I’m wandering through my archives, looking for interesting stuff to follow up on. I ran across A Titanic Feedback Reversal. The Backwards Brain Bicycle is a fun implementation of the problem that sank the Titanic:
NPR has a nice article on self-regulation in the textbook industry. It turns out that textbook prices are up almost 100% from 2002, yet student spending on texts is nearly flat. (See the article for concise data.)
Here’s part of the structure that explains the data:
Starting with a price increase, students have a lot of options: they can manage textbooks more intensively (e.g., sharing, brown), they can simply choose to use fewer (substitution, blue), they can adopt alternatives that emerge after a delay (red), and they can extend the life of a given text by being quick to sell them back, or an agent can do that on their behalf by creating a rental fleet (green).
All of these options help students to hold spending to a desired level, but they have the unintended effect of triggering a variant of the utility death spiral. As unit sales (purchasing) fall, the unit cost of producing textbooks rises, due to the high fixed costs of developing and publishing the materials. That drives up prices, promping further reductions in purchasing – a vicious cycle.
This isn’t quite the whole story – there’s more to the supply side to think about. If publishers are facing a margin squeeze from rising costs, are they offering fewer titles, for example? I leave that as an exercise.
I have a soft spot for breast cancer research, but I have to admit that it seemed a little silly when I started getting hay with pink baling twine.
But now it seems the Susan G. Komen foundation for breast cancer has really jumped the shark, with pink drill bits from oilfield service company Baker Hughes. Funding cancer care with revenue derived in part from pumping carcinogens into the ground, providing pinkwash for that practice, seems like rather unsystemic thinking. What’s next, pink cigarettes?
Not so fast?
Maybe Baker Hughes is deriving some enlightenment from the relationship. In a less-noticed bit of news:
As part of our ongoing commitment, we have adopted a new policy with respect to the information that we provide about the chemistry contained within our hydraulic fracturing fluid systems. Beginning October 1, 2014, Baker Hughes will provide a complete, detailed, and public listing of all chemical constituents for all wells that the company fractures using its hydraulic fracturing fluid products.
Facebook is down.
Runaway positive feedback is the culprit:
To make matters worse, every time a client got an error attempting to query one of the databases it interpreted it as an invalid value, and deleted the corresponding cache key. This meant that even after the original problem had been fixed, the stream of queries continued. As long as the databases failed to service some of the requests, they were causing even more requests to themselves. We had entered a feedback loop that didn’t allow the databases to recover.
The way to stop the feedback cycle was quite painful – we had to stop all traffic to this database cluster, which meant turning off the site. Once the databases had recovered and the root cause had been fixed, we slowly allowed more people back onto the site.
This got the site back up and running today, and for now we’ve turned off the system that attempts to correct configuration values. We’re exploring new designs for this configuration system following design patterns of other systems at Facebook that deal more gracefully with feedback loops and transient spikes.
It’s faintly ironic, since positive feedback of a different sort is responsible for Facebook’s success.
The model is a mixed discrete/continuous simulation of an individual sleeping, working and drinking. This started out as a multi-agent model, but I realized along the way that sleeping, working and drinking is a fairly ergodic process on long time scales (at least with respect to UFOs), so one individual with a distribution of behaviors over time or simulations is as good as a population of agents.
The model replicates the data somewhat faithfully:
- Alcohol is the dominant factor in sightings.
- I don’t party nearly enough to see a UFO.
Actually, now that I’ve built this version, I think the interesting model would have a longer time horizon, to address the non-ergodic part: contagion of sightings across individuals.
h/t Andreas Größler.
A cool video tracing the evolution of feedback control from Maxwell’s seminal paper on steam engine governors to quadcopters that can perform amazing feats.
A simple example of bathtub dynamics:
The flow of plastic bags into landfills is dramatically down from the 2005 rate. But the accumulation is up. This should be no surprise, because the structure of this system is:
The accumulation of bags in the landfill can only go up, because it has no outflow (though in reality there’s presumably some very slow rate of degradation). The integration in the stock renders intuitive pattern matching (flow down->stock down) incorrect.
Placing the flow and the stock on the same vertical scale, is also a bit misleading, because they’re apples and oranges – the flow of disposal has units of tons/year, while the accumulation has units of tons.
Also, initializing the stock to its 2005 value is a bit weird. If you integrate the disposal flow from 1980 (interpolating as needed), the accumulation is much more dramatic: about 36 million tons, by my eyeball.
The Tech Review Arxiv blog has a neat summary of new research on high blood pressure. It turns out that the culprit may be a feedback mechanism that can’t adequately respond to stiffening of the arteries with age:
The human body has a well understood mechanism for monitoring blood pressure changes, consisting of sensors embedded in the major arterial walls that monitor changes in pressure and then trigger other changes in the body to increase or reduce the pressure as necessary, such as the regulation of the volume of fluid in the blood vessels. This is known as the baroreceptor reflex.
So an interesting question is why this system does not respond appropriately as the body ages. Why, for example, does this system not reduce the volume of fluid in the blood to decrease the pressure when it senses a high systolic pressure in an elderly person?
The theory that Pettersen and co have tested is that the sensors in the arterial walls do not directly measure pressure but instead measure strain, that is the deformation of the arterial walls.
As these walls stiffen due to the natural ageing process, the sensors become less able to monitors changes in pressure and therefore less able to compensate.
“It’s Time to Retire ‘Crap Circles’,” argues Gardiner Morse in the HBR. I wholeheartedly agree. He’s assembled a lovely collection of examples. Some violate causality amusingly:
“Through some trick of causality, termination leads to deployment.”
Morse ridicules one diagram that actually shows an important process,
The friendly-looking sunburst that follows, captured from the website of a solar energy advocacy group, shows how to create an unlimited market for your product. Here, as the supply of solar energy increases, so does the demand — in an apparently endless cycle. If these folks are right, we’re all in the wrong business.
This is not a particularly well-executed diagram, but the positive feedback process (reinforcing loop) of increasing demand driving economies of scale, lowering costs and further increasing demand, is real. Obviously there are other negative loops that restrain this one from delivering infinite solar, but not every diagram needs to show every loop in a system.
Unfortunately, Morse’s prescription, “We could all benefit from a little more linear thinking,” is nearly as alarming as the illness. The vacuous linear processes are right there next to the cycles in PowerPoint’s Smart Art:
Linear thinking isn’t a get-out-of-chartjunk-free card. It’s an invitation to event-driven unidirectional causal thinking, laundry lists, and George Richardson’s Dead Buffalo Syndrome. What we really need is more understanding of causality and feedback, and more operational thinking, so that people draw meaningful graphics, employing cycles where they appropriately describe causality.
h/t John Sterman for pointing this out.