I recently got an email from newly graduated Math(s) major (mildly edited):
I am someone with a deep-seated desire to help the planet remain as habitable as possible in the face of the trials humanity is putting it through. I’d like to devote my career to this cause, but am young and haven’t chosen a definitive career path yet. My bachelors is in pure math and I am considering graduate study in either applied math or statistics. I’m curious what you would recommend to someone in my position. Between getting, say, a PhD in statistics vs. one in applied math, what positions me best for a career in the climate science community? What are its acute needs, where are the job opportunities, and how competitive is it?
My response was as follows (also slightly edited):
As you may know I too started out as a mathematician, and then moved to more climate related applications only in my post-doc(s).
I can’t possibly give you ‘the’ answer to your question – but I do suggest working from the top down. What do you see specifically as something where someone like you could have maximum impact? Then acquire the skills needed to make that happen. If that seems too hard to do now, spend time on the developing your basic toolkits – Bayesian approaches to statistics, forward modeling, some high level coding languages (R, python, matlab etc.), while reading widely about applications.
One of the things I appreciated most in finding my niche was being exposed to a very large number of topics – which while bewildering at the start, in the end allowed me to see the gaps where I could be most useful. At all times though, I pursued approaches and topics that were somewhat aesthetically pleasing to me, which is to say, I didn’t just take up problems just for the sake of it.
I’ve found that I get more satisifaction from focusing on making some progress related to big problems, rather than finding complete solutions to minor issues, but this probably differs from person to person.
But what do other people think? How should people prepare to work on important problems? Are there any general rules? What advice did people give you when you were starting out? Was it useful, or not? Any advice – from existing researchers, graduate students or interested public – will be welcome.
We don’t need more scientists to tell us what what they have already told us. No career choice anyone can make individually is going to help the planet’s survivability. The only collective political action (mass movements) that put pressure on the government and corporations to legislate radical changes has any hope of succeeding.
It’s fine to give someone career advice, but as peak oil facilitates more economic collapse, fewer and fewer people will have careers. This reality is already here, as colleges and universities cut enrollment. It behooves us to remember that it was only starting in the early 1960s that a university education became a widespread, middle-class phenomenon. It is as short-lived as the oil age.
My PhD is in the biomedical sciences and so I can’t offer specific advice to a maths major. But I think simply asking a leader in the field for advice is a very good first start. If the planet and specifically climate science is your interest then identifying the movers and shakers and seeking their input is valuable. But remember that a PhD can be more of a learning experience than a stepping stone to a defined career path. More than likely during your PhD and/or postdoc you will find your niche. What really matters is finding mentors to guide you. Don’t settle for second best, chase the best candidates until they accept you into their sphere.
Re : #43 and #49 Bayes. (Marcus, Gavin and David Benson)
How about David McKay’s
book on Bayesian inference and information theory?
That is only one of his interests;
here is another.
Re: previous.
It should have been Mackay.
> don’t need more scientists to tell us what what they have already told us
I refute it thus:
“We dodged a bullet” — Crutzen, in his Nobel Prize speech.
We need more scientists to tell us what’s unanticipated.
Geoff Wexler @53 — There are 6 reviews in
http://www.cambridge.org/gb/knowledge/isbn/item1159904/?site_locale=en_GB
These suggest to me that statistics is not the primary focus. However, decide for yourself.
Here
http://stats.stackexchange.com/questions/125/what-is-the-best-introductory-bayesian-statistics-textbook/24526#24526
many different texts with different focuses, from pragmatic to philosophical, have received brief reviews.
OK, my third comment, I hope it’s not too much.
The more I think about it, the more I feel this question may be ill-posed.
I believe that climate change will be one of the defining issues of our times. It will affect every human life on this planet. BILLIONS of people will be involved in action, one way or another, to try to keep our planet habitable (or not).
As others have noted, the world will need professionals of all stripes devoting their careers to keeping this planet habitable: lawyers, politicians, engineers, community organizers, and yes, scientists, to name a few. It would be hubris to assume that we, as scientists, are somehow the most important people in that effort.
Scientists do play role in this narrative, one I am proud to be a part of. Earth Science in general is multidisciplinary, climate change even more so. It sucks in technical people from all sorts of backgrounds with all sorts of skills. Whenever you decide you want to participate in Earth Science, THAT is the time to come over and start immersing yourself in the field. There is FAR more opportunity here than in, say, particle physics or pure math or applied math or statistics.
