Induction and Science

Two examples is nowhere near sufficient for induction to be useful.Yet there is no other way of identifying true regularities than by repeated observation. And it is from identifying regularities that science builds up it's picture of the workings of the natural universe.Induction, applied properly, works and is essential to science.

If there were 1000 examples - and if you could show that it was sufficiently unlikely to be due to chance - then it would certainly support the idea that there was something going on. If your result was replicated by other studies then we would havve more confidence, still, in it.And I have to ask why think that there are no true regularities (what are the "laws" of science but true regularities ?) and how science could work if it did not rely on them ? How could we, for instance, carry out spectrographic studies if we did not know the characteristic absorption and emission spectra of the elements ? If we could not trust even the simplest apparatus ? If every result was just chance, if every confirming replication was just a coincidence how could we do experimental science at all ?a

quote:

---

I'm dubious about generalization from just facts. You need an understanding of processes.

---

But don't we need to first establish the outcome of the process to obtain that understanding ? And isn't it true that in the case of gravity, for instance, that there are important aspects of the process that we still do not understand ? It seems to me that you are putting the cart before the horse here.a

If induction was no good then why would you ever think that a repeated observation was likely to indicate something worthy of further study ?As for spectrogtaphy we do have a theoretical understanding, now of how it works. But we didn't always have that and I would like to know how we could ever have worked out an understanding without accepting that the data was significant and indicated a genuine regularity.n

Yes, inductive reasoning stops at the point where we have made a generalisation. But if induction did not work then we wouldn't even get that far. All the other reaoning you mention is elaboration, not a denial of the validity of induction as a method of examinming empirical data.

I know what you are questioning. I am pointing out that you are also implicitly accepting it. If inducton were not a good way to identify patterns - regularities underlying phenonomena - then it would not be employed to find them.Your examples fail to understand that induction does not claim absolute reliability, nor that the reliability depends on taking an adequate sample, nor that induction is not to be taken as automatically overriding inferences made by other means, or even other inductive reasoning.

Experiments typically only give individual instances or at best only collections of individual instances. In other words by suggesting that experiment can validate a proposed natural law you are endorsing inductive reasoning.t

Let's try again. Generalising from collections of individual instances is induction. Experimental results are collectiosn of individual instances. The only way they can validate a supposed "law" is through induction.The trial-and-error aspect can track down which aspects of the conditions are important, but that, too fits within the inductive method (indeed the only way to conclude that a condition is important is through inductive reasoning).Your comments on a mechnaic adjusting timing are problematic for two reasons. Firstly we are talking about the validation of a proposed scientific "law" - which the mechanic is not doing. Secondly it is you, not I. that is suggesting that induction dows not work. Thus if the mechanic were using induction it is your views that would imply that there was a problem - not mine as you suggest..

I'm ASSUMING that a scientific law is a generalisation ? By definition a scientific law is a geneal statement. It is meant to apply generally, and not just in the specific instances tested.Newton's law of gravity is a general rule giving the atttactive force between two masses. Newton himself used celestial mechanics as his main observations. And we are still not certain of the underlying mechanisms.

Scientific laws are general statements derived from particular instances. That makes them generalisations.And since you are now saying that your comments on understanding the mechanics were not intended to deny that scientific laws were derived by induction it seems that you have run out of arguments.

Oh come off it.You're ignoring the argument for induction.You've presented strawman arguments against induction.The messages you present in your second point are yours and I agree that you do not deal with evidence and rely on bare assertion.FOr instance to deal in detail with your claims about Newton's law of gravity. Do you really think that Newton's law of Gravity is not a generalisation derived from specific instances ? Do you really think that it has been empirically derived by measuring the attractive force between every possible pairing of masses in this universe ? Or do you think that Newton derived it solely from first principles or more basic laws without direct reference to empirical evidence ?
You manage to write astronomy out of existence by insisting that the first experimental test of Newton's law was the first relevant observation. It;s rather an absurd idea - no scientific law would last for over a century unless it had a solid basis in empirical evidence - whcih must involve induction, since there is no other way to universalise a finite set of observations.Yiu say that scientiifc laws are derived from machanisms - but neither Newton nor Cavendish knew the meachanisms underlying gravity which are still not understood. And you have offered no way in which these mechanisms of yours could be discovered.The simple fact is that you don't know what you are talking about. Yet you insist that you must be right and won't hear the arguments against your view.

