Pols/Polls

When I was just out of college, Sunday night was my time for laundry. After the sun went down, I would run the washer and dryer (we had machines in the apartment I was sharing) and then iron my dress shirts. At the time, I lived in the greater Los Angeles area (right on the Orange County/Los Angeles County line). As I ironed, I listened to Rodney on the ROQ and Loveline. The former I considered to be, as a self-appointed music connoisseur, an important part of keeping on top of the avant-garde of rock culture. The latter was a secret, guilty pleasure.

Loveline‘s format was live, call-in radio where teens and young adults could have their relationship questions answered. An up-to-date Dr. Ruth for Generation X. I remember, now, one particular type of call. The caller set forth on a long, rambling tale of various problems in his life. Sometimes the problems were mundane and sometimes a bit absurd. Such calls seemed to happen a couple times a month, so this isn’t a particular caller that I am remembering, but the pattern. The “funny man” host, at that time it was Riki Rachtman, would toy the caller, usually making vulgar jokes, especially if the story was particularly amusing. In some of the more serious narratives, the caller left simply to spin his yarn. At some point, though, the “straight man” host – Drew Pinksy aka Dr. Drew – would cut the caller off and, seemingly unrelated to the story being told, ask “how often do you smoke pot every day?”

The caller would pause. Then, after a short but awkward silence, he would answer, “Not that much.”

“How much is ‘not that much’?” queries Dr. Drew.

“Oh, nine or ten times a day, I guess.”

The first time or two, it amazed me. It was like a magic trick. In a story that had absolutely nothing to do with drug use, or impairment, or anything implying marijuana, the Doctor would unerringly pull a substance abuse problem from out of the caller’s misfortune. The marijuana issues were obvious because they so often repeated, but the Dr. Pinksy’s ability to diagnose based serious issues from on seemingly scant facts ranged more broadly. After much, much ironing, I realized that as important as our individuality and uniqueness is to our identities, we humans are dreadfully predictable creatures, especially when we come under stress. Few of us get the chance to see it and few of us would be willing to admit it, but you get us in the right context and our behavior becomes disturbingly predictable.

Eventually, my compensation improved and I invested the early returns into getting my work shirts professionally pressed. Rachtman would fall out with Loveline co-host Adam Corrolla and subsequently found a home in professional wrestling. Corrolla and Pinsky took Loveline to MTV so that non-Angelenos could benefit from their wisdom. I never saw the television version of the show, nor listened to any version of it since they left KROQ. In fact, I don’t believe I’ve heard from Rachtman, Corrolla, or Dr. Drew since they were all together on Sunday night on my stereo.

So why do I dredge up old memories now?

I read three articles, all in about a 12 hour period, and together they got me thinking.

Article #1 is from a political blogger. He predicts a solid Trump win come November. He cites data on absentee ballot receipts relative to demographics and early-but-dramatic deviations from the predictions. The patterns, he explains, show that Trump voters are more active than expected while Biden voters are less involved. Because of the correlation between early voting and support for the Democrats, this early data might prove to be decisive.

Article #2 is a column from Peggy Noonan in the Wall St. Journal. Overall, the article is a self-congratulatory piece where she sees her predictions for a Biden win coming to fruition. Noonan is a lifetime Republican (she was a speechwriter for Ronald Reagan) but she has been anti-Trump from the get-go. Once it became clear that Trump would be nominated by Republicans for a second term, her support has focused on Joe Biden. Alone among those vying for the Presidency, at least to her, he represented the politics that she was used to – before Bush-Gore, before the Tea Party, before Donald Trump. She predicted that the vast middle of the American political body would gravitate to the old and the known and she now sees that she was proven right. As evidence, she cites polling data among college-educated women. The data say that this demographic has shifted so dramatically against President Trump that the result will be not just a Biden win, but a Biden landslide. A one-sided massacre, the likes of which should be entirely impossible in this hopelessly divided nation.

#3 is about State races. The bottom half of the ticket gets mostly ignored by the media and yet, if you’re a voter in American elections, this is where you have your best chance to influence policy. For most of us, our vote for President is already cast, whether we’ve voted early or not. We live in States where the outcome has been known since before the conventions and so, whatever our individual preference, we see which way our electors’ votes are going to fall. Even in a “battleground states” each voter is but one check mark in a sea of millions. The odds that your vote could decide the outcome are astronomically small. Contrast that with the election of State Representatives. There, vote totals are in the thousands, not the millions, and elections can be decided by a handful of votes. Add in a little pre-election advocacy, and the average citizen can have a real influence on the outcome. The lower house of State Government might seem like small potatoes compared to U.S. Senate, but States do have power and small elections occasionally produce big outcomes.

Article #3 presented polling data on State House races, making it one of the few that has been or will be written. The polling outfits aren’t particularly interested in these low level races because the public doesn’t show much interest. Furthermore, the calculus is considerably different than that which drives the national races and, often, it takes a politically savvy local person to understand the nuance. In the media’s defence, the biggest factor in many of these smaller races is what happens “at the top of the ticket.” Pro-Trump/anti-Trump sentiment is going to determine far more elections than the unique issues that impact Backwoods County in some smaller state. In fact, the data cited in this write-up was about the disparity in down-ballot voting between parties. Based on responses, Republicans look to be considerably less likely to vote for “their” State Representative candidates than Democrats.

The reason I saw this article was that it was being heavily criticized on social media – and criticized unfairly, in my opinion. First of all, in the macro sense, the article identified the top two predictors of State races, albeit obliquely (it was just poll data, no predictions of electoral results). What’s going to decide the close races at the State level is the relative turnout of pro- and anti- Trump voters, plus the motivation of those voters to consider all the other races that are on their ballots. However, the main complaint from the critics was of the polling methodology. The poll sample was just over 1000 respondents. How crazy is it, said the critics, trying to predict dozens and dozens of races from a poll which spoke to so few voters – maybe a handful from any given district containing thousands who will be voting next month?

