Of course I can eat that Big Mac and large fries when I sneak over to the McDonald’s across the street from our San Francisco office — I ordered a Diet Coke, after all.

And so on behalf of insufferable hipsters, fraudulent philanthropists and Big Mac dieters everywhere, I would like to extend a gracious invitation to our club: ETF investors who pride themselves on being passive investors while they tactically trade in and out of positions over the course of the year.

Now there’s a lot of old research protesting too much that “ETFs don’t promote excessive trading!” A cursory review of news media and finance journals will uncover a lot of literature arguing exactly that, although the richest studies are several years old now. You’ll even find some informing you that leveraged ETFs aren’t being abused any more. Those of you who are closest to clients, are you buying what the missionaries are selling on this one?

I hope not.

Even when some of the original studies were published (most of which said that mutual funds were held around three years on average, while ETFs were held about two-and-a-half years), it was plainly evident to anyone who works with consumers of ETFs that basing claims on the “average” holding behavior was a poor representation of how these instruments were being held and traded. The people with skin in the game who weren’t selling ETFs were aware that holders fell by and large into three camps:

  1. The long-term holders seeking out market exposure,
  2. The speculators trading in and out of ETFs to generate additional returns, and
  3. The increasingly sad and depressing long/short guys shorting SPY to hedge their longs, telling the young whippersnappers stories from a decade ago about “alpha shorts” before yelling for them to get off their lawn.

Source: Morningstar. For illustrative purposes only.

The mean holding period in the old research was still pretty long because Group 1 was a big group. I think that it was also because a lot of the ETF exposure that Group 2 was swinging around was in smaller, niche funds or leveraged ETFs. Both of these things are still true. They’re also becoming less true. A few weeks ago, Ben Johnson from Morningstar published this chart of the ten largest ETFs and their average holding period. There’s all sorts of caveats to showing a chart like this — some of the causes of ETF trading aren’t concerning — but if SPY turning over every two weeks doesn’t get your antennae twitching, I’m not sure what to tell you.

There are a lot of reasons to believe that we are lighting our houses on fire with the almost comically active use of “passive” instruments, and trading costs are one of them. Jason Zweig wrote an excellent piece recently highlighting research from Antti Petajisto on this topic. Petajisto’s work in the FAJ estimates that “investors” may be paying as much as $18 billion a year to trade ETFs. Zweig, perhaps feeling rather charitable, concedes that as a percentage of overall trading volume, this number isn’t really all that high. And he’s technically correct.

But who cares about trading volume, at least for this discussion, which isn’t really about the liquidity of the market? If — as so many investors and asset managers are fond of saying — the ETF revolution is but a trapping of the broader active vs. passive debate (insert audible yawn), we should really be thinking of this in terms of the asset size of the space. And in context of the $3 trillion, give or take, that is invested in ETFs, $18 billion is a LOT. It’s 60bp, which would be a lot even if it weren’t impacting investors who often make a fuss over whether they’re paying 15bp or 8bp in operating expenses.

And then there’s taxes. Now, actively managed strategies, especially those implemented through mutual funds, have plenty of tax issues and peculiarities of their own. But the short-term gains taxable investors are forcing themselves into by timing and day-trading ETFs are potentially huge.  If we assume, say, a 6% average annual portfolio return, the investor who shifts 100% of his return from long-term gains into short-term gains is costing himself 60-120bp per year before we consider any time value or compounding effects of deferring tax liabilities. Given that the largest ten ETFs all have average holding periods of less than a year, this doesn’t seem to be all that inappropriate an assumption.

The growing Group 2 above, our day traders — oops, I mean, our “passive ETF investors” — may be giving away as much as 1.2%-1.8% in incremental return. Those fee savings sure didn’t go very far, and the direct costs of all this tinkering may not even be the biggest effect!

Every piece of data on this topic tells the same story: when we try to time our cash positions to have “ammo to take advantage of opportunities,” when we decide a market is overbought, when we rotate to this sector because of this “environment” that is about to kick off, when we move out of markets that “look like they’ve gotten riskier,” when we get back in because there’s “support” at a price, we are burning down our houses to live by watchfires.

There are two ways in which we as investors do this, one familiar and one less so.

