When to "stop" doing science?

[rwp]

Yes, I agree. Please see post #122, and #124 where I agree that I messed up and left out an important part. It is a big/small mistake but an honest one with no intention to deceive. I meant to make a point with it.

I can't edit my post where I say the very definition of... but I hope that this long-winded discussion has shown that I am indeed sorry and have been punished for the crime.

The punishment didn't fit the crime, but that's just the witch's (my) opinion. I much preferred Wollery's gentle correction to being burned at the stake.

 

[T'ai Chi]

We do not have an a priori answer for this one, so we cannot have an arbitrary "truth" for it to converge on.

Just because you don't have an a priori you don't believe it can converge?

 

[TjW]

How will you know when it's converging? How will you know what it's converging to? How did belief get in here?

 

[T'ai Chi]

A graph of the results tending to some horizontal line, and deviations from horizontal getting smaller and smaller over time, is a good hint of convergence.

We don't "know" it is, that is the point. When are we satisfied that it would be. For example, a coin being 50-50.

 

[CFLarsen]

A graph of the results tending to some horizontal line, and deviations from horizontal getting smaller and smaller over time, is a good hint of convergence.

We don't "know" it is, that is the point. When are we satisfied that it would be. For example, a coin being 50-50.

But without a scale, a graph such as yours is worthless. You have an education in statistics, right? You have an IQ of around 160, right? Unless you made all that up (and there are indications that you did), how can you possibly not understand this?

How do you measure "truth"?

 

[Mercutio]

Just because you don't have an a priori you don't believe it can converge?

You misrepresent. It cannot converge on an a priori because there is none. It will (probably) converge on a number, but each flip merely narrows the range of possibilities. How many decimal places do we go before "THE TRUTH" is found?

We quite literally cannot know. With each flip, as our sampling distribution narrows, we are still left with an infinite number of possible "truths", albeit in an infinitessimal range. (And if we say our sampling distribution approaches "normal", as the central limit theorem says, we must understand that there is a finite probability of our parameter "truth" being outside--perhaps well outside--of our confidence interval.)

Fortunately, it is not the job of science to find absolute truths.

Doing stats without an understanding of probability is a bit like doing biology without an understanding of natural selection: One can go through a lot of the motions perfectly adequately, but gets tripped up on details that should be so simple.

 

[joobz]

I don't know Loki's wager. Is it like Pascal's?.

I had to look up Pascal's wager to answer. no, it's different.

Loki's wager(normally used as a logical fallicy) is based upon the story of Loki betting his Head. When he loses the bet and someone comes to collect he says, "You have all rights to take my head but none of my neck." which starts the debate about when does the head end and neck begin and what parts are neck and head. As a result, he keeps his head because the debate goes on.

I would like to get off of the probability theory debate.
Based on the initial Figure, It's Tai Chi's conjecture that we approach "truth" in a 2nd order underdamped response like fashion.(process control theory) In such a system, you never truly hit the target level because you approach the target level asymptotically. However, in reality we specify a thresold tolerance (say+_5%) and when our value hits that tollerance we claim victory.

If we accept this model, then he's right. Science has some end point. However, inherent in this view is that newer discoveries are less significant then past ones (look at magnitude changes as we approach "truthiness"). In fact, this does not hold true at all.

Consider the central dogma of genetics (DNA theory)
DNA encodes RNA that uses t-RNA to create proteins. This held true and describes much, yet the discovery of siRNA(which can block the creation of proteins) describes a whole new roll of RNA that wasn't previously appreciated. This isn't less signifigant. In fact the implications that RNA might do other things is huge.

Another example is RNA structures. For a long time, we only considered the tertiary structures of proteisn, yet RNA can also adopt interesting foldings and these folding effects may serve in ways completely unknown before.

Further, consider the transition from newtonian to Quantum mechanics. Consider the discovery of quarks.

We aren't approaching some setpoint of truth. We approach preconceived setpoints. But as we discover more, that setpoint changes. continually.

So, when do we stop science?
when no more questions are left to ask.
When will that happen?
?? I assume we'll know when we get there.
but until then, we keeping asking and then testing.
and asking and testing
and asking and testing
and asking and testing.

 

[T'ai Chi]

So one agrees it converges. ie, our measurments get closer and closer and closer to something, some parameter.

So how does one know when to stop and say 'it converges'?

 

[Grimoire]

So one agrees it converges. ie, our measurments get closer and closer and closer to something, some parameter.

So how does one know when to stop and say 'it converges'?

Even if it does converge, we do not stop. Newton's Laws converged. Everyone agreed, for 200 hundred years that Newton's Laws fully described the physical laws that govern the motion of objects. But then someone noticed a problem with the way that planets moved. They were just a little off from what Newton would predicate.

So now, despite a convergence, we suddenly get a divergence. That's the problem with your graph, nitpicking about "truth" and units aside, is that it is not an actual representation of what happens. As new observations are made, we must constantly re-evaluate the results, and make sure that they are consistent.

So, in the case of Newton's Laws, yes, we thought we had convergence. But suddenly, we get new observations that don't fit and it diverges. We need a new principle to explain the divergence.

Just because all the previous observations are in agreement with our theory, it does not mean the next will be.