For anyone starting out: a good career choice requires an intersection between the kinds of things you want to achieve (in this case, saving our planet) and your skills and aptitudes. I have only cursory knowledge of both these issues in your case, and there are likely no easy answers — it took me over 15 years to find that intersection! But maybe you can do it faster, if you work hard at knowing yourself, and you’re willing to try different things, and you’re mature.
Bert Rubash:
I agree with Bert. My division is spending considerable effort to replace existing Matlab and IDL code with Python, largely due to the prohibitive cost of licensing for both commercial languages. Take the trouble to learn Python, and avoid getting hooked on proprietary languages. It’ll make you more employable if nothing else.
I’d go with David Benson on this. Jaynes’ book is full of the sorts of insights that come only from someone who has been thinking about a subject for over 50 years. It is interesting that Gavin refers to it as a “classic”, as its publication date was only in the past 10 years after Jaynes’ death. That was probably the only way it would ever be published, given that Jaynes was never satisfied with it and kept tweaking it right up to his death.
My suggestion is to study what you find most interesting while ensuring that you progress your mathematical/coding/physical science skills. I wouldn’t be too specific beyond that and certainly not tie yourself to a must having computer language. You never know where your career may go in the future. For example, my PhD was on Schumann resonances which is primarily an electromagnetic phenomenon in the Earth-ionosphere cavity due to lightning. A little later did I think about the relationship between global lightning activity and climate (though the relationship is not straightforward). Much later did I realise the potential connection between the work I was doing on stochastic trends and autoregressive process and climate statistics. With the work being done in the 80’s and early 90’s in coded in FORTRAN. The move to C++ and R later was straightforward. So I would say, don’t tie yourself down too early – study what interests you.
A junior member of my department keeps trying to tell me computer science needs no mathematical background. I can only suppose he’s never had to solve an interesting problem. I would advise your student not to shy away from the hard stuff especially as applied to real-world problems. If you have a strong base and good ability to formalize you can switch fields and handle the complexity of multidisciplinary research much better than if you go for a narrow purely theoretical specialization and follow the line of least resistance. Aside from working climate scientists, John von Neumann is a great role model.
I second (third?) comments on Matlab. Aside from python, where there’s work going on to replace Matlab functionality, R is a great tool. And there’s no harm in learning a range of programming langages for when you just can’t make the thing you’re trying to smack home look like a nail.
I am sure that most colleges & universities in the U.S. offer student access to Matlab free. If not, the student version is very affordable. Sometimes it’s worth paying to not reinvent the wheel!
Wish I knew more R, it’s on my to-do list. I have dabbled in Python and should probably know more.
I have found that learning NCL, the NCAR Command Language, very useful — allows you to mine to vast amounts of climate data as you wish. FORTRAN remains the ultimate fall-back but only if you must.
Good luck, stick with the science it’s in short supply these days.
Maybe it is just youthfull exuberance, but after four or more years of university education and you express a deep-seated desire to help the planet remain as habitable as possible indicates a career in something other than science would be advisable. Deep-seated desires to help suggests you might be more suited to social programs than hard science. In all seriousness, you to biased out of the gate to be in the hard sciences.
First I have to say that what sidd says about the life of a graduate student in the hard sciences is outdated (at least in the U.S.) I have been at this for a long time (in Math/CS/Engineering depts), and from my point of view, the life of a grad student has improved significantly in the last 20 years. It is still intense, but not so exploitative as before.
The main point I want to make is that if you want to study climate change, I think you have to study numerical methods for partial differential equations. While this may sound like an esoteric, math-geek, topic, it is actually at the heart of all long term climate prediction.
In my (admittedly somewhat limited) opinion, the state of the art in multi-scale numerical modeling of the climate is currently lagging behind what scientists in fields like combustion or fusion are regularly adopting. So I will predict an increase interest in high-order,multi-physics, multi-scale numerical methods applied to climate in the next 10 years.
Learn this stuff, and you will be sought after.
Susan Bentsch McGrayne’s “The Theory That Would Not Die” is easy-to-read, fresh and breezy introduction to the history of statistics, with particular emphasis on the Bayesian vs Frequentist “wars”.
Marcus,
A book: “The Signal And The Noise,” by Nate Silver. I found about this site by reading it.