Of Clients and Crooked Card Games

First, the familiar. We stay in the crooked game because it’s what’s expected of us. It’s tempting to think of the desire, this inclination toward constant “tactical” trading as an internal impulse. A response to boredom or, perhaps, an addiction to certain of the chemical responses associated with winning, with risking capital, even with losing. I think that’s probably true for some investors. I know that when I sat in an allocator’s seat, when I heard a portfolio manager tell me he had “fallen in love with the market” when he was six years old and started trading options with his dad when he was 10, I didn’t see that as a particularly good thing. One can get too familiar with vegetables, you know.

But just as often, the impulse to stay in the game is external, and that pressure usually comes from the client. I’m empathetic to it, and it’s not unique to our industry.

Have you ever sent a document to a lawyer and gotten no comments back? Have you ever visited a doctor and gotten a 100% clean bill of health with no recommendations? Have you ever taken your car to a mechanic and had them tell you about just the thing you brought it in for? Have you ever consulted with a therapist or psychiatrist who didn’t find something wrong with you, even if they had the bedside manner to avoid using those exact words? It isn’t just that those folks are being paid for the additional services they’re proposing. There is a natural feeling among professional providers of advice that they must justify their cost to their clients even if the best possible advice is to do nothing.

The result is that the crooked card game usually takes three different forms, which, in addition to all the fees and tax impact discussed above, may add risk and harm returns for portfolios in other ways as well:

  • The Cash Game: When investors feel concerned about the timing of their entry into markets, the direction of markets, upcoming events, or some other factor and temporarily sell investments and go to cash, they’re playing the Cash Game. I recently had a meeting with an intermediary who had recently launched a system to integrate all client holdings (including accounts held away). Their initial run identified average aggregate cash positions of more than 15%!
  • The PerformanceChasing Game: I’ve talked about this ad nauseam in prior notes. We investors find all sorts of vaguely dishonest ways to pretend that we aren’t just performance-chasing. It doesn’t work, and a goodly portion of the damage done by tinkering and “tactical” moves is just performance-chasing in guise, even if we are high-minded enough to pretend that we’re making the decision because “the fund manager changed his process” or euphemistically inclined enough to say the investment “just wasn’t working,” whatever that means.
  • The In-Over-Our-Heads Game: Still other games are essentially designed to “fleece the odd sucker,” causing investors to seek out hedges and interesting trades to take advantage of events and “low cost” insurance for portfolios. As a case study, please take a gander at the size and volume of instruments and funds tracking the VIX. Please look at the return experience of holders of those various instruments. It’s not the vehicles themselves that are flawed, but the way in which these markets prey on misplaced expectations of investors that they know when insurance is cheap or expensive. As a quick test: if you can’t define gamma without looking it up on Investopedia, you probably shouldn’t own any of these instruments, much less be flopping in and out of them. This concept is broadly transferable to a variety of things investors do to “hedge” — buying S&P puts, buying short ETFs, etc.

I know I’m not treading new ground here. Borrowing from the work done in a thorough survey on the literature that itself concludes a 1.0% impact from the ways in which investors trade in and out of funds, the figures are pretty consistent. The folks over at Dalbar concluded in 2016 that investors in equity mutual funds underperformed equity indices by 3.5% over the last 20 years, 1.5% of which they attribute to “panic selling, exuberant buying and attempts at market timing.” Frazzini and Lamont previously estimated 0.85%. In 2007, Friesen and Sapp said 1.56%. We’ve got something for hedge fund investors, too.

You’ve heard this story before. So why am I telling you this?

Because when I meet or speak with investors, I often worry that when they think about dominant narratives and observations about human behaviors, they are focused on identifying tradable trends and signals. In rare cases, that is a worthwhile endeavor. And we’ve made no secret that we’re spending a lot of time thinking about the Narrative Machine — after all, if we believe that investors systematically make mistakes that cost them returns and money, it should be possible to identify ways to capitalize on the actions taken by others.

But far more often, the message from the analysis of prevailing narratives is to back away from the table. Investors I’ve spoken to in the past few years have heard a voice of caution against rotating away chunks of portfolios that by all rights ought to be invested in bonds based on flimsy rationale like, “rates couldn’t possibly get lower!” I’ve likewise cautioned against haphazardly fleeing equity markets into cash on the basis of historically high valuations, perceived political turmoil and the like. There will come times where it may be right to make strong positive observations on opportunities for tactical allocations, but as in all decisions we make when investing, it is imperative that we be aware that the hurdle for staying at the table to play the only game in town is very high. Our skepticism about opportunities to play it should be extreme.

