I disagree with this. While prediction is more convincing, I think that there is a lot of value in postdiction and data fitting in the sciences. I have a lot of collegues who only do postdictions, and never make predictions, and I do not think that they are any less scientists, much less pseudoscientists.
Of course there is, but not at the expense of testing predictions.
I’m skeptical. Would you please send me a lot of their names?
Technically, at least chronologically speaking, Kepler’s work was postdiction. Of course, we always refer to this as predictions. Because his predictions are still logically prior to the observations, even if they are chronologically posterior.
Another way of describing “logically” is that the only chronology that matters is when the person(s) using the hypothesis to predict know the answer, not when the event occurred.
It’s a really easy and powerful way for laypeople to learn. Construct a hypothesis, test. It doesn’t matter whether the answer has been known for centuries, as long as you don’t know it when you use the hypothesis to predict.
You should try it, Sal.
The entirety of papers V and VI on the Event Horizon Telescope (First M87 Event Horizon Telescope Results. V. Physical Origin of the Asymmetric Ring - IOPscience) is postdiction. Among them are many authors where this is their only project - these authors never make any predictions, just postdictions. I consider them real scientists.
Quite a bit of Genomics work is chronologically post diction, at least from an experimentalists point of view.
How do you explain this?
We compare the observed visibilities with this library and confirm that the asymmetric ring is consistent with earlier predictions of strong gravitational lensing of synchrotron emission from a hot plasma orbiting near the black hole event horizon.
They are testing predictions! Or are they misinformed?
We often call postdictions “predictions” when they are logically prior.
I consider it postdiction as the image is taken, and then it is compared to many theoretical computations of different models that spans a large portion of the possible configuration. The model that fits the image is then considered to be closest to be “correct”.
If they start with one model, and then test it against the observation, I will consider that prediction. Perhaps we have differrent notions of what predictions/postdictions are.
I sure don’t. That’s testing of predictions, and I’m pretty confident that no one considers the best model to be correct, just the best, and a foundation for refinement and testing of more predictions.
I don’t see how the number of hypotheses is relevant to any of these definitions, except that when we get a chance to test two that make different predictions with the same experiment, it’s a bonus. I’m sure that happens more often in cell biology and neuroscience than it does in astrophysics.
We clearly do.
Here’s more testing of predictions:
“Overall, the observed image is consistent with expectations for the shadow of a spinning Kerr black hole as predicted by general relativity. If the black hole spin and M87’s large scale jet are aligned, then the black hole spin vector is pointed away from Earth. Models in our library of non-spinning black holes are inconsistent with the observations as they do not produce sufficiently powerful jets. At the same time, in those models that produce a sufficiently powerful jet, the latter is powered by extraction of black hole spin energy through mechanisms akin to the Blandford-Znajek process. We briefly consider alternatives to a black hole for the central compact object. Analysis of existing EHT polarization data and data taken simultaneously at other wavelengths will soon enable new tests of the GRMHD models,”
Basically, the whole thing is about testing predictions!
@mercer, you are contrasting prediction with postdictions. The thing you are missing is that many so-called predictions are actually postdictions in one sense or another.
This doesn’t make them invalid if no cheating takes place. But prediction does not mean what you think it does.
Here’s my take:
Paper VI, for example, fits for the mass of the black hole. The way this was done is through fitting for models of black holes of different masses, and each is then compared to the image. The one that is closest to the image is then taken to give the correct mass of the black hole.
I would say that is a postdiction of the mass of the black hole, i.e. the correct mass is chosen because it is the one that fits the image. However, one can say that the M=Mtrue model correctly predicts the mass of the black hole.
Those are hypotheses that make empirical predictions.
That’s much clearer. The chronology pertains to when you have the data in hand. I wouldn’t call the hypothesis true, just the best hypothesis.
There is several ways to define pre and post. We can understand these terms in relation to:
- When the data was collected by someone, even though the data could be produced in the deep past.
- When the data was formed in an experiment done to test the prediction.
- When the data was seen by the person making the prediction, even if others already have seen it.
- Analytically, whenthe analysis of the data was done or reported.
- Logically (not chronologically) ordering reasoning
Molecular biologists place a premium on #2. However, several branches of science, often by necessity, are focused on defining prediction in relation to other points.
For this reason, postdiction and prediction are relative terms. One person’s postdiction is another’s prediction.
My use of the term “prediction” is consistent with its usage in that abstract that was agreed upon by ~200 authors.
Yeah, but the hypotheses spans the entire possible range of outcome, from M=0 to M=however large you want. One of the models in the hypotheses is bound to fit into the data. I would call that postdiction, as from the hypotheses alone, it’s impossible to say what the mass of the black hole is.
I wouldn’t, nor did the ~200 authors you chose to cite.
I see it like this:
Question: what is the mass of the black hole?
A prediction will be: the mass is M=5.
Then we can check if the prediction is correct by taking the image of the black hole.
A postdiction will be: first take the image of the black hole, then fit which M fits the data. This is what the EHT collaboration did.
Because astronomers very rarely use the word postdiction, and the collaboration wasn’t really interested in getting the prediction/postdiction terminology right.
I don’t see anything of the sort there. They are explicitly describing themselves as testing predictions:
" Consistent with the discussion in Section 2, we now adopt the working hypothesis that M87 contains a turbulent, magnetized accretion flow surrounding a Kerr black hole. To test this hypothesis quantitatively against the EHT2017 data we have generated a Simulation Library of 3D time-dependent ideal GRMHD models."
The simulation library contains the predictions.
That’s seems incredibly arrogant. If they are so wrong about basic terminology, then maybe you are just the person to straighten them all out!
Come on. I’ve never used the term in my writing, and there are 16 instances of “predict” and none of “postdict” in the paper YOU cited. Definitions are based on usage.
The simulation library contains parameters like the mass of the black hole, that can be tweaked to fit the data in the sense that I wrote:
It’s the truth though, astronomers do very rarely use the word postdiction… The EHT collaboration cares about taking pictures of black holes, not getting epistemological terms right.