OT: Article: The Peculiar Blindness of Experts (with regard to predicting the future)

[RU848789]

Fascinating article in the Atlantic. Decades of research on predictions (much of it by Philip Tetlock) have shown that credentialed authorities are "comically bad" at predicting the future, especially in their areas of expertise and this phenomenon holds for just about every field evaluated (economics, technology, politics, geopolitics, etc.). But the article goes on to show that reliable forecasting of the future is possible, but that it's usually made by "curious generalists" rather than deep subject matter experts. Worth a read.

https://www.theatlantic.com/magazin...hcIva5W9Ml7glICsUDrCUBcMEFJN4Kc3FIdhq2pwn8_2k

This article also makes a lecture I just attended make much more sense to me. The lecture was by John Warner, the "father" of Green Chemistry (he wrote the seminal text on green chemistry, i.e., waste-and-environmental-impact-sparing chemistry) and founder of the incredibly productive "invention house," Warner-Babcock. His comments were similar to what Tetlock found with experts and predictions.

Warner commented that when putting together teams to invent a new chemical product, which is "green," he makes sure he has some team members who aren't experts in the particular discipline (e.g., if they're working on a new biocatalysis product, he will include a physicist and an inorganic chemist on the team, along with, of course, some experts in biocatalysis), as he finds they often bring a more open, questioning perspective to discovery than world experts in a discipline. This has led them to more breakthrough discoveries than if they had followed the "usual" R&D mindset of having "experts" be the only technical team members, since they often don't "think outside the box" enough when trying to develop innovative products.

https://www.warnerbabcock.com/

 

[RutHut]

Interesting article. Thanks.

I'll make sure to post it on the next weather thread.:stuck_out_tongue_winking_eye:

 

[RU848789]

Interesting article. Thanks.

I'll make sure to post it on the next weather thread.:stuck_out_tongue_winking_eye:

l was wondering if someone would say that, lol...

 

[Caliknight]

Someone should have sent this to Al Gore 10 years ago and AOC yesterday.

 

[GoodOl'Rutgers]

Someone should have sent this to Al Gore 10 years ago and AOC yesterday.

One might argue that predicting dire results is about the same as predicting there is nothing to worry about.

 

[RUforester72]

So we're good for 6-8 wins this year?

 

[ashokan]

I recall the WSJ monkeys doing better at picking stocks (a couple years in a row) that market experts.

It long been clear academics don't do well in real world (McNamara in Vietnam). Currently many are lost between the ears on apocalyptic weather change.

 

[retired711]

I recall the WSJ monkeys doing better at picking stocks (a couple years in a row) that market experts.

It long been clear academics don't do well in real world (McNamara in Vietnam). Currently many are lost between the ears on apocalyptic weather change.

McNamara was not an academic. He was was President of Ford Motors at the time of his appointment as Secretary of Defense.

 

[ashokan]

McNamara was not an academic. He was was President of Ford Motors
.

He once taught at Harvard Business School (he was highest paid and youngest assistant professor at Harvard). He was a numbers geek - always indexing and crunching. At Ford, McNamara and others were known as "whiz kids." In government service, the statistical models didn't help. It turned out McNamara didn't know much of what he was doing and later admitted it.

The NYT obit said of McNamara "he tends to be caricatured as smart but not wise, obsessed with narrow quantitative measures but lacking in human understanding."

Back then, people were very worshipful of credentials. I don't think its quite than same now.


https://hbr.org/2010/12/robert-s-mcnamara-and-the-evolution-of-modern-management

 

[retired711]

He once taught at Harvard Business School (he was highest paid and youngest assistant professor at Harvard). He was a numbers geek - always indexing and crunching. At Ford, McNamara and others were known as "whiz kids." In government service, the statistical models didn't help. It turned out McNamara didn't know much of what he was doing and later admitted it.

The NYT obit said of McNamara "he tends to be caricatured as smart but not wise, obsessed with narrow quantitative measures but lacking in human understanding."

Back then, people were very worshipful of credentials. I don't think its quite than same now.


https://hbr.org/2010/12/robert-s-mcnamara-and-the-evolution-of-modern-management

He was indeed an assistant prof at Harvard Business School but that was in 1940-1943 -- way before he became Secretary of Defense in 1961. In between, he was in the Army Air Force and then at Ford . It's really stretching to call him an academic. I agree with the rest of your critique of McNamara -- just mentioning his name raises my blood pressure -- but his problem was not that he had been a professor.

 

[ashokan]

He was indeed an assistant prof at Harvard Business School but that was in 1940-1943 -- way before he became Secretary of Defense in 1961. In between, he was in the Army Air Force and then at Ford . It's really stretching to call him an academic. I agree with the rest of your critique of McNamara -- just mentioning his name raises my blood pressure -- but his problem was not that he had been a professor.