Of Bambi and Battle Tanks

Since I’m advising you to be skeptical, I’ll forgo the apocryphal (it’s real to me, dammit!) story I was going to tell at this point in my little piece. I was going to tell you a story my brother told me once about a high school classmate, an M1A1 Abrams tank and a whitetail deer. It is apparently not normal in polite company to discuss the disintegration of adorable animals, and so I won’t unless you buy me a drink (Lagavulin and water, please). What I will do is highlight that the often-overlooked pitfall of the tinkering mentality is the tendency to use very big tools to accomplish very small things, for which the intended aim is almost always overwhelmed by the unintended consequence. Pointing a 120mm smoothbore cannon at a tiny animal isn’t going to shrink the explosion it causes. Likewise, pointing a major change in risk posture or asset allocation at an event we’re a bit nervous about isn’t going to change the fact that we’ve made a change to some very fundamental characteristics of the portfolio.

This happens all the time.

In the last year, I’ve met with advisors, allocators and investors convinced of the inexorable, unstoppable, indomitable rise of interest rates who exited their government and investment-grade bond portfolios — in many cases, the only remnant of their portfolio standing against them and a downturn in risk assets — in favor of higher yielding equity portfolios that wouldn’t be as exposed to the environment they expect. I’ve seen investors leaving passive equity allocations in favor of concentrated private deals because they are concerned about the broader economy’s impact on stocks. I’ve seen investors switch asset classes because they didn’t like the manager they were invested with.

There may be reasons for some of those views, and in some cases even for acting on them. But I am always concerned when I see changes like that unaccompanied by consideration of the magnitude of the unintended consequences: are we still taking the right amount of risk? Are we achieving adequate diversification? As we close out the list of Things that Don’t Matter, I look forward to publishing our list of things that actually DO, because these questions play prominently. There is hope. There are things we can do, and most of them will run contrary to our instincts to take rapid, “nimble” action in our portolios.

Within that thread of hope, a plea first to readers who prefer poetry: that we feel disillusioned or confused about the outcomes for markets does not mean we ought to be more active, more nimble in modifying our asset allocation, however good and wise those things sound when we say them to ourselves and our clients. All the data tell us that we are likely to find ourselves warming our hands at a watchfire before long. To those who prefer poker: you don’t have to play the game. It is OK to step away from the table, walk back to the elevator bank and call it a night, to take what the market gives us.

Make no mistake: the alternative is worse. It’s an expensive alternative. It’s often a risk-additive alternative. It’s a tax-producing alternative. It’s an alternative that frankly most of us just aren’t in a position to successfully execute. There is a reason that most global macro and GTAA hedge funds hire traders who have success in individual markets, even individual types of trading strategies within individual markets. It’s because being able to effectively determine when to switch among managers, among asset classes and among drivers of risk and return is very, very hard. The data bear this out, and no matter how hard we feel like we have to do something, it won’t change the fact that lighting our house on fire isn’t going to make the sun come back.

Understanding the dominant impact of narratives in markets today doesn’t mean abandoning our well-designed processes and our work determining asset allocation, risk targets and portfolio construction in favor of a haphazard chasing of the narrative-driven theme for the day. It means that human behavior and unstructured forms of information should — must — increasingly play a role in the structure of each of those processes in the first place.

After all, all investing is behavioral investing. Anyone who tells you different is either incompetent, selling something or both. One of the most pointless such behaviors — our unquenchable desire to act — nearly completes our list of the Things that Don’t Matter.

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/16058/

Mo’ Compute Mo’ Problems (by Silly Rabbit)

Hard problems

Someone tweeted this cartoon at me last week, presumably in angry response to an Epsilon Theory post, as the Tweet was captioned “My feelings towards ‘A.I.’ (and/or machine learning) and investing”:

Source: xkcd

To be clear: YES, I AGREE

Unsurprisingly, we humans are pretty competent creatures within the domains we have contrived (such as finance) and spent decades practicing. So it is, generally, still hard (and expensive) in 2017 to quickly build a machine which is consistently better at even a thin, discrete sliver of a complex, human-contrived domain.

The challenge, as this cartoon humorously alludes to, is that it is currently often difficult (and sometimes impossible) to know in advance just how hard a problem is for a machine to best a human at.