It is this last bit, especially, made me think of Dr. Drew.

Before I get to that, though, let us ponder statistical methods for a second. When I first encountered some real-world applications of sampling and prediction, I was shocked with the rather small amount of collection that is necessary to model large amounts of data. If you know you have a standard distribution (bell curve), but you need to determine its parameters, you need only a handful of points to figure it all out. Another few points will raise your confidence in your predictions to very high levels. The key, of course, is to know what the right model is for your data. If your data are bell-curve like, but not strictly a standard distribution, your ability to predict is going to be lower and your margin of error is going to be higher, even after collecting many extra samples. If your data are not-at-all in a standard distribution, but you choose, anyway, to model it so (maybe not the worst idea, really), you might see large errors and decidedly wrong predictions. This is still all well understood. Much science and engineering has gone into designing processes such as the sampling of product for quality assurance purposes. We know how to minimize sampling and maximize production at a consistent quality.

But what about people? They are complex, unpredictable, and difficult to model, aren’t they? Can you really ask 1000 people what they think an use it to guess how millions of people are going to vote? Well, if you’re Dr. Drew, you’d know that people are a lot more predictable than we think we are. Behaviors tend to correlate and that allows a psychiatrist, a family physician, or maybe even a pollster to know what you are going to do before you do yourself. Furthermore, we are talking about aggregate outcomes here. I may have a hard time predicting whom you would vote for but, give me your Zip Code, and I can probably get a pretty accurate model of how you plus all your neighbors will vote.

That model, the underlying assumptions that we make about the data, is the key to accuracy and even validity. Is my sample random enough? Should it be random, or should it match a demographic that corresponds to voter turnout? If the latter, how do I model voter turnout? The questions go on and on and help explain why polling is done by a handful of organizations with long experience at what they do. If you really, really understand the underlying data, though, a very small sample will very accurately predict the full outcome. Maybe I only have to talk to married, college-educated women, because I know that the variation in their preferences will determine the election. Maybe all I need is the Zip Codes from absentee ballot returns. Or maybe, after I produce poll-after-poll with a margin-of-error of a percent or two, I’ll wind up getting the election outcome spectacularly wrong.

This is a fascinating time for those in the business of polling. Almost nobody was even close when it came to predicting the 2016 Presidential Election. Some of that was the personal bias of those who do the polling. I’d like to think that, more often than not, it was more often bad modeling of voters which led to honest, albeit rather large, mistakes. Part of me would really, really like to see inside these models. Not, as one might imagine, so I could try to predict the election results myself. Rather, I’d like to see how the industry is dealing with their failure last time around and how have they adjusted the processes (amidst very little basis for doing so) to try to get this election right. When I see simultaneous touting of both a Trump landslide and a Biden landslide, I know that somebody has got to be wrong. Is anybody about to get it right? If they are, how are they doing it?

This is something I’d like to understand.

Case/Point

My writing isn’t always as clear as it could be. Fortunately, the news of the day has provided me with an illustration of what I’m talking about.

I’ve seen other writers address the same point that I was making in my previous post. Many of them have done it far better than I did – or could. I saw graphic on-line that compared the “curve” for COVID-19 with a much bigger curve labeled Famine. Yesterday morning, the Wall St. Journal printed an editorial about why inflation might be a problem this time around (unlike 2008). This echos, albeit distantly, the on-line dread of “inevitable” hyper-inflation that must result from our current massive infusion of borrowed/printed money. I’ve seen other on-line discussion analyzing the rate of closures among restaurants and projecting that out to estimate economic damage. I should welcome all these opinions as those of fellow travelers because the majority of the articles that I read are focused on how things are already turning around and how quickly the economy will recover. Yet I feel like even those with whom I agree the most with are missing the real issue at play here.

Some, any, or all of the economic red flags might be at play now, soon, or maybe not until later. Or maybe not. My point was that there is simply no way to know.

So, here’s what happened instead.

“But,” you must be thinking, “this has nothing to do with viruses or restaurant shutdowns.” It does, though. Everything we do as a society is interrelated and nothing happens in isolation. Would a police officer have killed a man in broad daylight, on camera, if it weren’t for the environment that “quarantine” and “lock-down” engendered? I have a hard time believing that this is the case.

Folks have torched Targets before*, and life goes on. But could it, this time,  be the straw that breaks Target’s humped back when 28 stores shut down amidst government-imposed 50% capacity reductions, supply chain disruptions, and inflationary pressures?

It could be. Or maybe not. There is simply no way to know.

*OK. I don’t actually know whether Target has been the, eh, “target” of looting during past periods of civil unrest. A quick search doesn’t help me. In any case, I mean this metaphorically.

Lead/Lag

In control engineering, the concepts of stability and instability precede the design of the control algorithms themselves. As an engineering student, studying and understanding the behavior of dynamic system is a prerequisite to controlling them. As elementary it is as an engineering concept, it often seems to be beyond our collective comprehension as a society, even if it is well within our individual grasp. While most of us may scratch our heads when looking at a Nyquist Plot, we do understand that you’ve got to start backing off the accelerator even before you get up to your desired speed.

When I was a younger man, I had a roommate who drove a Datsun 280Z. This was a nifty little car. Price-wise, it was accessible to the young, the working class, without breaking the bank like. Not only was it affordable to buy but it was affordable to own. It was a pleasure to drive and, top it all off all, it was fairly impressive performance-wise. I nice example of the “roadster” style.

