Philosophy Forums
Forums Links Articles Gallery Chat
Style:



Register | Forgot Password

A New Kind of Science

printPrint


Page: 1 2 3 4

A New Kind of Science
DJPavel
Assistant Professor

Usergroup: Members
Joined: Oct 02, 2007
Total Topics: 8
Total Posts: 394
Posted 07/05/08 - 07:14 PM:
Subject: A New Kind of Science
quote post
#1
There is an interesting and provocative article on edge.org by Chris Anderson, called “The End of Theory”. Unfortunately, you cannot enter the debate unless you published several books and belong to their “Reality Club”. But I’m not complaining; with as much talent as I’ve seen on this board, I honestly think we can do better (and I don’t mean this in disrespect to those guys). I provide the links at the end of the post, but the major premise of the article is that as the amount of information available for analysis exponentially increases, it’s becoming practically impossible to make sense of the data in a traditional scientific way by building causal models and verifying predictions implied by the resulting theories. A few good examples in the article illustrate how objectives are achieved based on mathematical modeling only, without reliance on any causal theories. A lot of scientists and lay folks don’t want to call such methodology as “science”. They believe that statistical correlations might be necessary, but in no way sufficient conditions for science. Science, they claim, needs models. As Lee Smolin puts it (in an unrelated context), "Physics should be more than a set of formulas that predict what we will observe in an experiment; it should give a picture of what reality is. "

So, the question is, Can science retain its established (arguably) successful track record by substituting its narrative models of reality with reliable predictions based on statistical correlations only? Can we still call such methodology science?

Here’s my take on it. Let’s go back to basics and start from scratch. Why do we need knowledge? Well, because we have needs. The mere biological existence of our bodies necessitates solving problems. Knowledge allows us to meet these needs and we seem to have different ways of obtaining it. One of these ways we can call “science”.

Looking at it from such perspective, the notion of science is like the notion of sport – it is a social enterprise and its definition is a matter of convention. So, properly defining what science is and what it’s not is to meaningfully differentiate it from other domains of human inquiry. My criteria for such differentiation are those of utility and reliability.

Utility. The scientific method of inquiry must have value in addressing specific human needs. No scientist in the world needs to be preoccupied with studying ant diets, for example, for the lone purpose of extending ants’ longevity. Such research would only make sense if we could gain some insight from it to solve our own problems.

Reliability. Our cognitive capacity allows to solve problems in complex ways, but it becomes inefficient to reinvent the wheels every time we seek a solution. We thus rely on the existing knowledge. But in order to be successful in resolution of a complex task, the existing knowledge must be reliable. Otherwise, our bridges would collapse, the planes would fall, and we would never be able to launch a man in space. How do you achieve reliability? The proposed solutions must be testable and falsifiable.

These two criteria seem to me the bare minimum for calling something science. They are also sufficient conditions for preferring science over any other methods of inquiry (assuming you want to solve problems in the most efficient manner). Any other feature of science, such as accompanying narratives, requirement to meet only ‘good natured’ human needs, belief in transcendental universal laws, are value add-ons, a bonus, if you will, and not a requirement.

With such definition in mind, I agree with Anderson that we’re getting to the point of complexity that we can’t be building models and that’s still science. Companies like Google find innovative ways to solve problems simply by statistical correlations. Who cares that we don’t have the picture to visualize. The bottom line is it gets the job done, it works. When you take a prescription drug, do you require your doctor to give you a picture of how the medicine binds to the protein receptors in your cells and what chemical reaction that results in? Or do you simply take it because you know it works?

A few famous examples from the history that I find interesting and relevant that illustrate the utility without narration.

1. Cantor’s infinities. It doesn’t matter that you can’t picture one infinity “greater” than the other. You must accept such counterintuitive notion because it works. In fact, you can’t even deny it because in doing so you’ll contradict yourself.

2. Quantum Mechanics. Neils Bohr said that if you claim you understand QM, you don’t know anything about it. Fyemann added: “just do the math”. The experiments provide reliable predictions and there are upcoming computer technologies which will utilize them. Can we move on without “understanding the picture?”