If he was youngest and highest paid AP at Harvard the title fits. Plus that's where he got his template. A potential problem with learning is that you can learn more and more about less and less - and yet thinking you know more and more about more and more.

When I was in grad school, I was minoring in international relations, and I had to work on a project where statistical models were being proposed as predictors of future wars. I could see the idea was pretty zany, but a lot of people were taking it seriously. The rise of stats, algorithms, models - and then computers - bedazzled a lot of people who thought they were finding the keys to the kingdom.

That's all still a big problem actually, Business people think the computer people have everything nailed down. The computer people think Joe MBA from Wharton has the business knowledge nailed down. Then you get things like the financial crisis of 2008 where the computers failed to raise red flags because they were measuring the wrong things. Many people lost track of where the debt was. Dispersal of responsibility and overestimation of abilities

Dunning–Kruger effect
https://en.wikipedia.org/wiki/Dunning–Kruger_effect

 

[RU848789]

I recall the WSJ monkeys doing better at picking stocks (a couple years in a row) that market experts.

It long been clear academics don't do well in real world (McNamara in Vietnam). Currently many are lost between the ears on apocalyptic weather change.

Agree and disagree. The article mostly talks about Tetlock's research into political and economic predictions, not scientific ones, which are usually much more accurate. Even the Ehrlich predictions were largely correct on the population growth, but he was way off on the impact associated with that growth, since so many technologies related to food production improved dramatically, avoiding the famines Ehrlich thought would happen. With regard to global warming, my guess is we might see some similarities, in that the science is pretty strong on the warming part of the equation (if there are no technology "fixes" like gas sequestration technologies), but it's quite possible the impacts will be muted (maybe sea level rise will be less than predicted) or that humans will adapt reasonably well. Any "expert" has to concede that there are "unknown unknowns" out there that can wreak havoc with any prediction.

Anyway, I also wanted to differentiate weather from climate, since many appear to not fully understand the difference. Trying to predict both markets and weather beyond certain time points in the future are essentially useless endeavors, because these systems are chaotic in nature, meaning the numerical models used for future predictions will become less and less accurate with each time increment in the future, due to the fact that initial conditions can't be known fully accurately and those errors propagate out in time until the forecasts have such high uncertainty that they become worthless. Most experts in these fields know better than to try to predict beyond certain timepoints, but there are always exceptions, like some of the weather services that now try to predict specific weather beyond 10 days, which is mostly useless, and some even now publish 30 or 50 day weather forecasts, which are completely useless.

However, climate change "modeling" is quite different from weather forecasting/modeling, with the biggest difference being that climate forecasting is not nearly as affected by chaos as weather is, since the numerical models operate over very long time periods and while the initial conditions do matter, the key is that the forcing mechanism for global warming (overwhelmingly related to greenhouse gases like carbon dioxide and methane - and there's zero scientific disagreement that these gases trap infra-red radiation, keeping heat within the atmosphere - we wouldn't have life on earth if we didn't have the greenhouse effect), which is quite large, dwarfs the impact of the variabilities/perturbations from other initial conditions.

https://cleantechnica.com/2018/08/23/chaos-theory-does-not-invalidate-or-explain-global-warming/

Where there is uncertainty is over how much the planet will keep warming over the next 50-100 years, given projected increases in greenhouse gases, not whether the planet will keep warming. This is due to uncertainties in GHG levels (as technology might just save us somewhere down the road, making fossil fuel burning far less necessary; for example, at Warner's lecture the other day, he said his group has made major improvements to thin film solar cells for electronic devices, which he thinks will mean batteries won't even be needed, as these cells will work in ambient light), as well as uncertainties from other forcing mechanisms, like sun cycles/activity.

And, obviously, if there is uncertainty in the amount of warming, there will be uncertainty in the impacts from that warming, as well as difficult policy questions with regard to what to do to prevent potentially catastrophic impacts in the face of uncertainty of outcome. It was an easy call for the ozone hole, where the risk was extremely well understood and short term in nature (a few decades of continuing on the path we were on would've been catastrophic), plus the fix was a fairly simple technological fix, which made it far easier for governments and industry to work together, via the Montreal Protocol, to phase out the use of CFCs and HCFCs as refrigerants. The uncertainties are much greater, the impacts more uncertain, and the solutions much less agreed upon, with regard to global warming.