We took a few summer road trips together and, on more than one occasion, he asked me to drive so he could take a nap, sober up, or just monkey with his cassette deck. Before letting me drive for the first time, he gave me a bit of a run down on his car’s performance. “Don’t go over 90 mph,” he warned me, as the car would shake violently shortly after crossing the 90 mark on the speedometer. Needless to say, the car never shook itself apart while I was driving. Perhaps that was because, as a conscientious and responsible youth, I would never exceed the speed limit. Perhaps it was because it he wasn’t entirely correct about the circumstances under which his car would have vibration issues.

My point is, most of us have a gut understanding of frequency response and stability and the struggles of controlling it. The seriousness of the problem is exposed in the design of mechanical systems and, in particular, those that incorporate high-frequency rotation of components. Deeper understanding and mathematical analyses are necessary prerequisites to assembling a piece of machinery that will hurtle through the night at speeds approaching 100 mph. In the case of my friend’s Datsun, as cyclic energy is induced into a system, it is possible for those inputs to resonate in the spring-like manifestations of the system’s passive structure. Without proper analysis and design, a vehicle’s suspension system might well start to exhibit extreme vibration at high speeds. The same applies to any dynamic system. We are all familiar with the violently-shaking washing machine, whether we have one in own home or not.

Naturally, the mathematics apply to non-mechanical systems as well. Often the effects are far more serious than a shaking car or a jumping washing machine. In electric circuits, resonance can produce seemingly impossibly-high voltages and currents. Water hammer in a hydraulic system can crush equipment and cause explosions. The analyses that help us understand these physical phenomenon, I’ll argue today, would also help us understand interactions in social systems and the effect of a “black swan event,” if we allow them to.

It’s the Stupid Economy

The sometimes-sizeable gap between “gut feel” and mathematical certainty is particularly common to complex systems. Coincidentally, our body politic is eager to tackle the most complex of systems, attempting to control them through taxation and regulation. The global climate and national economies seem to be a recent, and often interconnected, favorite. I shall leave the arguments of climate science and engineering to others and, today, focus on the economy. When it comes the politics of the economy, I have noticed a pattern. When it comes to the intersection of economics and politics, the thinking is shockingly short-term. Shocking, because the economic environment may be the number one predictor for the outcome of an election. A good economy strongly favors the incumbents whereas economic misery almost guarantees a changing of the guard. You would think that if the economic conditions are what matter most to us, when it comes to our one contribution to the governance of society, we’d be eager to get it right. Yet, what seems matter most are the economic conditions on the day of the polling. Four years of economic growth doesn’t mean much if the economy tanks on the 30th of October.

In something of a mixed blessing, the recent political free-for-all has challenged this shortsightedness, at least somewhat. President Obama, for years, blamed his predecessor for recession and deficit spending, despite a negative economic climate persisting for years into his term. He even famously took credit for the positive economic indicators during his successor’s term. His opponents, of course, sought to do the opposite. The truth is far more nuanced than any care to admit, but at least popular culture is broaching the subject. Most of us know that if “the economy” is looking up the day after the President signs off on a new initiative, it wasn’t his signature that did it. Or, more accurately, it can’t possibly account for the entirety of the impact, which may take months or years to reveal its full effect.

Going Viral

We have a further advantage when it comes to talking about the interaction between the economy, the novel coronavirus, and the resultant economic shutdown. The media has inundated us with bell curves and two-week lags. Most of us can appreciate the math that says if a Governor closes bars and restaurants today, we shouldn’t yet be looking for the results in our statistics tomorrow. Nonetheless, our collective grasp of dynamic systems and probabilities is tenuous under the best of circumstances. Mix in high levels of fear and incessant media hype, and even things that should be obvious become lost in the surrounding clamor.

Shift the playing field to economics and the conversation gets even murkier.

“The economy” is at the high-end of complex and chaotic systems. It is, after all, not an external system that we can observe and interact with, nor is it subject to the laws of physics. Rather, it is the collective behavior all of us, each and every individual, and how we interact with each other to produce, consume, and exchange. Indeed, one might speculate on where the boundaries really lie. It seems a bit insensitive label everything as “economic activity” during a health crisis, but what is it that we can exclude? Waking and sleeping, each of us are in the process of consuming food, water, clothing, and shelter. Most social interactions also involve some aspect of contract, production, or consumption. Even if we can isolate an activity that seems immune to all that, all human activity still occurs within that structure that “society,” and thus “the economy, provides.

Within that framework, anyone who claims to “understand” the economy is almost certainly talking about a simplified model and/or a restricted subset of economic activity. Either that, or they are delusional. Real economic activity cannot be understood. Even if the human mind was vastly more capable, the interaction of every human being on the planet is, quite simply, unpredictable. Because of this, we use proxies for economic activity as a way to measure health and the effects of policy. GDP and GDP-growth are very common. Stock market performance substitutes for economic health in most of our minds and in the daily media. Business starts, unemployment numbers, average wages – each of these are used to gauge what is going on with the economy. However, every one of these metrics is incomplete at best and, more often than not, downright inaccurate in absolute terms.

Of course, it isn’t quite as bad as I make it out to be. GDP growth may contain plenty of spurious data, but if we seek to understand what is included and not included, and apply it consistently, we can obtain feedback that guides our policymaking. For example, we could assume that noisy prices associated with volatile commodities are not relevant to overall inflation numbers, or we can exclude certain categories when calculating GDP for the purpose of determining inflation. As long as we’re comparing apples to apples, our policy will be consistent.

What happens, though, when we get the economic equivalent of a hydraulic shock? In this case, our models of the economic world no longer apply and the world enters into an entirely unpredicted and unpredictable realm. We know this. What I want to explore, however, is what happens to our ability to “control” that system. My guess is it fails. It fails because we, again collectively, don’t appreciate the characteristics of dynamic systems. Yes, we understand it in terms of the heuristics we’ve traditionally used. Interest rates have to be raised before inflation kicks in to keep it from spiraling out of control. But what inflation will result from a $5 trillion stimulus at a time of 30% unemployment? Do we need higher or lower interest rates. In other words, when our traditional metrics fail us, will we truly appreciate the complex nature of the system?