3. Anthropic principle (in the weak form). This one deserves a separate thread, but given the fine tuning argument, you have to accept the multiverse, just like you have to accept the fact that the Earth is not designed for humans, it’s that humans filled an available niche as one of the possibilities produced by Earth’s environment. But accepting the multiverse based on such statistical inference without verification and testing flies in the face of traditional science. But there’s no other option. It’s either that, a very extreme coincidence, complete ignorance of the physical laws, or the Creator. None of the alternatives seems plausible. And considering things like “the coin tossing game” and biological speciation, the statistical inference in the Anthropic principle makes total sense. So, do you reject the multiverse because it’s just not traditional science, or do you accept such inference as a legitimate scientific method of investigation?


Here is the link to the article (you have to scroll down a little to the box):

http://www.edge.org/3rd_culture/anderson08/anders...


Here are the commentaries from the intellectual elite in the Reality Club:


http://www.edge.org/discourse/the_end_of_theory.html


DJP
Incision
Graduate
Avatar

Usergroup: Sponsors
Joined: Jan 04, 2008
Location: Utah
Total Topics: 9
Total Posts: 184
Posted 07/05/08 - 09:40 PM:
quote post
#2
I think there are trivial affirmative answers to both questions. Yes, doing what's been found to work will work. We'd still have something we could call "science," just because "science" is a vague word. But we should also consider whether we'd lose something without theories.

The theory of evolution isn't primarily a way to predict the future, it's mainly meant to tell us what really happened. Even if by using a horde of information we could predict new species more easily than we could by deduction from evolution, we wouldn't know where we came from without some kind of theory. Science may be more than just theorizing, but theories do something that statistics doesn't. Smolin's quote is right.
enkidu
Tenured Poster

Usergroup: Members
Joined: Jul 27, 2006
Total Topics: 18
Total Posts: 1254
Posted 07/05/08 - 09:44 PM:
quote post
#3
I don't think one should draw an opposition between "classical" science, and correlative methods, I would rather adopt the classic point of view that see them as complementary.

My first point is to say that what's going on with google is not so much a "paradigm shift" than a logical answer to an availability of data.
But these data's availability only concerns a few domains: For instance, to take an opposite example, in theoretical physics, we are facing a shortage of data rather than an inflation, which is why this domain is becoming increasingly mathematical, so much that some people have trouble calling it physics anymore, since we are left with a huge amount of hypothetical results without any means to verify them experimentally.
In other areas however, we have no problem to obtain data but no model is available, typically in Biology, because we are there considering such phenomena that take place at a scale that has been left uncharted by physics. It is a scale where we start encountering complex phenomena (a word on this shortly) that induces unpredictability, while it does not reach the simplicty of the newtonian macroscopic safe haven.
Complexity is really at the heart of this trend, and complexity typically is a phenomenological phenomenon, in other words, it is the domain of applied science much more than the one of fundamental science.

And that leads me to a second point that I see as capital to explain the sociological causes of this trend: We are living in a world where progress is dictated by economic necessity (I can argue more in more detail about this, Rene Thom already noticed it in the 60s). It's not a coincidence if Google is so put forward as the paradigmatic actor of this trend. It is immediate efficiency/utility which is here given the lead rather than genuine reliability, to take your distinction that is indeed valuable. Financing is working on this logic, especially private financing will favor quick rewards (and therefore efficiency) over long term uncertain investment (which is the requirement of fundamental science). The economical logic favors immediate utility and has no interest for theories and models (which are longer to develop and incur more cost),as long as it works, and it does, sometimes for very dubious reasons:
In finance (which is particularly concerned with this requirement for utility via correlation), there is something called the "self-fulfilling prophecy" which basically relies on the fact that if enough people follow a given arbitrary rule, this one becomes a useful rule for everybody to follow. This is particularly true in social sciences (which rely heavily on statistical tools) where psychology plays a fundamental role in the processes. But one can imagine a weak version of this phenomenon, especially in absence of an identified causal link, and we end up in thrall to a given correlation, so as this one become such an accepted evidence that all subsequent studies can only confirm it (because they implicitely take it for a causal link), we then fall into a dogmatic mode of thinking.

Let me then precise something, correlation is not something new, it has been there since the beginning of time, it has actually been the first tool for figuring out the surronding world. Clearly, today's correlation looks more sophisticated, a lot of data (in some limited cases), some fancy maths, that looks more impressive than the chance coindidence of a comet's observation and a bad harvest. The fact is however, that if you look a bit more closely to the maths the fanciness soon vanishes, and leave you with some valuable information but with a limited value, and it takes a clear understanding of this value to make the best of it; sometimes you get pretty close to causality, sometimes not so much, but you may still find some utility to the information you get, the trick is in knowing the difference, anyway, you'll always fall short of causality unless you put some effort into consolidating these data into a model, but that's where the cost becomes too high for the expected reward.