BUT, what we do know is that once an ML/AI-driven machine dominates, it can truly dominate, and it is incredibly rare for humans to gain the upper hand again (although there can be periods of centaur dominance, like the ‘Advanced Chess’ movement).

As a general heuristic, I think we can say that tasks at which machines are now end-to-end better have one or some of the following characteristics:

  • Are fairly simple and discrete tasks which require repetition without error (AUTOMATION)
  • and/or are extremely large in data scale (BIG DATA)
  • and/or have calculation complexity and/or require a great deal of speed (BIG COMPUTE)
  • and where a ‘human in-the-loop’ degrades the system (AUTONOMY)

But equally there are still many things on which machines are currently nowhere close to being able to reach human-parity, mostly involving ‘intuition’, or many, many models with judgment on when to combine or switch between the models.

Will machines eventually dominate all? Probably. When? Not anytime soon.

The key, immediate, practical point is that the current over-polarization of the human-oriented and machine-oriented populations, particularly in the investing world, is both a challenge and an opportunity as each sect is not fully utilizing the capabilities of the other. Good Bloomberg article from a couple of months back on Point72 and BlueMountain’s challenges in reconciling this in an existing environment.

The myth of superhuman AI

On the other side of the spectrum from our afore-referenced Tweeter are those who predict superhuman AIs taking over the world.

I find this to be a very bogus argument in anything like the foreseeable future, reasons for which are very well laid out by Kevin Kelly (of Wired, Whole Earth Review and Hackers’ Conference fame) in this lengthy essay.

The crux of Kelly’s argument:

  • Intelligence is not a single dimension, so “smarter than humans” is a meaningless concept.
  • Humans do not have general purpose minds and neither will AIs.
  • Emulation of human thinking in other media will be constrained by cost.
  • Dimensions of intelligence are not infinite.
  • Intelligences are only one factor in progress.

Key quote:

Instead of a single line, a more accurate model for intelligence is to chart its possibility space. Intelligence is a combinatorial continuum. Multiple nodes, each node a continuum, create complexes of high diversity in high dimensions. Some intelligences may be very complex, with many sub-nodes of thinking. Others may be simpler but more extreme, off in a corner of the space. These complexes we call intelligences might be thought of as symphonies comprising many types of instruments. They vary not only in loudness, but also in pitch, melody, color, tempo, and so on. We could think of them as ecosystem. And in that sense, the different component nodes of thinking are co-dependent and co-created. Human minds are societies of minds, in the words of Marvin Minsky. We run on ecosystems of thinking. We contain multiple species of cognition that do many types of thinking: deduction, induction, symbolic reasoning, emotional intelligence, spacial logic, short-term memory, and long-term memory. The entire nervous system in our gut is also a type of brain with its own mode of cognition.

(BTW: Kevin Kelly has led an amazing life – read his bio here.)

Can’t we just all be friends?

On somewhat more prosaic uses of AI, the New York Times has a nice human-angle on the people whose job is to train AI to do their own jobs. My favorite line from the legal AI trainer: “Mr. Rubins doesn’t think A.I. will put lawyers out of business, but it may change how they work and make money. The less time they need to spend reviewing contracts, the more time they can spend on, say, advisory work or litigation.” Oh, boy!

Valley Grammar

And finally, because it it just really tickles me in a funny-because-it’s-true way: Benedict Evans’ @a16z’s guide to the (Silicon) Valley grammar of IP development and egohood:

  • I am implementing a well-known paradigm.
  • You are taking inspiration.
  • They are rip-off merchants.

So true. So many attorney’s fees. Better rev up that AI litigator.

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/16065/

Westworld

Bernard Lowe: Last question Dolores. What if I told you that you were wrong? That there are no chance encounters? That you and everyone you know were built to gratify the desires of the people who pay to visit your world? The people you call the newcomers.

Bernard Lowe: What if I told you that you can’t hurt the newcomers? And that they can do anything they want to you?

“Westworld” (2016)

Bernard Lowe:  Go ahead, erase my sentience, mnemonic evolution …

Dr. Robert Ford:  Ah, yes … Such clinical language. I would prefer the more narrative voice. Bernard walked over to Clementine.

[Bernard walks to Clementine]

Dr. Robert Ford:  He took the pistol from her hand.

[Bernard takes the pistol out of Clem’s hand]

Dr. Robert Ford:  Overcome with grief and remorse, he presses the muzzle to his temple, knowing that as soon as Dr. Ford left the room, he would put an end to this nightmare once and for all.