In Control

During our imposed down-time, I re-watched an excellent film about the now-10-plus-year-old financial crisis induced by the housing market. The film The Big Short was made in 2015 based on Michael Lewis’ 2010 book of the same name. It dramatizes the subprime housing market collapse as seen by a handful of investors who saw it coming. As much as the story seems, today, in our distant past, there are those among us who feel that what we witnessed in 2008 was just the opening chapters in a longer tale. Whether a housing crisis is our past or our future, there are lessons to be applied to the present day.

The film’s story opens in 2005. Investor Michael Burry, reading the details of mortgage-backed security prospectuses, determines that the housing market is unstable and the financial instruments built upon it are doomed to fail. Unable to take a contrarian financial position using existing instruments, he commissions the creation the Credit Default Swap to allow him to bet that the mortgage market will fail. The film concludes when Burry, and several others who bet against the housing market, liquidate their positions at a profit, sometime after Spring, 2008. The real-life Burry had actually been analyzing data from 2003 and 2004 before making his predictions and his commitment. Burry later wrote a piece for the New York Times saying that the housing market failure was predictable as much as four or five years out.

Putting this another way, by 2004 or 2005, the massive financial crisis of 2008-2010 had already happened, we just didn’t realize it yet. One might argue that sometime in those intervening four years, sanity might have come over America’s banks and the prospective home-owners to whom they were lending, but of course it didn’t. The reasons are many why it didn’t; why perhaps it couldn’t. Thus the events that all-but-inevitably put us on the road to global financial advance happened [four, five, more?] years in advance of what we would consider the start of the crisis. Unemployment numbers didn’t recover until 2014. That implies that for the individual, perhaps someone becoming unemployed and being unable to find a new position circa 2014, the impact of the collapse may have taken more than a decade to manifest itself.

Again, lets look at it from a different angle. Suppose I wanted to avoid the tragedy to that individual who, in 2014, became unemployed. Let’s imagine that, as a result of his lack of employement, he died. Maybe it was suicide or opioid addiction. Maybe the job loss turned into a home loss and his whole family suffered. Suppose as a policy maker, I wanted to take macro-economic action to avoid that unnecessary death. How soon would I have had to act? 2003? 2000? Sometime in the 1990s?

Next Time

All of this comes to mind today as a result of the talk I am seeing among my fellow citizens. People are angry, although that isn’t entirely new. Some are angry because their livelihoods have been shut down while others are angry that folks would risk lives and health merely to return to those livelihoods. In the vast majority of cases, however, the fear is about near term effects. Will my restaurant go bankrupt given the next few weeks or months of cash-flow? What will the virus do two weeks after the end of lockdown? Will there be a “second wave” next fall? A recent on-line comment remarked that, although the recovery phase would see bumps along the road, “We’ll figure it out. We always do.”

Statistically, that sentiment is broadly reflected in the population at large. A summary of poll data through the end of March (http://www.apnorc.org/projects/Pages/Personal-impacts-of-the-coronavirus-outbreak-.aspx) suggested similar thinking. A majority of those currently out-of-work see no problems with returning “once it’s over.” In fact, a majority figure that by next year they’ll be as good as or better off financially than they are now. Statements like “we’ll get through this and come out stronger than ever” can be very motivational, but extending that to all aspects of economic and financial health seems a bit blind.

We’re losing track the macro-economic implications for the personally experienced trees. We’ve all see the arguments. Is it better to let grandpa die so that the corner burger shack can open back up a few weeks earlier? The counter argument cites the financial impact of a keeping the economy mostly-closed-down for a few more weeks. This isn’t the point, though, is it? On all sides of the argument it seems that the assumption is that we can just flip everything back on and get back to business. We are oblivious to the admittedly unanswerable question – how much damage has already been done?

Unprecedented

Word like “historic” and “unprecedented” are tossed around like confetti, but not without reason. In many ways our government and our society have done things – already done things, mind you – that have never happened before in the history of man. At first, the “destruction” seemed purely financial. Restaurants being shut down meant as loss in economic activity; a destruction of GDP. But is that even a real thing? Can’t we just use a stimulus bill to replace what is lost and call it even? But as April turns into May, we’re starting to see stories of real and literal destruction, not just lost opportunity. Milk is dumped because it can’t be processed. Vegetables are plowed under. Beef and chickens are killed without processing. This is actual destruction of real goods. Necessary goods. How can this go away with a reopening and some forgivable loans?

None of the experience gained through the financial crises of my lifetime would seem to apply. Even the Great Depression, while correct in magnitude, seems to miss the mark in terms of methodology. We’re simultaneously looking at a supply shock, a consumer depression, and inflationary fiscal policy. It’s all the different flavors or financial crisis, but all at the same time. Imagine a hydraulic shock in a some rotating equipment where the control system itself has encountered a critical failure. I’ve decided that, for me, the best comparison is the Second World War. Global warfare pulled a significant fraction of young men out of the workforce, many never to return. Shortages ravaged the economy, both through the disruption of commerce as well as the effects of rationing. A sizeable percentage of the American economic output was shipped overseas and blown up; gone.

Yet we got through it. We always do.

But we did so because we were willing to make sacrifices for the good of the nation and the good of the free world. We also lost a lot of lives and a lot of materiel. If “we” includes the citizens of Germany or the Ukraine, the devastation to society and culture was close to total, depending on where they called home. So, yes, civilization came through the Second World War and, as of a year or so ago, were arguably better than ever, but for far too many that “return to normalcy” took more than a generation. Will that be the price we have to pay to “flatten the curve?”