Edited by enkidu on 07/05/08 - 09:49 PM
ManiacJack
aka W
Avatar

Usergroup: Members
Joined: May 29, 2008
Location: San Mucus, TX
Total Topics: 37
Total Posts: 530
Posted 07/09/08 - 05:49 PM:
quote post
#4
Are we talking about nat. sciences?

Because the answer is no- it doesn't work. That forces the correlation.

We need a value system for correlation, not empty space. We have to define the consciousness for nat. science. Science done your way produces fallible theories at best; i.e. Global Warming.

Future Tense
Passed Relief

the Escapist wrote:
Bullshit, self-deception, self-aggrandizement.

Explains everything, really...
enkidu
Tenured Poster

Usergroup: Members
Joined: Jul 27, 2006
Total Topics: 18
Total Posts: 1254
Posted 07/09/08 - 06:31 PM:
quote post
#5
ManiacJack wrote:
We need a value system for correlation, not empty space. We have to define the consciousness for nat. science. Science done your way produces fallible theories at best; i.e. Global Warming.


Global warming is more a conclusion than a theory (in the scientific sense of the term), and it is not exclusively, not even primarily, based on correlations, but rather on well-developed mathematical models directly issued from physics. The uncertainty as to the results of these models does not come from statistics, but merely from the fact that these models are stochastic.
The general public however may get the impression that these conclusions are exclusively based on correlations, but it's not true, it simply is easier to convey these results in terms of correlations, the real science behind it being much too complex to be of any use to the public.

Edited by enkidu on 07/09/08 - 06:35 PM
Kelby
Cognitive Philosopher
Avatar

Usergroup: Members
Joined: Jan 03, 2008
Location: The Fire Nation
Total Topics: 10
Total Posts: 248
Posted 07/15/08 - 12:12 PM:
quote post
#6
I stumbled across the article at work and I’m glad to see this post. There is much involved with this issue and I have hardly the space and time, nor the intelligence, to touch on all the issues….mainly because I am ignorant of all the issues. But I do believe the reactions to such a phenomenon are fair and honest, and looking at it from such a distance may perhaps be the best position for observation.

The claim by Djpavel that science is “a social enterprise and its definition is a matter of convention” is passable, but I would beware of attributing conclusive finales to a word, leaving the word relative and “open” to complete interpretation. In the wake of such relativism, the word would lose all meaning. Of course, I am not saying that Djpavel meant it in this way. I am just sending a word of caution towards interpretation. Moving from this point, it is true that utility and reliability are inextricably tied, but there are underlying nuances that act as the glue, and these nuances are to be looked at more closely. Enkidu began to touch on these nuances when he said that “it is immediate efficiency/utility which is here given the lead rather than genuine reliability.” This is where the sifting begins. A simple question of reliability vs. truth vs. utility is in question.

If utility is truth, and utility is tied to reliability, then science, according to pavel’s “differentiations” would suffice and it would be evidently valuable. However, if what makes something reliable is the fact that it has utility, and truth is not dependent on utility, then science, according to pavel’s “differentiations” would be an endeavor based on a type of pragmatism. But it is not utility that makes something true; rather, something has utility because it is true.

Information is obviously abundant and it is not hard to see that such information is getting harder and harder to synthesize. However, I do not believe complete systemization is necessary for science to progress. It has been shown time and time again that empirical studies can continue to produce results by using only a small bit of information. I see no reason to believe that the new way of advancing could not be called science, especially if science is “a matter of convention.” Science, yes, as a means of gaining empirical knowledge, can continue, but philosophy on the other hand is a different issue.

Philosophy synthesizes and with the advent of exponentially growing information is the advent of relativism, as we experience it, for any sort of information can be synthesized to advocate any line of thinking. We are limited on how much knowledge we gain, and our minds synthesize the limited knowledge to fit coherently together. I guess one could say we are all coherence-theory followers, no matter how much we know the problems with coherence theory. Science can continue, and information WILL grow and continue to grow. However people will begin to synthesize their own choice of information to create philosophical truths that best explain their experiences, and these systemizations will be wholly unique. Meaning and truth will be completely relatively experienced.

