Question:
Let me try to sketch (in a peaceable manner), the way in which Bayesianism works and it’s relevance to the current topic.
Suppose that we have an urn with an unknown ratio of black and white marbles, assumed to be distributed randomly: we reach in and pull out one marble after another, and each is black. Say we do it 40 times. The hypothesis that the majority of the marbles are black has now increased its probability over what it started out to be. This is true even if there are, say, 10,000 marbles in the urn.
Now suppose someone were to look at the wardrobes of 40 Basenotes members, chosen randomly, and were to note that there is a tendency toward Serge Lutens ownership, far beyond the reference population. (This is the Bayesianism at work.) He says on this forum ‘Basenoters love Serge Lutens’. It would be a mistake for me to post a reply saying ‘Hey, I’m a Basenoter and I DON’T love Serge Lutens’ [nor do I, because where I live we have no chance to sample them!] It would be a mistake because when the person said ‘Basenoters love Serge Lutens’ that is not refuted by an exception, or even a raft of exceptions. (Linguists have a name for these common generalizations that resist refutation by some number of counterexamples, but I don’t know what that name is.) The claim meant nothing more than that there was a tendency to Lutensophilia in Basenotes, and there may be such a tendency even if quite a few prefer other scents. Now if Serge Lutens himself gets wind of this fact he may well decide to advertise on Basenotes, because he can market his fragrances to this niche. There need be no ‘biological essence’ to being a Basenoter that is implied or being traded upon by so marketing Serge Lutens fragrances.
HTH
--------------------------------------
Oh, and I'll just add that: suppose someone says that the above stereotypes Basenoters as Lutensophiles, then the right response is not to say `Hey, I'm a Basenoter and I don't like Serge Lutens', but rather, `Hey, you are missing all of these other tendencies by focussing on just that one. We're also mad for Lorenzo Villoresi!' (or choose something that is in a lot of wardrobes).
That's the thing about tendencies: a lot can coexist at once.

Answer:
I think this post is a whole lot of fun. Greyhueofdoubt said it inspired him to rejoin his love of statistics, or something close to that. (It was made to another thread.)
I tug it out and make it sail its own journey as a thread because I sort of think it has something wise in it, that touches on many of the ways we discuss scent and generalizations on the forum.
It also makes me think we've got to get Eluard some SL samples.
--Chris

Answer:
I definitely agree that there is definitely some rare wisdom in Eluard's thought provoking post. It sums up what I believe: Many 'absolutes' can coexist peaceably in an ever evolving forum.

I do hope that this wisdom will deter / temper the vitriolic personalised condemnation, of those that express an alternative / ill informed view. Respect, Tact and Consideration for all.

Answer:
I think that the only problem with this would be defining the 'reference population'.
Anyone?
Anyone with 2 or more perfumes?
Anyone with merely a passing interest in perfumes?
We might wind up defining a reference population by using ourselves as the baseline- that data would be largely useless.
"Electrical Engineers are 1,000 times more likely to understand maxwell's laws than random people on the street"
vs
"electrical engineers know as much about maxwell's laws as other electrical engineers"
Or I might be talking nonsense- I am falling asleep at the keyboard while I wait for my flight.
-ben

Answer:
Interesting thread and posts.
I have only come across Bayesian logic in a legal context. There have been attempts to persuade the courts (in the UK at least) to allow a Bayesian quantification of different elements of evidence in criminal trials. The attempts have been in cases involving evidence that necessarily produces probability ratios, (particularly DNA ratios). Juries and judges presumably find it difficult to balance hard-edged statistical data ('If the accused were innocent there would be a one-in-a-million chance of a DNA match') with other evidence (alibi, inconsistent eye witness report).
These attempts have all been rejected by the courts on the grounds of a) the complexity of assigning numerical weightings to all types of evidence, b) the undesirability of replacing trial by jury with trial by statistical evidence.
To move from the firm ground of reporting legal developments to the treacherous waters of my own interpretation - I think that whatever the benefits of Bayesian logic in arriving at more or less tentative generalisations about groups of people, it seems to be unsuited to deliver conclusions about individuals. Perhaps the offence that sometimes flares up when generalised discussion takes place is a result of this. [Sorry, I'm running out of steam in this argument and I need to get back to work - I haven't expressed myself well, but look forward to other contributions].
------------------------
Hell, work can wait 5 minutes!! Looking at the linguistics of Eluard's argument I think I would, with respect, differ from him in places. Eluard suggests that a response to the assertion that 'Basenoters love SL' indicating that you were a Basenoter who did not love SL would be a mistake.
Even if the OP had included all the research that Eluard describes in their posting, it would be an error on their behalf to arrive at the particular conclusion. If the OP wanted to assert that there is a tendency towards Lutensophilia amongst the Basenotes community, then that is what they ought to have said. Through the constraints of my legally trained mind, I suppose what I am saying is that the burden of proof is on the person posing the question. If they do not want to be misinterpreted, then they must communicate what it is precisely they are asking or asserting.