 

[joobz]

So one agrees it converges. ie, our measurments get closer and closer and closer to something, some parameter.

So how does one know when to stop and say 'it converges'?

you missed the point again.
Conceptually. Yeah we hope to be approaching the truth.

but we've given you the answer. The simplist answer we have. We stop "science" when we stop having questions to ask that we can then test.

 

[joobz]

So now, despite a convergence, we suddenly get a divergence. That's the problem with your graph, nitpicking about "truth" and units aside, is that it is not an actual representation of what happens. As new observations are made, we must constantly re-evaluate the results, and make sure that they are consistent.

So, in the case of Newton's Laws, yes, we thought we had convergence. But suddenly, we get new observations that don't fit and it diverges. We need a new principle to explain the divergence.

Just because all the previous observations are in agreement with our theory, it does not mean the next will be.

Nicely stated.
Thank you

 

[Hellbound]

IN a nutshell:

We "stop" when all the results of experiment and observation match the theories within the range of error allowed by current technologies and situations.

When technology or the situation changes to make those error bars smaller, and they no longer agree with the theory, then it starts up again.

Technology changes that caused re-starts: Better telescopes and record-keeping to notice the discrepency in orbits, leading to Relativity theory as an expansion of Newtonian mechanics.

Situational discrepencies: Problems reconciling relativity and the standard model at certain extremes, leading to current research into string theories, M-theory, and similar fields.

 

[Mercutio]

So one agrees it converges. ie, our measurments get closer and closer and closer to something, some parameter.

No. Our (you switched voice between third person and first person plural) confidence interval narrows, certainly, although it may (and will) jump around a bit depending on measurements...the "parameter" you speak of is a hypothetical construct, based on a theoretically infinite population; any data set we take is necessarily finite. Thus, the parameter in this case is estimated, and our estimate must always include a confidence interval (typically 95%, but 99% and others are done) recognizing that our finite sample cannot tell us the "truth" if the "truth" is an exact value of a parameter (which, of course, an a priori probability can give us).

So how does one know when to stop and say 'it converges'?

One has had this poorly phrased question answered in this thread, if one reads and understands.

 

[T'ai Chi]

One has had this poorly phrased question answered in this thread, if one reads and understands.

I must have missed it. You're welcome to point it out exactly, and specifically how it applies to the general question. Most answers I've seen simply take the word 'tolerance' and put it in different terms.

 

[Grimoire]

I must have missed it. You're welcome to point it out exactly, and specifically how it applies to the general question. Most answers I've seen simply take the word 'tolerance' and put it in different terms.

Try post #149.

 

[CFLarsen]

You misrepresent. It cannot converge on an a priori because there is none.

That's where T'ai Chi goes wrong. He does assume an a priori - which, in his case, is God. He is really arguing that we are approaching a scientific truth that will eventually prove God.

 

[joobz]

I must have missed it. You're welcome to point it out exactly, and specifically how it applies to the general question. Most answers I've seen simply take the word 'tolerance' and put it in different terms.

Post #147

I answered you directly your question

So, when do we stop science?
when no more questions are left to ask.
When will that happen?
?? I assume we'll know when we get there.
but until then, we keeping asking and then testing.
and asking and testing
and asking and testing
and asking and testing.

It's a vague answer.
But you ask a vague question. Don't expect better.

Want a better answer? Ask a better question.

Now, my complaint is multiple posts have given similar answers but you continually ignore those answers and only respond to the points that you wish to respond to. All you got out of my post was that "Aha! someone agrees it converges."

When the point to my claim was convergence as you hold it is not an accurate model. And we will stop science when there is no more questions to ask.

 

[joobz]

That's where T'ai Chi goes wrong. He does assume an a priori - which, in his case, is God. He is really arguing that we are approaching a scientific truth that will eventually prove God.

I've seen this critique of T'ai Chi in multiple posts of his. It seems that he likes to post threads that challenge science, reason, scepticisim....
But i don't get the feeling it's because he believes in god or believes in what he says. It seems to me that he's just a troll or enjoys challenging peoples views. So, the question is, do we respond because we like to argue as well or do we really believe that we are trying to convince him of something?

 

[Mercutio]

So, the question is, do we respond because we like to argue as well or do we really believe that we are trying to convince him of something?

Or because in previous cases (say, with Iacchus), we got feedback from lurkers who said they learned from and appreciated the threads despite the utter lack of demonstrated understanding on the part of one participant.

 

[Hellbound]

I just get a kick out of the threads. It's like the old "Who's on First?" bit.

TC: I'd like to talk about pie.

JREF: Okay, what about pie?

TC: See this pie? It's interesting.

JREF: Um, okay, sure. But it is fattening.

TC: Sorry, but that doesn't address any interesting points.

JREF: Um, what interesting points? I was commenting about the pie.

TC: Yes, but you obviously only focus on the side issues and ignore the main point.

JREF: What main point?

TC: If you don't know, I'm not obligated to tell you.

JREF: BUt you started this! You wanted to talk about pie! What about pie?!

TC: Typical, make it about the arguer, ignore the argument. Guess you can't argue against pie, huh?

JREF: Wha...? Is this Candid Camera?