As an ending note, take the fascinating field of neuroscience for example. It is becoming more and more clear that philosophy and neuroscience must be paired because simple information gathering has not helped us in understanding key issues in neuroscience, i.e., free will and consciousness. People like Daniel Dennett, John Searle, Bennett, Hacker, and Patricia Churchland all see the need for philosophical clarity and understanding in order to advance successfully in neuroscience. This is debatable, but I virtually agree with them. And Churchland points out that we can still have theories that do not encompass all knowledge. Unification does not need to be reliant on all known knowledge and unknown. Then again, theorizing is reliant on philosophy, and “theorizing about the brain and the behavior it produces will require both an understanding of fine grained facts about nervous systems and large-scale framework conceptions.” (Churchand)

Perhaps my post is unorganized and discursive, but I expressed my concerns. Simply, I feel science can progress without theorizing within certain domains coupled with strict interpretations of utility and reliability. But “domains” nonetheless. Not all domains of science can progress just as not all questions are best explained by science. Theorizing is still a necessity to progress in certain spheres of discussion.

Embodied Cognition: http://www.iep.utm.edu/e/embodcog.htm#H2
enkidu
Tenured Poster

Usergroup: Members
Joined: Jul 27, 2006
Total Topics: 18
Total Posts: 1254
Posted 07/16/08 - 07:25 PM:
quote post
#7
Kelby wrote:
...
Perhaps my post is unorganized and discursive, but I expressed my concerns. Simply, I feel science can progress without theorizing within certain domains coupled with strict interpretations of utility and reliability. But “domains” nonetheless. Not all domains of science can progress just as not all questions are best explained by science. Theorizing is still a necessity to progress in certain spheres of discussion.

Yes, I agree with you.
I singled out this last paragraph because I felt it highlighted a key perspective, that I may not have explicited in my earlier post.
Basically, the greatest risk of an over-reliance on data gathering and statistical treatment of these data, in order to draw correlations, and be somehow content with that, is the end of speculation.

Science is still suffering from the positivist myth which says that it is rooted in observation, and solely in observation, entertaining the idea of science as the mere, almost natural, expression of truth. This is obviously absurd to anybody who has considered this problematic a bit, but it may not be for most people. And in a market driven world, most people have their say, they are the one, indirectly who decide of what is important (worth financing) and what is not.

I believe this is the locus of the problem. Most scientists are very well aware that data collections and statistics do not make science, they know that speculation is the key driving force of science, that systematization, theoretization is a necessary requisite for the progress of knowledge, but the market ignores that, and it is mostly the market that allocates the funds, that finances the research. The market indeed only thinks in terms of utility, not in terms of truth (from its point of view, it's actually the same thing).

And the worse in this situation is that having the protagonists (philosophers, scientists) acknowledging the problem is simply not enough. The true challenge is in finding the means to solve the problem, and this solution can only be political, and endeavour to remove this problematic from the blind hand of the market.

Edited by enkidu on 07/16/08 - 07:30 PM
Kelby
Cognitive Philosopher
Avatar

Usergroup: Members
Joined: Jan 03, 2008
Location: The Fire Nation
Total Topics: 10
Total Posts: 248
Posted 07/16/08 - 11:57 PM:
quote post
#8
enkidu wrote:

And the worse in this situation is that having the protagonists (philosophers, scientists) acknowledging the problem is simply not enough. The true challenge is in finding the means to solve the problem, and this solution can only be political, and endeavour to remove this problematic from the blind hand of the market.


I agree with this on several fronts. William Barrett in his wonderful book on existentialism, Irrational Man, makes note that if the philosopher were keen he/she would realize philosophers have less and less influence on the world. The last great philosopher that had a wide influence in political and communal life was Dewey, and we are facing a type of bankruptcy in the ubiquitous influences of philosophical thought.
There was a time when philosophical quandaries were relatively widespread, i.e., France when Gutenberg created his magnificent invention. The philosophes were read widely, especially since the popular way to go write was centered around a vernacular emphasis. The lingua franca was drastically altered from Latin, the language of the scholars, to a more encompassing reach which in turn initiated Nationalistic facets.
Today, such widespread reading is not segregated to specific fields because the “elites” are no longer in charge of what type of information is worth publishing. Anything and everything is publishable. So I agree with Enkidu…that politics is our best bet to have the biggest impact, as it seems. We cannot tell people to think, question, and speculate in essays or philosophical books and expect a collective initiative. Politics gathers up and inspires in waves, and perhaps what needs to be done is propose political ideas that require such speculation.

