somebody wrote:
>> This is a classic Bayesian vs frequentist dilemma. I can't use past
>> experience to predict the result of *different* actions in the future.
> 
> Unless there is a correlation. Are you claiming there's no commonality
> between past research and future research? If funding caused past medical
> research to succeed because of a causal chain of relations (ie more money,
> more brains, more experiments, more cures), you seem to be saying that such
> causal relations cannot be dependent on in the future. Why not? I personally
> don't think this very particular point in time is special.

	I repeat my concern: Correlation studies are fine when the experiment 
isn't changing. I can talk about humans and their behavior, saying that 
there is a correlation between how much time they spend in school and X 
behavior. I can expect that to work over a wide range of humans 
(assuming, say, gender, race and income level are unrelated). But that's 
because I'm talking about applying the *same* treatment to the *same* 
kind of problem. Humans more or less are the same and don't change much 
with time. Society does, but if what I'm correlating is relatively 
robust with societal changes, the correlation will stand.

	I can't imagine a correlative study with what you're suggesting, 
because the problem is continually changing. Money was pumped in, and 
some results were obtained. So I'd bet that if you pump money into 
another country's scientific endeavors who are uninformed of the 
discovery, then perhaps they'll also discover the same thing.

	I can't say that since money was pumped in before and we got great 
results, then if you pump in more you'll get even more achievements 
(particularly given that we don't even know if those achievements are 
realizable by the laws of this universe).

	Do you really think that as long as we pump enough money into it, 
Moore's law will hold forever?


-- 
Ground yourself, THEN hug your motherboard!


                     /\  /\               /\  /
                    /  \/  \ u e e n     /  \/  a w a z
                        >>>>>>mue### [at] nawazorg<<<<<<
                                    anl

Post a reply to this message