 

 

The not-so-Friendly Skies

This past weekend, the Wall St. Journal published a front-page article detailing the investigation into the recent, deadly crashes of Boeing 737 MAX aircraft. It is a pretty extensive combination of information that I had seen before, new insights, and interviews with insiders. Cutting to the chase, they placed a large chunk of the blame on a regulatory structure that puts too much weight shoulders of the pilots. They showed, with a timeline, how many conflicting alarms the pilots received within a four-second period. If the pilots could have figured out the problem in those four seconds and taken the proscribed action, they could have saved the plane. The fact that the pilots had a procedure that they should have followed means the system fits within the safety guidelines for aircraft systems design.

Reading the article, I couldn’t help but think of another article that I read a few months back. I was directed to the older article by a friend, a software professional, on social media. His link was to an IEEE article that is now locked behind their members-only portal. The IEEE article, however, was a version of a blog post by the author and that original post remains available on Medium.

This detailed analysis is even longer than the newspaper version, but also very informative. Like the Wall St. Journal, the blog post traces the history behind the design of the hardware and software systems that went into the MAX’s upgrade. Informed speculation describes how the systems of those aircraft caused the crash and, furthermore, how those systems came to be in the first place. As long as it is, I found it well worth the time to read in its entirety.

On my friend’s social media share, there was a comment to the effect that software developers should understand the underlying systems for which they are writing software. My immediate reaction was a “no,” and its that reaction I want to talk about here. I’ll also point out that Mr. Travis, the blog-post author and a programmer, is not blaming programmers or even programming methodology per se. His criticism is at the highest level; for the corporate culture and processes and for the regulatory environment which governs these corporations. In this I generally agree with him, although I could probably nitpick some of this points. But first, the question of the software developer and what they should, can, and sometimes don’t understand.

There was a time, in my own career and (I would assume) in the career of the author, that statements about the requisite knowledge of programmers made sense. It was probably even industry practice to ensure that developers of control system software understood the controls engineering aspects of what they were supposed to be doing. Avionics software was probably an exception, rather than the rule, in that the industry was an early adopter of formal processes. For much of the software-elements-of-engineering-systems industry, programmers came from a wide mix of backgrounds and a key component of that background was what programmers might call “domain experience.” Fortran could be taught in a classroom but ten years worth of industry experience had to come the hard way.

Since we’ve been reminiscing about the artificial intelligence industry of the 80s and 90s, I’ll go back there again. I’ve discussed the neural network state-of-the-art, such as it was, of that time. Neutral networks were intended to allow the machines to extract information about the system which the programmers didn’t have. Another solution to the same category of problems, again one that seemed to hold promise, was a category called expert systems, which was to directly make use of those experts who did have the knowledge that the programmers lacked. Typically, expert systems were programs built around a data set of “rules.” The rules contained descriptions of the action of the software relative to the physical system in a way that would be intuitive to a non-programmer. The goal was to be a division of labor. Software developers, experts in the programming, would develop a system to collect, synthesize, and execute the rules in an optimized way. Engineers or scientists, experts in the system being controlled, would create those rules without having to worry about the software engineering.

Was this a good idea? In retrospect, maybe not. While neural networks have found a new niche in today’s software world, expert systems remain an odd artifact found on the fringes of software design. So if it isn’t a good idea, why not? One question I remember being asked way-back-when got to the why. Is there anything you can do with a rule-based system that you couldn’t also implement with standard software techniques? To put it another way, is my implemented expert system engine capable of doing anything that I couldn’t have my C++ team code up? The answer was, and I think obviously, “no.” Follow that with, maybe, a justification about improved efficiencies in terms of development that might come from the expert systems approach.

Why take this particular trip down memory lane? Boeing’s system is not what we’d classify as AI. However, I want to focus on a particular software flaw implicated as a proximate cause of the crashes; the one that uses the pitch (angle-of-attack) sensors to avert stalls. Aboard the Boeing MAX, this is the”Maneuvering Characteristics Augmentation System” (MCAS). It is intended to enhance the pilot’s operation of the plane by automatically correcting for, and thereby eliminating, rare and non-intuitive flight conditions. Explaining the purpose of the system with more pedestrian terminology, Mr. Travis’ blog calls it the “cheap way to prevent a stall when the pilots punch it” system. It was made a part of the Boeing MAX as a way to keep the airplane’s operation the same as how it always has been, using feedback about the angle-of-attack to avoid a condition that could occur only on a Boeing MAX.

On a large aircraft, the pitch sensors are redundant. There is one on each side of the plane and both the pilot and the co-pilot have indicators for their side’s sensor. Thus, if the pilot’s sensor fails and he sees a faulty reading, his co-pilot will still be seeing a good reading and can offer a different explanation for what he is seeing. As implemented, the software is part of the pilot’s control loop. MCAS is quickly, silently, and automatically doing what the pilot would be doing, where he to have noticed that the nose of the plane was rising toward a stall condition at high altitude. What it misses is the human interaction between the pilot and his co-pilot that might occur if the nose-up condition were falsely indicated by a faulty sensor. The pilot might say, “I see the nose rising too high. I’m pushing the nose down, but it doesn’t seem to be responding correctly.” At this point, the co-pilot might respond, “I don’t see that. My angle-of-attack reports normal.” This should lead them to deciding the pilot should not, in fact, be responding from the warning produced by his own sensor.

Now, according the the Wall St. Journal article, Boeing wasn’t so blind as to simply ignore the possibility of a sensor failure. This wasn’t explained in the Medium article, but there are (and were) other systems that should have alerted a stricken flight crew to an incompatible difference in values between the two angle-of-attack sensors. Further, there was a procedure, called the “runaway stabilizer checklist” that was to be enacted under that condition. Proper following of that checklist (within the 4 second window, mind you) would have resulted in the deactivation of the MCAS system in reaction to the sensor failure. But why not, instead, design the MCAS system to either a) take all available, relevant sensors as input before assuming corrective action is necessary or b) take as input the warning about conflicting sensor reading? I won’t pretend to understand what all goes into this system. There are probably any number of reasons; some good, some not-so-good, and some entirely compelling; that drove Boeing to this, particular solution. It is for that reason I led off using my expert system as an analogy; since I’m making the analogies I can claim I understand, entirely, the problem that I’m defining.