…is a relatively new book that is, essentially, *all* about Bayesian Statistics applied to a wide variety of fields, from baseball, cards, and the economy, to weather and climate science. Well, well worth the read, as it’s constantly pointing out the strengths and weaknesses of Bayesian logic (spolier alert: weaknesses are few, compared to other models). Nate Silver has also, recently, yet again proved himself a serious authority on Bayesian statistics by predicting (among other things) the last U.S. Presidential Election and related races with fairly stunning precision; or, as he would say – accuracy. His blog, for a quick primer, recently published an AMA transcript where he answers a lot of questions about his process http://fivethirtyeight.blogs.nytimes.com
[Response: Unfortunately, there are some things that Nate gets seriously wrong about climate change/climate modeling, etc. See my piece at Huffington Post on this. –mike]
Good luck to you.
A little physics to go along with that math would go a long way in improving her/his understanding of what she/he is modeling.
I recommend you read Mike’s blog article (in response to post #68). If nothing else, this makes clear the difference between being smart and KNOWING a field.
Many responses here have mentioned tools: C++/Python/MATLAB/IDL/NCL/Fortran, Bayesian Statistics, PDE Solvers, etc. These are (some of) the technical tools that underlie all hard sciences these days (and many “soft sciences” as well). Sure, anyone wanting to be a scientist should have these skills. But more important than any of them is the CONTENT of the field. We are studying the Earth, not just solving PDEs or grinding data through statistical models.
First and foremost, a curiosity about the EARTH is required. When I do something, I ask myself “what does this tell us about the Earth?” If I’ve built a cool piece of software that uses sophisticated statistics but in the end it tells us nothing, then I’m wasting my time as a scientist.
As for Statistics… this tool shines through when you want to measure small things accurately — for example, the likely winner of a presidential election that’s close to 50-50 split. But climate change is different, less subtle. You don’t need fancy statistics to see that the glaciers and ice sheets are melting, or that CO2 is going up, or that measured temperatures are increasing. Statistics is useful, don’t get me wrong. But progress is made by scientists who USE statistics, not by statisticians who wonder whether they might apply their tools to climate change. That is true, even when the scientists use outdated methods of computer science, numerical analysis, statistics, etc. Better statistics will lead to better papers. But in the absence of better statistics, scientists will make do with inferior tools.
I disagree with Comment #65. There are certainly scientists who are motivate purely by curiosity. But there are plenty who want their curiosity to fit a larger social or human goal. Nothing wrong with that. Scientists are people, not robots, and it is only natural that we care about the subject of our study. That does not, in and of itself, introduce a “bias” into one’s work.
Anyway, you need to find a career path that fits your interests, skills and aptitudes. And from the sound of it, your skills and aptitudes seem to make you at least as well suited to be a scientist as any of the other careers mentioned.
[Response: Well put Bob. Thanks for stopping by :-) –mike]
I agree very much with with Bob F but would just like to return to..
Some others here, know much more about this topic than me , but in general for other subjects , I would usually recommend reading more than one book at once, especially if there is a chance to download one free. In the case of the book I suggested, you could do the opposite of what its author suggests. David MacKay suggests that Chapters 3 and 37 are not essential and could be skipped. I think that you could perhaps skip most of the rest of the book and browse or read just those two chapters and parts of chapter 2. They are not written as if they are restricted to information theorists. You don’t have to follow every line.
There is also a tiny “Further Reading” list at the end of chapter 37.
Bob F:
This needs to be emphasized! There are many smart AGW “skeptics”, but almost none who know the field. Those that take the trouble to know it, like R. Muller, tend to lose their skepticism.
Regarding statistics and climate, I recommend some of the talks from a workshop at NCAR, October 2007, see Strange scholarship …, pp.67-70. Among the good talks (and one very bad one), was Statistical Issues Involving (Climate) Computer Models by Jim Bergman, including p.17 “How can statisticians help?” but especially, p.19
‘How can statisticians become involved?
The Key: Becoming involved in a ‗team environment‘ with scientists.
Facilitating infrastructure:
• NCAR, where teams operate
• SAMSI (and NPCDS), where teams can be formed
• National labs (both LANL and LLNL have climate/stat teams)
• Large interdisciplinary grants available today
Barriers:
• Statistics cannot generally fund involvement of statisticians in other
disciplines which, in turn, rarely have much money for statistics.
• Shortage of statisticians
• The time needed for a statistician to get deeply involved with
another science and to also learn the statistics needed for it.
• Scientists often have a hard time judging what they can do
themselves and when they should seek statistical help.’