Answer:
Only one thing wrong with what you said...
...we can test them in Australia, at Mecca Cosmetica in Sydney, Melbourne and Brisbane. True, not the bell jars, not the entire range, but about 12-14 of them.
Ged

Answer:
Hi ben — have a safe flight!
The reference class would not include Basenoters, but might, for a Serge Lutens fragrance (since it’s unisex) be the fragrance buying public of Europe and the U.S. say. Since the tendency toward SL ownership is much, much higher on Basenotes than the reference class, SL might decide that it would make sense to cast a net here. If he could find a sizeable niche that has a greater tendency to SL ownership than Basenotes then he might go there instead, though I doubt if such exists.
You might say SL ownership at Basenotes is totally unsurprising given its nature, but that was just to make the general point. If it was discovered that cigar smoking was very much higher on Basenotes than the general population then Cohiba might just decide to cast a net here as well (though there are easier ways to reach cigar smokers, because there are specialized magazines).
Hi Michael
The problem that the U.K. has is unique in the world. There is an entrenched anti-Bayesian group in the Universities (centered on the London School of Economics) that are determinedly opposed. This group was led for a long time by Sir Karl Popper. Since the 90s there has been an attempt to roll back their influence, but it has not been well handled IMO, and there has certainly not been much time for it to take effect. Thus judges and legislators who are opposed can always find credible experts who will present the arguments against (the problem of fixing priors, which you allude to is the main one). This is completely unlike the situation in the U.S. where Popper had little to no influence and probabilism has held the day since the 1940s.
The deep contradiction of the U.K. system is that the concept of reasonable doubt is simply a qualitative version of the more quantitative Bayesian concept. The underlying idea is the same: the weight of evidence compels belief when it reaches a certain point. Quantitative measures are more precise, not less, in the total evidential mix.
On the other matter of the generalizations: the problem is that they often do not work in ordinary discourse as they do in logic classes. In logic classes a single exception refutes (“tests” or “proves” in the medieval sense) a generalization, but many generalizations in ordinary use do not behave that way. Thus suppose someone said “All you Basenoters really love Serge Lutens!” I could reply: “Yes we do, though, personally I don’t”. What I’ve said is strictly absurd by the test of the logic class, but you would understand it perfectly. Similarly, if you said “Crows are black”. The truth of this is robust against an albino crow or two. Many ordinary generalization express tendencies — and, yes, someone could be precise and say it that way, but we grasp what they mean in most situations even when they express it as a generalisation. (Or we should!) If someone French says: “The English are perfidious”, we don’t take them to mean every single English person, man, woman or child, and it would be folly to reply, “Well, my uncle Albert isn’t perfidious!”
Thanks Ged — I had no idea of this store you mention. I’ll track it down.

Answer:
From your example, I'd make the hypothesis that on the balance of probability, members of Basenotes who fill out their wardrobe details have a propensity to own more Serge Lutens scents than other scents typically owned by members of Basenotes who fill out their wardrobe details.
If one knew the amount of Serge Lutens scents sold as a percentage of all scent sales in the world, one could then compare those percentages and make a further hypothesis about the procurement habits of members of Basenotes who fill out their wardrobe details and those of the scent buying population at large.
Also, I am a member of Basenotes and I do not love any scent. I don't think it is possible to be in a state of unrequited love with a bottle of scent (well...... not for any significant length of time, anyway)
The biggest over-generalisations typically arise with the use of the verb "to be", such as in "I am a Basenoter" or "I am an engineer". While it is true that I do devote a considerable amount of time doing things that members of Basenotes do and doing things that those practicing engineering do, I do numerous other things as well. Similarly, I do many of the things typically associated with women who look after their home - but I don't say "I am a houseperson" - which would be the logical thing to say if what I do does indeed determine what I am.
Renato

Answer:
I sort of think this is sort of misplaced in the Male Fragrance Discussion. I am uncertain concerning the wisdom aspects of this but definitely appreciate the chance to learn a few things beyond olfactory opinions.

Answer:
I think this does belong here - some of it strays away from the central topic (i.e. my contributions ) but the overall theme is how we should approach debate on fragrance choices that involves generalisations. I'm a bit surprised that there have not been more responses as this is an issue that sometimes seems to turn threads into 'train wrecks'.
I entirely agree with Eluard's general thrust that we ought not to let generalisations derail our conversations about scents. I think we have a difference of emphasis on how best to approach this - but its very late and I ought to go to bed - so I'll take up why he is wrong and I am right tomorrow