Back then, a fellow engineer and enthusiast for technologies like expert systems and fuzzy logic (a promising technique to use rules for non-binary control) explained it to me with a textbook example. Imagine, as we’ve done before, you have a self-driving car. In this case, its self-driving intelligence uses a rule-based expert system for high-level decision making. While out and about, the car comes to an unexpected fork in the road. In computing how to react, one rule says to swerve left and one says to swerve right. In a fuzzy controller, the solution to conflicting conclusions might be to weight and average the two rule outputs. As a result, our intelligent car would elect to drive on, straight ahead, crashing into the tree that had just appeared in the middle of the road. The example is oversimplified to the point of absurdity, but it does point out a particular, albeit potential, flaw with rule-based systems. I also think it helps explain, by analogy, the danger lurking in the control of complex systems when your analysis is focused on discrete functions.

With the logic for your system being made up of independent components, the overall system behavior becomes “emergent” – a combination of the rule base and the environment in which it operates. In the above case, each piece of the component logic dictated swerving away from the obstacle. It was only when the higher-level system did its stuff that the non-intuitive “don’t swerve” emerged. Contrasting rule based development with more traditional code design, the number of possible states may be indeterminate by design. Your expert input might be intended to be partial, completed only when synthesized with the operational environment. Or look at it by way of the quality assuredness problem it creates. While you may be creating the control system logic without understanding the entire environment within which it will operate, wouldn’t you still be required to understand, exhaustively, that entire environment when testing? Otherwise, how could you guarantee what the addition of one more expert rule would or wouldn’t do to your operation?

Modern software engineering processes have been built, to a large extent, based on an understanding that the earlier you find a software issue, the cheaper it is to solve. A problem identified in the preliminary, architectural stage may be trivial. Finding and fixing something during implementation is more expensive, but not as expensive as creating a piece of buggy software that has to be fixed either during the full QA testing or, worse yet, after release. Good design methodologies also eliminate, as much as possible, the influence that lone coders and their variable styles and personalities might have upon the generation of large code bases.  We now feel that integrated teams are superior to a few eccentric coding geniuses. This goes many times over when it comes to critical control systems upon which people’s lives may depend. Even back when, say, an accounting system might have been cobbled together by a brilliant hacker-turned-straight, avionics software development followed rigid processes meant to tightly control the quality of the final product. This all seems to be for the best, right?

Yes, but part of what I see here is a systematization that eliminates not just the bad influences of the individual, but their creative and corrective influence as well. If one person had complete creative control over the Boeing MAX software, that person likely would never have shipped something like the MCAS reliance on only one of a pair of sensors. The way we write code today, however, there may be no individual in charge. In this case, the decision to make the MCAS a link between the pilot’s control stick and the tail rudder rather than an automated response at a higher level isn’t a software decision; its a cockpit design decision. As such it’s not only outside of the purview of software design, but perhaps outside of the control of Boeing itself if it evolved as a reaction to a part of the regulatory structure. In a more general sense, though, will the modern emphasis on team-based, structured coding methodology have the effect of siloing the coders? A small programming team who has been assigned a discrete piece of the puzzle not only doesn’t have responsibility for seeing the big picture issues, those issues won’t even be visible to them.

In other words (cycling back to that comment on my friend’s posting many months ago), shouldn’t the software developers should understand the underlying systems for which they are writing software? Likely, the design/implementation structure for this part of the system would mean that it wouldn’t be possible for a programmer to see that either a) the sensor they are using as input is one of a redundant pair of sensors and/or b) there is separate indication that might tell them whether the sensor they are using as input is reliable. Likewise the large team-based development methodologies probably don’t attract to the avionics software team the programmer who is also a controls engineer who also has experience piloting aircraft – that ideal combination of programmer and domain expert that we talked about in the expert system days. I really don’t know whether this is an inevitable direction for software development or if this is something that is done for better or for worse as we look at different companies. If the latter, the solutions may simply be with culture and management within software development companies.

So far, I’ve mostly been explaining why we shouldn’t point the figure at the programmers, but neither of the articles do. In both cases, blame seems to be reserved for the highest levels of aircraft development; at the business level and the regulatory level. The Medium article criticizes the use of engineered solutions to allow awkward physics as solutions to business problems (increasing the capacity of an existing plane rather than, expensively, creating a new one). The Wall St. Journal focuses on the philosophy that pilots will respond unerringly to warning indicators, often under very tight time constraints and under ambiguous and conflicting conditions. Both articles would tend to fault under-regulation by the FAA, but heavy-handed regulation may be just as much to blame as light oversight. Particularly, I’m thinking of the extent to which Boeing hesitated to pass information to the customers for fear of triggering expensive regulatory requirements. When regulations encourage a reduction in safety, is the problem under-regulation or over-regulation?

Another point that jumped out at me in the Journal article is that at least one of the redesigns that went into the Boeing MAX was driven by FAA high-level design requirements for today’s human-machine interfaces for aircraft control. From the WSJ:

[Boeing and FAA test pilots] suggested MCAS be expanded to work at lower speeds so the MAX could meet FAA regulations, which require a plane’s controls to operate smoothly, with steadily increasing amounts of pressure as pilots pull back on the yoke.

To adjust MCAS for lower speeds, engineers quadrupled the amount the system could repeatedly move the stabilizer, to increments of 2.5 degrees. The changes ended up playing a major role in the Lion Air and Ethiopian crashes.