You claim that your examples are not strawmen. Alright. Can you show me an example of an actual use of induction, which is considered to be valid and useful which uses a mere two observations ?If not then your original example is a strawman.I also point out that youir own claims confirm that induction WAS heavily used in science at the point they were written:

quote:

---

Carnap, when studying degree of confirmation, concluded that for laws of physics the degree of confirmation is zero. Both Kuhn and Feyerabend made strong cases against traditional induction-based philosophy of science.

---

By the way you can't usefully invoke Hume or Goodman either since they don't refer to any difference in our positions. I don't and never have claimed that induction is a full equal of deduction. [quote] qsDo you really think that Newton's law of Gravity is not a generalisation derived from specific instances ?[/qs] Yes. [/quote]OK. I also note that you think it absurd that he worked it out by doing all the measurements required to provide direct empirical proof. or that he arrived at it through sound deductive arguments. But you don't say how he did work it out.IN Message 60 you claim that

quote:

---

As far as I can tell, the first specific observation, of which it could possibly be said to be a generalization, was the measurement made by Cavendish in 1798. That's 111 years after Newton proposed his law of gravity.

---

You now say that it is absurd to think that Cavendish's observation was the first and thus I should not mention that you said what you said.As to the rest it seems that you are the one arguing like a creationist. You don't offer any reasonable allternative to induction. You just insist that Newton must have used some other methood - while apparently not even being aware of Newton's use of astronomical data or indeed of any signfiicant data at all.Just like you insist that the problem reproted by Hume and Goodman must have been solved and that there must be a valid method of universalising a finite set of observations. Or at least one better thna inductionAs for your mechanic argument the real problem is that you have changed your argument. First you stated that scientific "laws" were based on known processes (which is true of some, but these rely on lower level "laws", which fails to help your position). Now you argue that we can universalise the results of a set of trial and error tests without using induction. OK, how do you do it ? Or is it just that you assume that there must be a way because your argument needs one ??

I asked you for an example of an inductive arguemnt considered to be valid based on only two observations. That psychologists claim that people often make inductions on two examples does not meet tht criterion. People often fall into fallacious thinking - the "gambler's fallacy" is one example. Thus even if your unreferenced claim is true you have not given any evidence that such examples are considered valid.Your example specifically relied on using far too few observations. To say that if you had used more it ould have still worked ay best begs the question.To continue, you ask how a quote which refers too "the traditional induction-based philosophy of science" can mean that science used induction heavily - when the context is scientific methodology. Well i is tellign you that the philosophy of science was heabily based on scientific use of nduction and since the philosphy shoucld reflect the practice the conclusion is obvious. You also might want to try explaining Carnap's statement before assuming that it has nothing to do with induction, too.I have no idea why you consider my comment on Goodman's "grue" problem to be inadequate. All it does is point out a potential problem with induction. Since I accept that induction is fallible then why should I be especially concerned with it ?Your comment on Kepler's laws also shows a problem with your view. If Kepler's laws are not based on empricial observations then they do not help your case. If they are then they are based on specific instances and they still do not help your case. Also, Newton also used other specific observations - Galileos' experiemnts with projectiles - as a basis for his work.Your attitude to "insults" is also completely unwrothy of someone in a moderator posiiton. You started this subtrhead by being insulting. But now you insist that iy is "unwarranted" for anyone ot make statements you interpret as "insultinh". Which includes pointing out the impliucatiosn of your arguents.Remember it is your assertion that science uses a method superior to induction. In that context you raised the arguments of Hume and Goodman - as arguments agaisnt induction. If your proposed methodology suffers from the same flaws then Hume and Goodman are irrelevant to the discussion. Thus you gave implicitly claimed to have a method which excapes those problems.The rest of your claims of "misrepresentation" similarly fail.If you actually intend to have a serious discussion then I suggest that you actually present your methodology and show how it removes the need for induction.