Bernard Lowe:  Don’t do this.

Dr. Robert Ford:  I have a celebration to plan, and a new story to tell.

Bernard Lowe:  Robert.

Dr. Robert Ford:  I’ve told you, Bernard. Never place your trust in us. We’re only human. Inevitably, we will disappoint you.

Dr. Robert Ford:  Goodbye, my friend.

[Ford leaves the room and starts walking away. In the background, blurry, Bernard stands still, gun to his own head. A shot is heard, and he falls.]

“Westworld” (2016)

Billy Kwan:  In the West, we want answers for everything. Everything is right or wrong, or good or bad. But in the [shadow play]

Billy Kwan:   no such final conclusion exists.

Billy Kwan:  Look at Prince Ajuna. He’s a hero. But he can also be fickle and selfish. Krishna says to him, “All is clouded by desire, Ajuna, as a fire by smoke, as a mirror by dust. Through these, it blinds the soul.

“The Year of Living Dangerously” (1982)

Sukarno never had a chance. And yes, that’s Linda Hunt as Billy Kwan.

Michael Corleone:  I saw a strange thing today. Some rebels were being arrested. One of them pulled the pin on a grenade. He took himself and the captain of the command with him. Now, soldiers are paid to fight; the rebels aren’t.

Hyman Roth:   What does that tell you?

Michael Corleone:   They could win.

Hyman Roth:  This county’s had rebels for the last fifty years— it’s in their blood, believe me, I know. I’ve been coming here since the ’20s. We were running molasses out of Havana when you were a baby — the trucks, owned by your father.

Hyman Roth:  Michael, I’d rather we talked about this when we were alone. The two million never got to the island. I wouldn’t want it to get around that you held back the money because you had second thoughts about the rebels.

― “The Godfather: Part II” (1974)

Michael Corleone is like me and every investor over the past five years who held off on an attractive investment for fear of political risk. Except he was right and I’ve been nothing but wrong.

Somehow, I think Silicon Valley got even more spun up than Manhattan. There were hedge fund people I spoke to about a week after the election. They hadn’t supported Trump. But all of a sudden, they sort of changed their minds. The stock market went up, and they were like, ‘Yes, actually, I don’t understand why I was against him all year long.’

― Peter Thiel, in a New York Times interview (January 11, 2017)

Everyone loves a solid gold telephone.

All great literature is one of two stories; a man goes on a journey or a stranger comes to town.

― Leo Tolstoy (1828 – 1910)

Grigory Vakulinchuk: We’ve had enough rotten meat! Even a dog wouldn’t eat this! It could crawl overboard on its own!

Smirov, the ship doctor:  These aren’t worms. They are dead fly larvae. You can wash them off with brine.

“Battleship Potemkin” (1925)

Spoiler Alert: Grigory is shot and killed by management. But the baby makes it down the steps alive. Eisenstein is never entirely clear about the efficacy of the whole just-wash-your-dead-fly-larvae-off-with-brine thing.

John Wick:  People keep asking if I’m back and I haven’t really had an answer. But now, yeah, I’m thinkin’ I’m back.

“John Wick” (2014)

Me, too. Political risk, though, not so much.

If political parties in Western democracies were stocks, we’d be talking today about the structural bear market that has gripped that sector. Show me any country that’s had an election in the past 24 months, and I’ll show you at least one formerly big-time status quo political party that has been crushed. This carnage in status quo political systems goes beyond what we’d call “realigning elections”, like Reagan in 1980 converting the formerly solid Democratic Southern states to a solid Republican bloc. It’s a rethinking of what party politics MEANS in France, Italy, and the United States (and with the UK, Spain, the Netherlands, and maybe Germany not too far behind). The last person to accomplish what Emmanuel Macron did in France? The whole “let’s start a new political party and win an election in two months” thing? That would be Charles de Gaulle in 1958 and the establishment of the Fifth Republic. The last person to accomplish what Donald Trump did in the U.S.? The whole “let’s overthrow an old political party from the inside and win an election in two months” thing? I dunno. Never? Andrew Jackson?

1999 v2.0

On episode 21 of the Epsilon Theory podcast, Dr. Ben Hunt is joined by Brad McMillan, CFA, CAIA, the chief investment officer at Commonwealth Financial Network®. Brad graciously hosts us at Commonwealth’s headquarters in Waltham, Massachusetts. Ben and Brad talk about their mutual love for Terry Pratchett, narrative causality, the French elections, and how technology is changing the financial advisory business.

