Archive for the ‘Modelling’ tag
Measurement, Error, and Uncertainty
Norm Costa recently published on 3 Quarks Daily his thoughts on “PSYCHOLOGICAL SCIENCE: MEASUREMENT, UNCERTAINTY, AND DETERMINISM” (part 1 and part 2).
A lot of his thoughts are mostly relevant to psychology, but I will try and extract some thoughts that are relevant to an environmental modeller. Mostly the thoughts that are relevant are related to measurement, error and uncertainty.
I do have a hard time on commenting on Costa’s thoughts, but I think they are well worth to be read and thought about!
Measurement
- a measurement is a comparison to a standard;
- standards do not last;
- I recently came across a funny unit: 1 knot, which is the velocity of a vessel which travels one minute of geographic latitude in one hour
Photo by wfyurasko
Error
Costa does treat “error” not so much in a sense as in “measurement error”, but in terms of “what is research” and how do scientists deal with research and progress in research.
I like comparing scientific psychology with psychics because it dispels many false notions of science and makes room for psychology in the pantheon of science. For example:
- It dispels the notion that ‘real’ science is exact, objective, and dispassionate; Also, it blunts the objections of those who dismiss psychological science as inexact, subjective, and self-absorbed.
- It dispels the notion that there is such a thing as an exact science; Rather there are sciences that deal with relatively smaller errors of measurement (physics,) and others that deal with relatively larger errors of measurement (econometrics.) Psychology lies between the two in terms of the size of errors of measurement.
- Few scientists are objective and dispassionate in the absolute; Rather, science, by the way science is conducted, is self-correcting, in the long run, and keeps scientists on the straight and narrow.
- Scientists are not free of biases, preconceptions, misconceptions, and personal agendas; Rather, these can fuel the energy, motivation, and creativity of scientists.
- No scientific knowledge is absolute, unchanging, or final; Rather, all scientific knowledge is proximate and provisional, and only represents the best we can produce to this point in time. We can count on better data in the future superseding present-day knowledge.
Uncertainty
In his discussion on uncertainty, Costa talks quite a bit about Heisenberg’s uncertainty principle and how it relates to psychology. Here are two relevant passages:
For Heisenberg and all of quantum physics, UNCERTAINTY IS A PROPERTY OF NATURE. For scientific psychology, uncertainty is a function of the limitation of our measurement tools, not a property of nature.
Einstein’s view of the world was a deterministic one: Knowledge of an outcome is certain, provided all factors and antecedents are known. For quantum physics, outcomes are uncertain (probabilistic) because uncertainty is a property of nature, not a function of inadequate measuring. Einstein’s objection to this is captured in his famous statement, “God does not play dice.” In fact, God does play dice. God spends the whole damn day playing dice. Paradoxically, Einstein, the man who gave birth to quantum physics, could not accept Heisenberg’s ‘Uncertainty Principle’ and was forever bypassed by science and left to doter in the backwater of physics.
Design
This is the story of a designer, who worked at google. He explains how design at google is extremely data driven.
Yes, it’s true that a team at Google couldn’t decide between two blues, so they’re testing 41 shades between each blue to see which one performs better. I had a recent debate over whether a border should be 3, 4 or 5 pixels wide, and was asked to prove my case.
Picking up the google story, Scott Stevenson argues in this mini-pamphlet how important designers’ judgments in software design are.
How does this relate to planetwater.org or my work? I am obviously not mainly in software design. What I have done recently is I have been going through a lot of iterations of one problem with a master’s level student of mine. It has been a lot of fun at times and it has been frustrating at times. We could have tackled the problem by “just do it”. However, we took a lot of paths away from the main path, and we did learn a lot. Maybe it was not the most direct way, maybe we also created a lot of “failures”, maybe we could have been quicker. We did learn a lot, and our end-result is very good.
It remains, generally, that the boundary between design, statistics, environmental modelling, and even art is very interesting. It might be an art by itself. And dreaming is a big part of it.
Jacob Bear Short Course – Day 4
The course is over. Instead of blogging immediately about day 4, I spent the evening in Torino and hung out with some people from the course. At this point I have to point out how nice the city of Torino is, how nice and willing to help the people are. In the town, during the days of my visit, I was asked at least three independent times, if I needed help! On the final evening, we sat down on a bench on one of the plazas, and an elderly man started to talk to us in Italian, slowly and very well understandably. He ended up walking with us through the old city four over an hour and pointed out places of interest. It was just wonderful!
