Thursday, August 20, 2009

What Do You Do With Theories That Make Both Testable and Untestable Predictions?

I would like people's opinions on this:

I keep alluding to how modern cosmologists think the universe is actually an eternal structure, perhaps even a multiverse with many different types of universes.

Look, I'm not just making this stuff up because it is fun. It is a real prediction of inflation. I will quote Alan Guth himself: (This guy is a father of inflation, and if anyone is going to win the Nobel Prize for it its him.)
In this paper I have summarized the arguments that strongly suggest that our universe is the product of inflation. 
He names several predictions of inflation that have all been confirmed. It is a remarkable theory. In addition, inflation makes other where predictions people are divided about their testability:
Next I turned to the question of eternal inflation, claiming that essentially all inflationary models are eternal. In my opinion this makes inflation very robust: if it starts anywhere, at any time in all of eternity, it produces an infinite number of pocket universes... For that reason it is important for us to learn more about the evolution of the multiverse during eternal inflation.
He then says(which is my whole point):
It is the success of [the testable] predictions that justifies spending time on the more speculative [predictions] of inflationary cosmology.
So my question: what are we to do with such a theory? It makes several predictions, most of which can be tested and so far have past every test with flying colors. However, there are these other predictions that may not be testable. What is the more rational thing to do?
  1. Dismiss the "untestable" predictions as philosophy?
  2. Say, "since all predictions we can test pass with flying colors, the theory is probably true so we should take seriously all predictions, even those (perhaps) not testable"?
  3. Something else.
As you may have guessed, I fall into the second camp. Inflation is real. It has been verified on nearly every level it can possibly be verified. (Still a couple left I admit). It actually predicts the universe is eternal and that there are probably an infinite number of "pocket universes" out there in addition to ours. This is a real prediction and it is my opinion we should take it just as seriously as we take the others.

So what are you thoughts? What should we do with such a theory? What do we do with these possibly untestable predictions of inflation?


  1. The idea of inflation is a good concept, but here's the problem I have with saying that it has testable predictions: it is a testable theory in the sense that the concept fits observations, but as far as I understand it, the theory relies on some essentially ad hoc field to drive inflation. So in some ways, it's not surprising that inflation works as well as it does - you can make inflation any way you want because we have absolutely no idea what drives it.

    In some ways its like saying the light-bulb elves have exactly the properties to fit our observations of light bulbs. Now of course it relies of far better physical reasoning, but it still seems to be satisfying observations rather than making predictions.

  2. "as far as I understand it, the theory relies on some essentially ad hoc field to drive inflation"

    That's not true, you do not need some scalar field to drive inflation. The predictions are independent of the mechanism. It is true we don't know what the mechanism that drives in inflation is, but we do know what the predictions are.

    Inflation is the idea that at some time in history the comoving Hubble radius (aH)^(-1) was getting smaller. That's it, nothing more and nothing less.

    This made very specific predictions that were made well before these things were established:

    1. Flatness (Nobody at the time knew whether our universe was flat. It turns out it is really flat, the kind of flat that is hard to believe without the comoving Hubble radius shrinking at some point, or in other words: inflation)
    2. Nobody was sure until COBE, after inflation was proposed, that there were scale invariant anisotropies in the CMB. These were a prediction of inflation. They have to be at in very precise levels and be scale invariant if inflation was true.
    3. Etc.. I will blog more on this later.

    Look, I will not deny we don't know the mechanism for sure. But we are sure the comoving Hubble radius at some point shrank, or inflation, is real and as Guth and others have noticed, all ways you look at the issue suggest that the universe one inflation begins, continues eternally creating other pocket universes.

    My point is is, as Guth says, eternal inflation seems pretty robust, despite the mechanism that caused it.

  3. By the way, I am glad Nick is here to keep the theorists honest. But, like a good theorist, I need to stand by my claims. :)

    Thanks again Nick.

  4. I suppose that I'm not trying to bad mouth inflation - I understand that it solves some very real and longstanding issues - but it still does so by employing a physical field that is created simply to fit the theory.

    In many ways, whatever drove inflation and whatever is driving the acceleration of the universe today (so-called dark energy) are a lot like the 19th century concept of aether. Aether was postulated to exist because nobody could get their heads around a universe without it and it "solved" some nice problems with theories like electromagnetism. I'm not saying inflation didn't happen (it almost certainly did) or that the expansion of the universe isn't accelerating (it almost certainly is), but I think the inflaton and dark energy exist simply because we can't get our heads around a universe without them (that also fits the evidence).

  5. I agree inflation, dark energy, etc... are to be viewed with more skepticism then some very established laws in physics, like the inverse square law of gravitation.

    But I even think comparing it with "the 19th century concept of aether" is not fair.

    Aether, as you know failed the famous test it was given: Michelson/Morley.

    But inflation, dark energy, dark matter, acceleration, have withstood every experiment thrown at them. In every instance, the predictions you get from these theories are confirmed.

    I'm not saying they shouldn't be considered with a healthy amount of skepticism, but you can't compare them to theories that failed in experimental settings. Nor can you compare them to theories that haven't made experimental predictions since they have.

  6. Nick, I believe was your point which is a good one that: Inflation, dark energy, dark matter, etc... are black boxes. This is true.

    But, we know enough about these black boxes to make predictions and so far so good.

  7. My non-physicist perspective is that I agree with option #2. The boundary between science and pseudoscience is fuzzy, and testability is often invoked to help guide judgment of which side ideas fall on. It is important, yes. But to make a big deal about it in the context you describe is overdoing it and unnecessary, IMO.

  8. Thanks Jared*. I'm glad to see that there are more #2s out there.

    Sorry I this post was overdoing it.

  9. "Inflation, dark energy, dark matter, etc... are black boxes"

    I think you hit what I'm getting at in your last comment, Joe. It's not that I don't believe the expansion of the universe is accelerating or that shortly after the Big Bang the universe underwent a brief period of super-luminal acceleration. It's just that our mechanisms for explaining these things are black boxes. We don't know what did it, simply that something did it in the same way that aether gave light something to propagate through (and people liked the idea of a universal rest frame).

  10. Joseph,

    To be clear, I did not mean that you were overdoing it. I meant that people who act like untestability automatically puts a hypothesis on weak epistemic grounds are being too pedantic.


To add a link to text:
<a href="URL">Text</a>