To put this in context, the MCAS system was created to prevent an instability at some high altitude conditions, conditions which came about as a results of larger engines that had been moved to a suboptimal position. Boeing decided that this instability could be corrected with software. But if I’m reading the above correctly, there are FAA regulations focus on making sure a fly-by-wire system still feels like the mechanically-linked controls of yore, and MCAS seemed perfectly suited to help satisfy that requirement as well. Pushing this little corner of the philosophy too far may have been a proximate cause of the Boeing crashes. Doesn’t this also, however, point to a larger issue? Is there a fundamental flaw with requiring that control systems artificially inject physical feedback as a way to communicate with the pilots?

In some ways, its a similar concern to what I talked about with the automated systems in cars. In addition to the question whether over-automation is removing the connection between the driver/pilot and the operational environment, there is, for aircraft, an additional layer. An aircraft yoke’s design came about because it directly linked to the control surfaces. In a modern plane, the controls do not. Today’s passenger jet could just as well use a steering wheel or a touch screen interface or voice-recognition commands. The designs are how they are to maintain a continuity between the old and the new, not necessarily to provide the easiest or most intuitive control of the aircraft as it exists today. In addition, and by regulatory fiat apparently, controls are required to mimic that non-existent physical feedback. That continuity and feedback may also be obscuring logical linkages between different control surfaces that could never have existed when the interface was mechanically linked to the controlled components.

I foresee two areas where danger could creep in. First, the pilot responds to the artificially-induced control under the assumption that it is telling him something about the physical forces on the aircraft. But what if there is a difference? Could the pilot be getting the wrong information? It sure seems like a possibility when feedback is being generated internally by the control system software. Second, the control component (in this case, the MCAS system) is doing two things at once; stabilizing the aircraft AND providing realistic feedback to the pilots by feel through the control yoke. Like the car that can’t decide whether to swerve right or left, such a system risks, in trying do both, getting neither right.

I’ll sum up by saying I’m not questioning the design of modern fly-by-wire controls and cockpit layouts; I’m not qualified to do so. My questions are about the extent to which both regulatory requirements and software design orthodoxy box in the range of solutions available to aircraft-control designers in a way that limits the possibilities of creating safer, more efficient, and more effective aircraft for the future.

Artificial, Yes, but Intelligent?

Keep your eyes on the road, your hands upon the wheel.

When I was in college, only one of my roommates had a car. The first time it snowed, he expounded upon the virtues of finding an empty and slippery parking lot and purposely putting your car into spins.  “The best thing about a snow storm,” he said. At the time I thought he was a little crazy. Later, when I had the chance to try it, I came to see it his way. Not only is it much fun to slip and slide (without the risk of actually hitting anything), but getting used to how the car feels when the back end slips away is the first step in learning how to fix it, should it happen when it actually matters.

Recently, I found myself in an empty, ice-covered parking lot and, remembering the primary virtue of a winter storm, I hit the gas and yanked on the wheel… but I didn’t slide. Instead, I encountered a bunch of beeping and flashing as the electronic stability control system on my newish vehicle kicked in. What a disappointment it was. It also got me thinkin’.

For a younger driver who will almost never encounter a loss-of-traction slip condition, how do they learn how to recover from a slide or a spin once it starts? Back in the dark ages, when I was learning to drive, most cars were rear-wheel-drive with a big, heavy engine in the front. It was impossible not to slide around a little when driving in a snow storm. It was almost a prerequisite to going out into the weather to know all the tricks of slippery driving conditions. Downshifting (or using those number gears on your automatic transmission), engine breaking, and counter steering were all part of getting from A to B. As a result*, when an unexpectedly slippery road surprises me, I instinctively take my foot off the brakes/gas and counter-steer without having to consciously remember the actual lessons. So does a car that prevents sliding 95% of the time result in a net increase in safety, even though it probably makes that other 5% worse? It’s not immediately obvious that it does.

On the Road

I was reminded of the whole experience a month or so ago when I read about the second self-driving car fatality. Both crashes happened within a week or so of each other in Western states; the first in Arizona and the second in California. In the second crash, Tesla’s semi-autonomous driving function was in fact engaged at the time of the crash and the drivers hands were not on the wheel six seconds prior. Additional details do not seem to be available from media reports, so the actual how and why must remain the subject of speculation. In the first, however, the media has engaged in the speculation for us. In Arizona, it was an Uber vehicle (a Volvo in this case) that was involved and the fatality was not the driver. The media has also reported quite a lot that went wrong. The pedestrian who was struck and killed was jaywalking, which certainly is a major factor in her resulting death. Walking out in front of a car at night is never a safe thing to do, whether or not that car self-driving. Secondly, video was released showing the driver was looking at something below the dashboard level immediately before the crash, and thus was not aware of the danger until the accident occurred. The self-driving system itself did not seem to take any evasive action.

Predictably, the Arizona state government responded by halting the Uber self-driving car program. More on that further down, but first look at the driver’s distraction.

After the video showing such was released, media attention focused on the distracted-driving angle of the crash. It also brought up the background of the driver, who had a number of violations behind him. Certainly the issue of electronics and technology detracting from safe driving is a hot topic and something, unlike self-driving Uber vehicles, that most of us encounter in our everyday lives. But I wonder if this exposes a fundamental flaw in the self-driving technology?

It’s not exactly analogous to my snow situation above, but I think the core question is the same. The current implementation of the self-driving car technology augments the human driver rather then replaces him or her. In doing so, however, it also removes some of the responsibility from the driver as well as making him more complacent about the dangers that he may be about to encounter. The more that the car does for the driver, the greater the risk that the driver will allow his attention to wander rather that stay focused, on the assumption that the autonomous system has him covered. In the longer term, are there aspects of driving that the driver will not only stop paying attention to, but lose the ability to manage in the way a driver of a non-automated car once did?