In particular, while math/statistics skills are always important, it is another expression of the idea that one has to know something about the science as well. It is all too easy to generate statistics and nice graphs … that don’t actually contribute any insight into anything real.
On computing: as I’ve often said for 35 years, as in this:
‘USE EXISTING TOOLS
Work at the highest level possible’
Anything one can do to *find* code and *use* it rather than write it is good … because then, if you write it, you may actually solving something useful. [UNIX shell programming was created, in part, because I kept finding people struggling to write C code … that mostly duplicated the effects of some combination of existing UNIX commands.] If I were still teaching CMPSC, I’d probably have a few programming assignments where the task was to be solved without writing much if any code, but by studying the various repositories for what already existed and using that, maybe with a little glue.
Of course, some people must write in FORTRAN or C++ or C, either for structural reasons or performance, but most people should be able to avoid needing to do that.
Bob F at ~70 says it well and says it right.
As a layperson, I’ve been holding back here. But it is very important that you write clearly and communicate by any means possible. Right now, the science is getting clearer by the day, and the obfuscation more forceful. We live in a marketing-based society and our model will not create the community of all we need to solve the many problems we face.
The letter seemed clear and to the point, which is part of what is needed. Being able to tell a story that grabs people’s attention would be a great additional asset.
I agree that a person with aptitude in math should consider pursuing engineering to “help the planet.” Many years ago one of my physics professors said I should go into photovoltaics if I wanted to stop global warming. (I should have listened to him.) Two other things that come to mind are writing software used to design energy efficient buildings, including cost analysis (which could have a very large immediate impact), or working on computational chemistry or computational engineering for clean technology applications.
Response @ 68 Mike: Thanks for the fine article, and for pointing it out here, and for taking the time to do it when there’s so much else to do.
My advice is: study logic. I am stunned by the lack of it on the part of self-important think-tank carpetbaggers, and others, who come feigning high pedigrees.
The most common fallacy I find could be called, “pounding around the nail,” or maybe just,”changing the subject,” though Aristotle had a different term.
In addition to all the specific tools commented on here, logic is an asset–as RC contributors habitually demonstrate.
Nate Silver has been out of logic so long, on cli-sci, it looks like sarcasm to him.
It’s a missed opportunity, as you wrote, because he “could have applied his considerable acumen and insight to shed light on this important topic.”
Silver’s success sorting out demographic statistics for political relevance has gone–um, fallen–to (the wrong part of) his head.
On cli-sci, he is fighting for the wrong industrial revolution: the former one.
This may be a charitable way to understand the virulence of the attack you have been–excuse the word–forced to document. It’s the last industrial revolution fighting to continue.
One ideal paradigm deserves another.
That’s why I appreciate John Schellnhuber’s view of least-cost analysis and the global optimum, at minute 43:10 here:
http://www.youtube.com/watch?v=etl0FBDghtA
My view is that we could use more and better statisticians in the field of climate change. There are many examples of the wrong statistical methods being used to interpret climate data normally by climate scientists whose primary knowledge is in physics / climatology
#77 Colin Aldridge: cite?
Mike @68: the lesson – whenever you are asked about your field by an outsider, ask them first what they know.
Thanks for posting the link; I never read anything by Nate other than when his competition is partisan pollsters. Thanks also for the reminder of you book: I should get a copy.
On statisticians in climate science or any other science for that matter:
The thing is that most scientists have little formal training in statistics. In my case, while doing particle physics research for my PhD, it was simply assumed that you would pick up what you needed. As a result, a lot of us go only as far as necessity and inclination take us. The downside of this is that often our statistical analysis would be cringeworthy to a professional statistician, even if we get it more or less correct. The upside is that when we do learn a technique, we tend to see how it applies, since we have an application readily at hand.
The obverse side of this coin is that some statisticians may apply an analysis that is patently unphysical–e.g. approximating global temperatures as a random walk.
Snow’s Two Cultures have continued to bifurcate multiple times since the 50s. What is needed is researchers sufficiently broad to bridge the gaps between them.
Water is the big issue of today and the future. So I recommend Hydrogeology and Water Resources – in which I have a Masters, and have been working for 25 yrs.
I did a BA in Mathematics, went on to do an MSc in Environmental Science and eventually after years working went back to college to do a PhD in Atmospheric Physics. The most useful degree was Environmental Science. I work now in the area of EnviroInformatics and build environmental management software (mainly geospatial). There is a demand of people with IT and mathematical skills who understand environmental problems (from scientific perspective).