However, as per the most recent paper on this, coauthored by Robert Kopp, from RU (and others).a 2C increase in the next 80 years (the prediction if the Paris accords are followed, i.e., the likely best case, asssuming no huge surprises or huge tech breakthroughs) would be moderately to seriously impactful, with a 1-3 foot rise in sea level while a 5C increase (if greenhouse gas emissions keep increasing, unchecked) would likely be catastrophic for many with up to a 6 foot rise in sea levels being possible. Since we know outcomes like these are at least plausible, if not likely, it does seem like "doing nothing" which is the Trump approach, is foolhardy, especially when many of the things that could be done (improving efficiency, reducing usage, funding a "Manhattan Project" for alternative energy research, etc.) would likely be good for both the country and the world's environment.

https://www.pnas.org/content/early/2019/05/14/1817205116

 

[GoodOl'Rutgers]

McNamara was not an academic. He was was President of Ford Motors at the time of his appointment as Secretary of Defense.

And one of the lessons of WWII was that logistics played a pivotal role. There was a legit argument that organizational skills could be important. But we're always fighting the last war.. right?

RU#s.. has there ever been an accurate climate model yet? Whatever Al Gore was using was all wrong.. how do we know these models have gotten any better?

 

[RutHut]

So we're good for 6-8 wins this year?

Well, rutgersal is an expert on RU football, so take that for what it's worth.

 

[mdk01]

I recall the WSJ monkeys doing better at picking stocks (a couple years in a row) that market experts.

I believe it was darts.

 

[RU848789]

And one of the lessons of WWII was that logistics played a pivotal role. There was a legit argument that organizational skills could be important. But we're always fighting the last war.. right?

RU#s.. has there ever been an accurate climate model yet? Whatever Al Gore was using was all wrong.. how do we know these models have gotten any better?

View embedded media

GOR - that's such a loaded question, without having done any research, as it implies all climate models are wrong, which is way off base. Yes, most of the climate models have been reasonably accurate in "hindcast" predictions ("predicting" what we already know happened in the past, which is a good way to check a model for biases) and in "forecast" predictions (predicting the future climate). The attached summary of the performance of many of the major models used over the last few decades (including the IPCC models, which are the "official" ones for the world climate science community) is quite informative. Below is the conclusion from the article analyzing model performance.

None of this guarantees that future temperature increases from the models (and related outcomes like sea level rise) will be correct, but I would hope it makes people aware that major impacts are at least moderately likely in the next 50-100 years. As I said above, I'd rather focus on the most likely model outcomes, not the doomsday scenarios and actually have political conversations about what to do about the most likely impacts and how to mitigate them/prevent them.

"Climate models published since 1973 have generally been quite skillful in projecting future warming. While some were too low and some too high, they all show outcomes reasonably close to what has actually occurred, especially when discrepancies between predicted and actual CO2 concentrations and other climate forcings are taken into account.

Models are far from perfect and will continue to be improved over time. They also show a fairly large range of future warming that cannot easily be narrowed using just the changes in climate that we have observed.

Nevertheless, the close match between projected and observed warming since 1970 suggests that estimates of future warming may prove similarly accurate."

https://www.carbonbrief.org/analysis-how-well-have-climate-models-projected-global-warming

The problem with Al Gore is Al Gore. He simply went off the reservation to try to play up the worst case scenarios that might happen and, by doing so, IMO, he set climate science and climate discussions back, somewhat, although he did also help raise awareness of the potential risks of climate change. Every model has confidence intervals for its outputs and Gore often quoted the worst case. On the other hand, that video you posted is also inaccurate, as it doesn't contain the full quotes from Gore or any analysis of what he actually said.

For example, Gore did predict that the Arctic sea ice could be gone by 2014, during the summer, as summer sea ice levels are the lowest - he never said the Arctic would be completely free from sea ice year round by the summer of 2014, as that video implies. Regardless, he was obviously wrong, although summer sea ice levels are way down (see below) and it's thought there will be ice free summers within 30-50 years and April 2019 had the lowest recorded Arctic sea ice levels in April in modern history.

Gore also said that sea levels could rise by 20 feet if the Antarctic ice sheets melted, but he didn't "predict" that that was going to happen, per se, as the video implied. And that estimate is reasonably correct, although the likelihood of that happening in the near future is pretty low.

https://www.snopes.com/fact-check/ice-caps-melt-gore-2014/

 

[Caliknight]

One might argue that predicting dire results is about the same as predicting there is nothing to worry about.

Maybe. But in this case, those who predicted the latter are right.

 

[Barnaby&Neill]

Someone should have sent this to Al Gore 10 years ago and AOC yesterday.

I’m a curious generalist (refer to article summary if you’re wondering why I mention this). I predicted emitting egregious amounts of greenhouse gases into the atmosphere would cause the climate to get warmer, lo and behold it has.