A politic that inspires speculation by proxy would be a lethal striking force on Ignorance.

Embodied Cognition: http://www.iep.utm.edu/e/embodcog.htm#H2
Kwalish Kid
Tenured Poster
Avatar

Usergroup: Members
Joined: Sep 26, 2004
Total Topics: 22
Total Posts: 2992
Posted 07/17/08 - 05:32 AM:
quote post
#9
DJPavel wrote:
A few famous examples from the history that I find interesting and relevant that illustrate the utility without narration.

1. Cantor’s infinities. It doesn’t matter that you can’t picture one infinity “greater” than the other. You must accept such counterintuitive notion because it works. In fact, you can’t even deny it because in doing so you’ll contradict yourself.

I'm not sure that this has anything to do with science, let alone your take on its methodology. What is the application of Cantor's infinities that you have in mind?
2. Quantum Mechanics. Neils Bohr said that if you claim you understand QM, you don’t know anything about it. Fyemann added: “just do the math”. The experiments provide reliable predictions and there are upcoming computer technologies which will utilize them. Can we move on without “understanding the picture?”

Yet we cannot even "do the math" without some narrative. We have to describe the proper experimental set-up, we have defined entities in the theory that serve as the locus of determining what calculations to perform. We may not have a complete understanding of these entities, but we still use them.
3. Anthropic principle (in the weak form). This one deserves a separate thread, but given the fine tuning argument, you have to accept the multiverse, just like you have to accept the fact that the Earth is not designed for humans, it’s that humans filled an available niche as one of the possibilities produced by Earth’s environment. But accepting the multiverse based on such statistical inference without verification and testing flies in the face of traditional science. But there’s no other option. It’s either that, a very extreme coincidence, complete ignorance of the physical laws, or the Creator. None of the alternatives seems plausible. And considering things like “the coin tossing game” and biological speciation, the statistical inference in the Anthropic principle makes total sense. So, do you reject the multiverse because it’s just not traditional science, or do you accept such inference as a legitimate scientific method of investigation?

This might deserve its own thread, because your take on the anthropic principle, like the take of many starry-eyed physicists who want to go beyond the currently available observations, is simply wrong. There is nothing that the weak anthropic argument forces us to accept. It is only certain assumptions about what universes or areas of the universe are "typical" and what the probabilities are for the initial conditions of the universe or areas of the universe. Assumptions in these areas are, to my mind, completely unwarranted as of yet.

The weak anthropic principle says only that any theory that we propose should not have as one of its consequences that we do not exist. This is no stronger than, to steal a comment of philosopher of science Carrie Klatt, the requirement that no theory that we propose should not have as one of its consequences that jellyfish do not exist.

"Scientific truth is always paradox, if judged by everyday experience, which catches only the delusive nature of things." - KM, V, P and P

"A fishnet is made up of a lot more holes than strings, but you can't therefore argue that the net doesn't exist. Just ask the fish." - Jeffrey Kluger

"…Love of God and compassion and empathy leads you to a very glorious place, and science leads you to killing people." -Ben Stein [This is included for the irony.]
Cadrache
Tenured Poster

Usergroup: Members
Joined: Dec 09, 2006
Location: AB, Canada
Total Topics: 43
Total Posts: 1002
Posted 07/18/08 - 02:32 PM:
quote post
#10
*shrugs* We can logically and in reality *not exist*. If you take into account all the dimensions and the like, and base your theory of everything from a starting point where some dimensions do not interact with said original dimensions and these said 'hidden dimensions' are the ones in which humans inhabit, then humans do not exist.

Even with the application of Bell's localization theorums, at a certain amount of differences (change) in dimensions, you end up at a point where you can observe movements of some points within observable dimensions, but are unable to observe how they are moving or that they are even connected. (when applied mathematically, you end up at infinities as well as some areas that what-his-name did when acquiring the poincare conjecture; where things do not end up at a finite number.)
Download thread as

Page: 1 2 3 4



You don't have permission to post.

Please login or register.

30 total queries
This page was created in 1.65 seconds
Memory used: 7545148 bytes
Server Status: time since last reboot is 246 days, 4:59, load average: 2.34, 2.31, 2.22