On the last day we covered heat transport and transport with fluids of variable density, especially sea water intrusion. From a historical point of view it’s interesting that because of sea water intrusion, density dependent models were the first “contamination” models to be developed. That is before dispersion was developed, and hence sea water intrusion was treated with sharp interfaces. We learned about the “Hele Shaw Model“, which Jacob Bear has used to model sea water intrusion before the use of computers was feasible. Bear developed during his M.Sc. thesis a horizontal Hele Shaw model. His first bookhas a full section on constructing Hele Shaw models. The idea seems from a former time, but such a model could have its uses for education!
In the afternoon, Dr. Rajandrea Sethi gave a presentation on how his group models colloid- and nano-particle transport under saturated conditions.
These were just amazing four days in Torino. It was such an interesting approach – to hear essentially a short but complete version of porous media theory in four days. Jacob Bear as teacher for this short course was amazing. Every word he uses has a meaning, everything he says builds up consecutively, and he stresses the important points. I will have many ideas to write about in the next little while for sure! 🙂
The Hydrological Snark
A colleague of mine pointed me to an extraordinary paper. It is funny, but there is much truth below the surface. It seems to be on hydrological modelling, but I think it is for anybody who deals with the analysis of complex systems.
The paper’s title is “The Hunting of the Hydrological Snark”, here is a link to the paper’s site at Wiley’s. The full reference is
V. Andréassian, N. Le Moine, T. Mathevet, J. Lerat, L. Berthet, and C. Perrin. The hunting of the hydrological snark. Hydrological Processes, 23:651–654, 2009.
The paper describes the steps required to hunt the hydrological “snark”, an “hypothetical, unknown and unseen monster”, which supposedly is a hydrological system, the hunt is a perfect model of the snark, and the hunting party the hydrological modellers. I’d highly recommend it for reading!
Environmental Modelling – Financial Modelling
I’ve spent a fair bit of my time as a grad students trying to model groundwater systems. There have been long discussions on modelling and its philosophy. Today was a PhD defence of such a grad student. Typical thoughts afterwards tend to be “… if we only had more data… ” or “… if we only knew the underlying processes better and could describe them… ” or “… if we only had more computing power available… ” .
I think a key trait of an environmental modeller should be his or her awareness of the quality of the data and the limitations regarding the quality of the predictions of the model. So when I read the headline “Climate Models Trump Financial Models” I was a bit worried. Sure, climate change is a pressing problem, and there is a lot of effort put into improving climate models. However, for me as an environmental engineer it seems that global financial markets should be better understood than global warming. And if it’s only because finance has been around for longer than environmental engineering! 🙂
I had never put much thought into how financial models might work, so this article brought up an interesting point for me, even though I’m not quite sure what I think about it:
“Climate models are very complex but you more or less understand the basic physics or chemistry,” said Derman. “[Finance papers] look like physics but a lot of the similarity is syntactic more than semantic.”
For example, stock options are priced with the Black-Scholes model, which says that stock price movement can be seen to move like the random movements of particles suspended in a liquid, i.e. Brownian motion. But stock price models differ from particle models because they describe the aggregate actions of people.
I guess quite a few processes are modelled like Brownian Motion. Why not stock option pricing. This just means though that we don’t know the underlying processes of stock option pricing better, so we could model the pricing in a more deterministic way. Then I thought a bit and it occurred to me that if you asked me, I probably couldn’t explain to you what is going on in the financial world these days. Why are we having this crisis? What are the underlying processes? Then I found this wicked little movie that explains these processes:
The Crisis of Credit Visualized from Jonathan Jarvis on Vimeo.
Based on this video, this seems to be a fairly straight forward process… 🙂 But I agree, there is a human factor involved, which admittedly doesn’t make things easier. I wonder what would happen, if you would throw an environmental modeller, a psychologist, and a finance person into a room for a while!
It seems like Weird was on a roll writing about financial models: They also point out, that the length of the finger of a trader is an indication for his degree of success. For me, that seems a little far-fetched! However, there’s again one interesting thought:
But leading up to the crisis — and underlying public acceptance of the mistakes and wrongdoing that produced it — was a widespread belief in the fundamental rationality of free markets and economic behaviors. That assumption may need to be revisited.