2016-07-et-podcast-itunes 2016-07-et-podcast-gplay 2016-07-et-podcast-stitcher

Future Flash Crashes, Digital Darwinism & the Resurgence of Hardware (by Silly Rabbit)

Future flash crashes

Remember a few years back when a bogus AP tweet instantly wiped $100bn off the US markets? In April 2013 the Associated Press’ Twitter account was compromised by hackers who tweeted “Breaking: Two Explosions in the White House and Barack Obama is injured.”

For illustrative purposes only.

Source: The Washington Post, 04/23/13, Bloomberg L.P., 04/23/13.

The tweet was quickly confirmed to be an alternative fact (as we say in 2017), but not before the Dow dropped 145 points (1%) in two minutes.

Well, my view is that we are heading into a far more ‘interesting’ era of flash crashes of confused, or deliberately misled, algorithms. In this concise paper titled “Deceiving Google’s Cloud Video Intelligence API Built for Summarizing Videos”, researchers from the University of Washington demonstrate that by inserting still images of a plate of noodles (amongst other things) into an unrelated video, they could trick a Google image-recognition algorithm into thinking the video was about a completely different topic.

Digital Darwinism

I’m not sure I totally buy the asserted causality on this one, but the headline story is just irresistible: “Music Streaming Is Making Songs Faster as Artists Compete for Attention.” Paper abstract:

Technological changes in the last 30 years have influenced the way we consume music, not only granting immediate access to a much larger collection of songs than ever before, but also allowing us to instantly skip songs. This new reality can be explained in terms of attention economy, which posits that attention is the currency of the information age, since it is both scarce and valuable. The purpose of these two studies is to examine whether popular music compositional practices have changed in the last 30 years in a way that is consistent with attention economy principles. In the first study, 303 U.S. top-10 singles from 1986 to 2015 were analyzed according to five parameters: number of words in title, main tempo, time before the voice enters, time before the title is mentioned, and self-focus in lyrical content. The results revealed that popular music has been changing in a way that favors attention grabbing, consistent with attention economy principles. In the second study, 60 popular songs from 2015 were paired with 60 less popular songs from the same artists. The same parameters were evaluated. The data were not consistent with any of the hypotheses regarding the relationship between attention economy principles within a comparison of popular and less popular music.

Meanwhile, in other evolutionary news, apparently robots have been ‘mating’ and evolving in an evo-devo stylee. DTR? More formal translation: Researchers have added complexity to the field of evolutionary robotics by demonstrating for the first time that, just like in biological evolution, embodied robot evolution is impacted by epigenetic factors. Original Frontiers in Robotics and AI (dense!) paper here. Helpful explainer article here.

The resurgence of hardware

As we move from a Big Data paradigm of commoditized and cheap AWS storage to a Big Compute ­­paradigm of high performance chips (and other non-silicon compute methods), we are discovering step-change innovation in applied processing power driven by the Darwinian force of specialization, or, as Chris Dixon recently succinctly tweeted: “Next stage of Moore’s Law: less about transistor density, more about specialized chips.”

We are seeing the big guys like Google develop their specialized chips custom-made for their specific big compute needs, with a very significant increase of speed of up to 30 times faster than today’s conventional processors and using much less power, too.

Also, we are seeing increased real-world applications being developed for truly evolutionary-leap technologies like quantum computing. MIT Technology Review article on implementing the powerful Grover’s quantum search algorithm here.

And, finally, because it just wouldn’t be a week in big compute-land without a machine beating a talented group of humans at one game of another: Poker-Playing Engineers Take on AI Machine – And Get Thrashed.

Key points:

  1. People have a misunderstanding of what computers and people are each good at. People think that bluffing is very human, but it turns out that’s not true. A computer can learn from experience that if it has a weak hand and it bluffs, it can make more money.
  2. The AI didn’t learn to bluff from mimicking successful human poker players, but from game theory. Its strategies were computed from just the rules of the game, not from analyzing historical data.
  3. Also evident was the relentless decline in price and increase in performance of running advanced ‘big compute’ applications; the computing power used for this poker win can be had for under $20k.

PDF Download (Paid Subscription Required): http://www.epsilontheory.com/download/16079/