Offline Oami

  • *
  • Posts: 88
    • View Profile
What makes one theory better than other?
« on: May 20, 2017, 08:32:48 PM »
In science there are different kinds of theories. While some theories may peacefully coexist, some can not. When two (or more) theories conflict with each other, we may need to decide (assuming we are interested in the topic in the first place), which theory is better.

So, what methods do you actually use to determine, which theory to believe in? There may be several different answers, but I'll put my criteria here.

In order for theory A to be better than theory B, one of the following three conditions must be met:
1) B is proven false and A is not.
2) Neither is proven false, but A explains things better than B.
(That is: it explains a greater number of things or more important things relevant to the topic.)
3) Neither is proven false, and both explain things equally well, but A is simpler than B. (This is also known as the Occam razor.)

It is worth noting that if a theory is false, it might be proven false: but if a theory is true, it can never be proven true. If a theory is true, we can test it, and we will have the results that are predicted by the theory: but if we test a theory a million times and every time get the right result, that doesn't prove that the million-and-first test will also give the right results, instead of failing because of some reason that the first million tests didn't take into account.

Let's have an example. First, a problem: why does a flashlight work? And then four theories to answer it:
1) It has normal fireflies in it.
2) Strawberries are red.
3) It has fireflies in it, and those fireflies are so small that they cannot be detected, only their light can be detected, and they have friends in the switch that telepathically tell them when the switch is pressed and it's time to start glowing, and other friends in the battery who telepathically tell them that the battery is old enough so then don't need to glow anymore. We don't yet know where the fireflies get their energy and what that has to do with the battery.
4) The switch closes an electic circuit, and the resulting electic current causes a reaction in the material used in the light bulb, making it glow.

Theory 1 is the worst. It fails at step 1 when compared to all others: we can prove it false by breaking up the light bulb and seeing that there are no normal fireflies.
Theory 2 is slightly better. It cannot be proven false, but it fails at step 2 when compared to theories 3 and 4: it doesn't explain things.
Theory 3 is an improvement of theory 1. It takes into account what caused theory 1 to fail in the first place. However, it fails at step 3 when compared to theory 4: it is more complex.
And so, theory 4 is the best one here.
« Last Edit: June 07, 2017, 12:10:58 AM by Oami »


Offline ﮎingulaЯiτy

  • *
  • Posts: 20
  • Rationality prevails.
    • View Profile
Re: What makes one theory better than other?
« Reply #1 on: August 11, 2017, 12:21:38 AM »
Your post appears to be a solid breakdown and covers the basics quite well. However, if I was trying to explain the them to someone completely scientifically illiterate, I'd probably elaborate more on condition 2.

The notion of "explaining things better" may seem a tad arbitrary to a newcomer unfamiliar with scientific rigor. The strawberry hypothesis :) will be obviously inferior to the other theories because it is completely non sequitur. The mistake I see people making most often, is crediting a hypothesis due to its descriptive power (explanations made after the relevant observations). Any hypothesis with descriptive power would still be far inferior to models with predictive power (falsifiable explanations offered in advance of new empirical observations).

The strawberry example has neither descriptive nor predictive power so it may not prepare people for noticing the shortcomings of retroactive explanations, when they are looking for a "better explanation".

I'd also remark that theories are made even stronger by their applicable scope. Ex. If you've seen "A Beautiful Mind", John Nash won his Nobel Prize for the Nash Equilibrium, because his game theory approach worked in so many different disciplines besides economics. It created insights in topics he had never considered like sociology, politics, insurance, conflict analysis, fiances, etc. I recently watched a numberphile video in which squared squares' network graphs gained insights from electrical models of circuits with various amps.

Even under a smaller less-abstracted scope, we can also use consistency checks against other working theories tangential to your current hypothesis. It's like sculpting the shape of a puzzle piece and then testing how it fits with its neighbors in the puzzle. Ex. Successful biology theories suggest that the nano fireflies hypothesis would require an oxygen mixed atmosphere in the flashlight for them to breathe and to fuel their bioluminescent reactions. In contrast, our models of electrochemical reactions suggest an incandescent filament would rapidly burn out in the presence of oxygen.

Consequently, we can make a prediction about the construction of the flashlight prior to investigation. This means we can ascertain the accuracy of an explanatory model, by measuring its predictive success. If sealed part of the flashlight houses a vacuum (or at least the absence of an oxidizer; noble gasses work because they are inert), the incandescent filament hypothesis is credited. If oxygen is present, the nano-fireflies bioluminescence hypothesis would be credited.

...I suspect you essentially knew all of these things, but I took this thread as an invitation to try to collaboratively narrow down the definitions and criteria.

Ultimately this thread seems to dissect the tenets of rational investigation, but I'll try not to hijack your thread and ramble about why Occams Razor is valid, or how a phenomenon could be mysterious to some particular person, but there could be no phenomena mysterious of themselves. ("Mysterious answer" is a contradiction in terms, and worshiping a sacred mystery is just to worship your own ignorance.)

In disciplines other than mathematics, the term "theory" denotes the highest possible status an explanatory model can ever attain by rigorously proving itself.


Offline CriticalThinker

  • *
  • Posts: 159
  • Polite and Pragmatic
    • View Profile
Re: What makes one theory better than other?
« Reply #2 on: September 10, 2017, 07:44:07 PM »
If you drop the word theory and replace it with hypothesis, you're basically on track.  Statement #2 is neither hypothesis nor theory as it's not testable or falsifiable.

A theory is formed from a lot of hypothesis testing under a broad series of variables.  Something doesn't move from hypothesis to theory until it has been tested from every different angle that we can image and still be supported by the evidence.  Multiple conflicting hypotheses about a single testable set of variables do not coexist after the test has been made and the falsified ones identified.  They are dismissed as they have been proven false.  Even when a concept has moved from hypothesis to theory, it is never safe from falsification.  As technology advances, new variables are discovered or measurement capabilities are improved, it can always lead to a scenario where the theory is falsified and we move on from there.

A belief that is held onto despite falsifying evidence is called a Dogma.  I'll use your flashlight experiment to illustrate.  You're told statement #1 is true and upon breaking the bulb and seeing the absence of fireflies, you continue to state that statement #1 is still true.  It is now officially a dogma, not a hypothesis or theory.  You have falsified the statement through reasonable experimentation yet you haven't dismissed it.  The only way that dogma can go back to being a hypothesis at this point is for someone to develop a new technology that somehow demonstrates that there really are fireflies in the bulb that you couldn't see earlier.  At this point new experiments must be conducted to see if this holds true under a variety of different circumstances like time of day, temperature, day of the month or any other variable that you can dream up which could again falsify your hypothesis.

Thank you,

Quod gratis asseritur, gratis negatur


Offline Rama Set

  • *
  • Posts: 9845
  • Round and round...
    • View Profile
Re: What makes one theory better than other?
« Reply #3 on: September 10, 2017, 09:44:00 PM »
Just wanted to elaborate on dogma. Dogma isn't something that is believed even after it has been proven false. Dogma is something whose belief in it does not rely on it being true or false.
Th*rk is the worst person on this website.

Re: What makes one theory better than other?
« Reply #4 on: December 16, 2017, 05:21:32 PM »
Occam's razor doesn't always work very well when given to "magical thinkers".

After all, doesn't Newton's theory as to why the universe is held together "in God's hand" answer the question much more simply than modern theories of "Dark Energy"?

For that matter, is it possible that simplifying the many paradoxes of scientific measurement by theorizing "dark energy" is also a failure to seek the more complicated--yet more accurate--hypothesis?