Excellent,my first time reading this blog. I have no advanced education or degrees. I am Joe average.
Knowing all the math and physics you could possibly ever want is not going to do anything if you dont figure out how to change Americans mindsets on how they currently live.
99% of the people just want to gas up the suv and go to the mall to buy more electronics gadgets to distract themselves from there problems.
Climate change isnt on there radar at all!!!
Excellent, first time reading this blog. I am Joe average. I have no advanced degrees.
Knowing all the math and physics in the world isnt going to make a bit of differance until someone figures out how to change the current mindset of Americans
99% of people just want to gas up there suv and go to the mall to buy more electronic gadgets and crap.
Climate change isnt even close to being on there radar!!
This is a great thread and has helped me think about what I want to do. I’m a “retired” applied statistician and software developer. I have an MS plus 30 hours and am considering going back to school to get a PhD that would prepare me to support some kind of climate change effort – a purposely vague goal. My thoughts are to get a PhD in statistics with additional courses in “big data” techniques and also some numerical analysis. I spent 35 years working in industry and (as has been mentioned above) realize that many scientists and engineers do not get enough training in statistics to be able to apply it efficiently to their problems. I think I could help in that area. I would feel that going back to grad school and updating my skills would have been worth it if I could help some solar panel manufacturer increase yield by 10% or help develop an algorithm that would optimize a power grid or help a battery company properly design a set of experiments to evaluate a new battery design.
Re- Comment by Medina64 — 28 Jan 2013 @ 2:25 AM
Go for it! So far In this thread the best advice regarding training in climate science has been provided by the professionals and students in this area, so I recommend that you ask some statisticians who are doing what you think you would like to train for and ask them how best to get there.
Steve
I am not a climate scientist: I work as an epidemiologist on the health effects of climate (and other environmental issues too). But my backgroung is in fluid mechanics, the proof that your entire professional life is not defined by your diplomas :-)
From my daily work, I understood that studying the climate is important, but if we want to change things we also need to study the impacts, and to understand the adaptation needs and possibilities for many different topics.
To do so, we need more people able to work on an integrated fashion, willing to share and develop new methods on a multidisciplinary / interdisciplinary basis. So if you already master enough maths and stats, you may want to broaden your view and study environmental or social sciences…
The new graduate posed the issue as follows: “I am someone with a deep-seated desire to help the planet remain as habitable as possible in the face of the trials humanity is putting it through.” This problem has many dimensions (scientific/technical, economic, sociopolitical, as starters), and any credible solution has to address these many dimensions in parallel. This calls for a systems engineering/systems optimization/operations research type of approach, and would be well suited for a person with good mathematical and logical skills. Having good climate science results is a necessary, but not sufficient, condition for addressing the overall problem. The climate science is central to setting the requirements for any technical, economic, and sociopolitical solutions. But, most importantly, all the components of the overall system must mesh.
With the caveat that I have not been convinced that GW = AGW or that climate can be successfully modeled to make valid predictions, I do find myself largely in agreement with at least some of the ideas in most all of the above. I especially find #3, #5, #19, #22, #75, #80, #85, and #88 relevant. As many have suggested, I also would advise to consider a broad approach to hard science, engineering, and mathematics. This would enable you to adapt to whatever the future needs of society might entail as opposed to a narrow paradigm of the moment.
Concerning statistics: As noted by others, there are methodologies labeled “statistics” and then, with all due respect to Bayesians, there are other methodologies also so labeled. During my college years at an institution largely devoted to science and engineering, analogue computers were just beginning to be replaced with mainframe digital computers. During a summer internship, I also had the opportunity to work at a research facility that was engaged in silicon crystal growth. (The long — hopefully long — cylindrical silicon crystals would go elsewhere to be sliced and diced into chips.) My job entailed writing a Fortran program that used the output of the crystal growth furnace / kiln computer to do a statistical analysis of the data (temperatures, rate of rotation and vertical speed, etc.) that would then be plotted graphically and also used as re-inputs to the crystal growth kiln computer. The job entailed not only learning Fortran IV on the job but also learning statistics. My vocabulary thus increased with words such as “correlation,” “significant correlation,” “chi square”, etc. The advantage of the statistics we used was that the results of calculations could eventually be measured. That is, as a result of our work the crystal growth procedure either improved or else it didn’t. I can thus identify with the experience and goals of #85 Medina64. But I am skeptical about using statistical inference to predict future climates or even next week’s weather. Thus I would be leery of “big data techniques” if the goal is to infer a supposed truth certainty of the paradigm.