I guess you have forgotten the point of the request already. You were supposed to be producing a real example to show that your fictional example in the OP was not a strawman.

quote:

---

The number of observations was not the point

---

But it is very much the point. Induction relies on statistics, You need a good body of observations to make valid inductions. By choosing a hopelessly small number you set up a strawman. If you don't care about that then that speaks very poorly of you.

quote:

---

That might be a fair conclusion if the literature on philosophy of science actually had clear evidence of induction being used.

---

In other words you are now claiming that you do not understand what the quote you produced meant. [quote] qsIf you actually intend to have a serious discussion then I suggest that you actually present your methodology and show how it removes the need for induction.[/qs] It would be pointless. You have been consistently failing to respond to the parts of my posts that were relevant to such a discussion. [/quote]That's false. You've consistently failed to deal with the real problem all through this discussion. Your invocation of trial and error doesn't address it all - it's only a way of collecting observation. You've denied that you meant either deriving scientific laws from direct observation or by deduction. Use of earlier laws simply raises a bootstrapping problem.So far as I can tell the real reason why you are refusing to talk about your methodology is that you don't HAVE a sensible alternative to induction. And the low quality of the rest of your post only reinforces the idea that you are not even trying to hold a serious discussion at all.

quote:

---

Your harping on this is an unnecessary diversion. If my thought experiment was a strawman, then most of the literature on induction consists of strawman arguments.

---

Since it was your initial argument in the OP then it seems relevant. It seems more likely that the reason you want to stop talking about it is because I am right. The final assertion about arguments for induction is a clear non-sequitur, the more so since a strawman argument must be an arguet AGAINST something.

quote:

---

We were not discussing statistical sampling. But if you want to include that, then most ordinary measurements consist of a statistical sample of size one that is taken as valid.

---

Induction is at the least very closely related (the statistical methods of hypothesis testing are used to analyse experimental results). And when you say "most ordinary measurements consist of a statistical sample of size one" are you simply referring to ordinary measuremnts of a single object (which would be irrelevant) or are you asserting that we can validly derive a general statement about a population from a single measurement on a single individual ?

quote:

---

Stick to discussing the issues, instead of making accusations against the person.

---

Let me remind you that you chose to make accusations rather than discussing the issues in starting this subthread.

quote:

---

The fact is, that for many scientific laws, those laws are themselves
prerequisite to the possibility of making the observations that the law is alleged to inductively generalize. If that is what you mean by "bootstrapping problem", then it is enough to demonstrate that induction could not possibly be the correct explanation.

---

Your initial statement is trivially true in the sense that we cannot observe something that does not exist. It may even be true in the sense that the observer could not exist or would be unable to make observations in many cases. However neither of these would help your case. We are not talking about the existence of a "law", we are talking about how we can know that it applies.However that is not what I menat by a "bootstrapping problem". What I mean is that if you wish to derive a "law" of physics from one or more more basic "laws", it is necessary to know the more basic "laws" first. Thus such a method is forced into an infinite regress.And pointing out that there are no available alternatives is not intended ot be a formal proof that induction is necessary. After all such an argumnt could be easily falsified if it failed to enumerate a possibility. However it is up to you to actually show that there is such a possibility, the more so since you allege that scientists have been using it for centuries.

quote:

---

Stick to discussing the issues, instead of making allegations about the person

---

By which you mean that I shouldn't mention that you were making allegations against my person as an excuse to avoid discussing the issues. And presumably I shouldn't mention that you have just done it again. Why don't you leave off the "do as I say don't do as I do" and actually describe this method of yours ?