Naturally, all of this can be designed into the self-driving system itself. Even if a car is capable of, essentially, driving itself over a long stretch of a highway, it could be designed to engage the driver every so many seconds. Essentially requiring unnecessary input from the operator can be used to make sure she is ready to actively control the car if needed. I note that we aren’t breaking new ground here. A modern aircraft can virtually fly itself, and yet some part of the design (plus operational procedures) are surely in place to make sure that the pilots are ready when needed.

As I said, the governmental response has been to halt the program. In general, it will be the governmental response the will be the biggest hurdle for self-driving car technology.

In the specific case of Arizona, I’m not actually trying to second guess their decision. Presumably, they set up a legal framework for the testing of self-driving technology on the public roadways. If the accident in question exceeded any parameters of that legal framework, then the proper response would be to suspend the testing program. On the other hand, it may be that the testing framework had no contingencies built into it, in which case any injuries or fatalities would have to be evaluated as they happen. If so, a reactionary legal response may not be productive.

I think, going forward, there is going to be a political expectation that self-driving technology should be flawless. Or, at least, perfect enough that it will never cause a fatality. Never mind that there are 30-40,000 motor vehicle deaths per year in the United States and over a million per year world wide. It won’t be enough that an autonomous vehicle is safer than than a non-autonomous vehicle; it will have to be orders-of-magnitude safer. Take, as an example, passenger airline travel. Despite a rate that is probably about 10X safer for aircraft over cars, the regulatory environment for aircraft is much more stringent. Take away the “human” pilot (or driver) and I predict the requirements for safety will be much higher than for aviation.

Where I’m headed in all this is, I suppose, to answer the question about when we will see self driving cars. It is tempting to see that as a technological question – when will the technology be mature enough to be sold to consumers? But it is more than that.

I recall see somewhere an example of “artificial intelligence” for a vehicle system. The example was of a system that detected a ball rolling across the street being a trigger for logic that anticipates there might be a child chasing that ball. A good example of an important problem to solve before putting an autonomous car onto a residential street. Otherwise, one child run down while he was chasing his ball might be enough for a regulatory shutdown. But how about the other side of that coin? What happens the first time a car swerves to avoid a non-existent child and hits an entirely-existent parked car? Might that cause a regulatory shutdown too?

Is regulatory shutdown inevitable?

Robo-Soldiers

At roughly the same time that the self-driving car fatalities were in the news, there was another announcement, even more closely related to my previous post. Video-game developer EA posted a video showing the results of a multi-disciplinary effort to train a AI player for their Battlefield 1 game (which, despite the name is actually the fifth version of the Battlefield series). The narrative for this demo is similar to that of Google’s (DeepMind) chess program. The training was created, as the marketing pitch says, “from scratch using only trial and error.” Without viewing it, it would seem to run counter to my previous conclusions, when I figured that the supposed generic, self-taught AI was perhaps considerably less than it appeared.

Under closer examination, however, even the minute-and-a-half of demo video does not quite measure up to the headline hype, the assertion that neural nets have learned to play Battlefield, essentially, on their own. The video explains that the training methods involves manually placing rewards throughout the map to try to direct the behavior of the agent-controlled soldiers.

The time frame for a project like this one would seem to preclude them being directly inspired by DeepMind’s published results for chess. Indeed, the EA Technical Director explains that it was earlier DeepMind work with Atari games that first motivated them to apply the technology to Battlefield. Whereas the chess example demonstrated ability to play chess at a world class level, the EA project demonstration merely shows that the AI agents grasp the basics of game play and not much more. The team’s near-term aspirations are limited; use of AI for quality testing is named as an expected benefit of this project. He does go so far as to speculate that a few years out, the technology might be able to compete with human players within certain parameters. Once again, a far cry from a self-learning intelligence poised to take over the world.

Even still, the video demonstration offers a disclaimer. “EA uses AI techniques for entertainment purposes only. The AI discussed in this presentation is designed for use within video games, and cannot operate in the real world.”

Sounds like they wanted to nip any AI overlord talk in the bud.

From what I’ve seen of the Battlefield information, it is results only. There is no discussion of the methods used to create training data sets and design the neural network. Also absent is any information on how much effort was put into constructing this system that can learn “on its own.” I have a strong sense that it was a massive undertaking, but no data to back that up. When that process becomes automated (or even part of the self-evolution of a deep neural network), so that one can quickly go from a data set to a trained network (quickly in developer time, as opposed to computing time), the promise of the “generic intelligence” could start to materialize.

So, no, I’m not made nervous that an artificial intelligence is learning how to fight small unit actions. On the other hand, I am surprised at how quickly techniques seem to be spreading. Pleasantly surprised, I should add.

While the DeepMind program isn’t open for inspection, some of the fundamental tools are publicly available. As of late 2015, the Google library TensorFlow is available in open source. As of February this year, Google is making available (still in beta, as far as I know) their Tensor Processing Unit (TPU) as a cloud service. Among the higher-profile uses of TensorFlow is the app DeepFake, which allows its users to swap faces in video. A demonstration compares the apps performance, using a standard desktop PC and about a half-an-hour’s training time to produce something comparable to Industrial Light and Magic’s spooky-looking Princess Leia reconstruction.

Meanwhile, Facebook also has a project inspired by DeepMind’s earlier Go neural network system. In a challenge to Google’s secrecy, the Facebook project has been made completely open source allowing for complete inspection and participation in its experiments. Facebook announced results, at the beginning of May, of a 14-0 record of their AI bot against top-ranked Go players.

Competition and massive-online participation is bound to move this technology forward very rapidly.

 

The future’s uncertain and the end is always near.

 

*To be sure, I learned a few of those lessons the hard way, but that’s a tale for another day.