It has only been in recent years that I have stumbled across “Bayesian statistics” in my reading. As I understand matters, a great deal of skepticism is called for, at least in some uses. A problem stems from the introduction of subjective beliefs (including hypotheticals about hard data) as opposed to just hard empirical data and calculations that can be tested in the real world. This problem has been further compounded by the failure to always be transparent regards such subjectivity and underlying assumptions. Mario Bunge wrote an interesting article on this titled “In Praise of Intolerance to Charlatanism in Academia.” See Bunge’s article in THE FLIGHT FROM SCIENCE AND REASON edited by Gross, Levitt, and Lewis and published by the New York Academy of Sciences, 1996. Bunge wrote (p.103):
“When confronted with a random or seemingly random process, one attempts to build a probabilistic model that could be tested against empirical data; no randomness, no probability. Moreover, as Poincare pointed out long ago, talk of probability involves some knowledge; it is no substitute for ignorance. This is not how the Bayesians or personalists view the matter: when confronted with ignorance or uncertainty, they use probability — or rather their own version of it. This allows them to assign prior probabilities to facts and propositions in an arbitrary manner — which is a way of passing off mere intuition, hunch, or guess for scientific hypothesis. In other words, in the Bayesian perspective there is no question of objective randomness, randomization, random sample, statistical test, or even testability; it is all a game of belief rather than knowledge.”
Is my skepticism regards Bayesian inference (and the public’s skepticism regards AGW while accepting GW) misplaced? Also note, at Amazon, some of the seemingly quite expert reviews of Jaynes’ book. See also RealClimate’s informative FAQs regards modeling.
In any case, several of my more specific suggestions for any mathematician, scientist, or engineer who is just starting out, and whether or not you accept an extreme version of AGW, would be:
(1). a. Spend some time with GoogleMaps. Google the country as a whole, any region, or any locale. Then ask yourself this question: Does the country really need more “infrastructure”? It has often occurred to me that some disestablishmentarianism has long been called for.
b. Consider ‘the Traveling Salesman Problem’ especially ‘biological approaches to.’
(2). Consider the notion, in economics, of cost externalities, both positive and negative externalities. If our goal is to be as unbiased as possible, who ought to pay for or benefit from such currently unaccounted for externalities? Ask the same question regards what are termed incremental costs. Who ought to pay for school buses driving 5 miles out to the suburbs or for mail delivery 20 miles into the country? Etc.? Who ought to pay for the externalities of global trade? Etc.?
(3). Study some quantum physics and nanotechnology so that we might get on with the space elevator (and also origins of life issues). Contemplate the question, Would adding or deleting mass to the Moon-Earth system change the system’s orbit about the Sun or is the orbit totally dependent upon initial conditions?
(4). As scientists, engineers, and mathematicians, consider how many MBAs and Wall Street, real estate, and commodities “money changers” you all will be needing to manage you and what comparative values you would place on such managements.
All those of you who already have multimillion dollar or larger bank roles, come work with me ASAP.
Bradley Efron
A 250-year argument: Belief, behavior and the bootstrap
Bull. Amer. Math. Soc. 50:1 January 2013
is an engaging, after dinner styled, 18 page paper which places Bayesian and frequentist notions of statistics in contrast, overlap and perspective. He ends with “[t]he two philosophies, Bayesian and frequentist, are more orthogonal and antithetical. And of course, practicing statisticians are free to sue whichever methods seem better for the problem at hand — which is just what I do.
“Meanwhile we can all get ready to wish Bayes rule a very happy 250th birthday [this] January.”
> free to sue whichever methods seem better
[use]
“seem”?
Here’s the Efron article — link to the abstract; PDF of full text available there
DOI: http://dx.doi.org/10.1090/S0273-0979-2012-01374-5
PII: S 0273-0979(2012)01374-5
http://www.ams.org/journals/bull/2013-50-01/S0273-0979-2012-01374-5/home.html
Hank Roberts @91 — Brad Efron invented (discovered?) the bootstrap method. Surely one of the greatest statisticians ever. So what seems to him is probably right.
:-)
Ah, here’s something at my level, mostly:
http://lesswrong.com/lw/774/a_history_of_bayes_theorem/
I would also add the words of Joseph Campbell: “Follow your bliss”