I talked about David W. Hogg's "Why do we do astrophysics?" in our seminar today.
-
"You can’t do science if you don’t live within a network of trust. You have to trust your coauthors, you have to trust the literature, and you have to trust the machinery and tools that you use."
"A trusted partner is one that takes responsibility for their work."
5/6
@vicgrinberg This is the hill upon which science will either live or die on.
-
@vicgrinberg This is the hill upon which science will either live or die on.
@masek I honestly think that either of the points mentioned arel.
-
@masek I honestly think that either of the points mentioned arel.
@vicgrinberg IMHO if you win that point, all the others will be won at the same time.
I am not an astrophysicist or even a scientist (other than by education).
My job is mostly strategy. Foresee where and when battles will be fought and why... and to win them before anyone else knows they will happen.
We live in a time of historical crisis not seen for centuries.
José Ortega y Gasset defined such historical crisis as a change of fundamental ways of thinking.
The last crisis for him was the Renaissance where the belief-based mindset battled the knowledge-based mindset. The belief-based mindset lost.
Now we see the knowledge-based mindset being attacked. And the best way to defeat it would be to make a discourse about knowledge impossible.
As long as that discourse is alive, science cannot be vanquished.
But by destroying the network of trust, you make discourse impossible. You end up "my book says" vs. "the other book says" and with no way of resolving that other than by belief.
That is why my finger pointed there...
-
R relay@relay.infosec.exchange shared this topic
-
@vicgrinberg IMHO if you win that point, all the others will be won at the same time.
I am not an astrophysicist or even a scientist (other than by education).
My job is mostly strategy. Foresee where and when battles will be fought and why... and to win them before anyone else knows they will happen.
We live in a time of historical crisis not seen for centuries.
José Ortega y Gasset defined such historical crisis as a change of fundamental ways of thinking.
The last crisis for him was the Renaissance where the belief-based mindset battled the knowledge-based mindset. The belief-based mindset lost.
Now we see the knowledge-based mindset being attacked. And the best way to defeat it would be to make a discourse about knowledge impossible.
As long as that discourse is alive, science cannot be vanquished.
But by destroying the network of trust, you make discourse impossible. You end up "my book says" vs. "the other book says" and with no way of resolving that other than by belief.
That is why my finger pointed there...
@masek @vicgrinberg this is something I have been writing toward from a personal angle, as well as from a technology angle.
I note I am not alone in this. There are other isles of coherence. They kerp showing up.
There is a convergence here we are all mapping.
-
"You can’t do science if you don’t live within a network of trust. You have to trust your coauthors, you have to trust the literature, and you have to trust the machinery and tools that you use."
"A trusted partner is one that takes responsibility for their work."
5/6
@vicgrinberg Underrated comment
-
@vicgrinberg IMHO if you win that point, all the others will be won at the same time.
I am not an astrophysicist or even a scientist (other than by education).
My job is mostly strategy. Foresee where and when battles will be fought and why... and to win them before anyone else knows they will happen.
We live in a time of historical crisis not seen for centuries.
José Ortega y Gasset defined such historical crisis as a change of fundamental ways of thinking.
The last crisis for him was the Renaissance where the belief-based mindset battled the knowledge-based mindset. The belief-based mindset lost.
Now we see the knowledge-based mindset being attacked. And the best way to defeat it would be to make a discourse about knowledge impossible.
As long as that discourse is alive, science cannot be vanquished.
But by destroying the network of trust, you make discourse impossible. You end up "my book says" vs. "the other book says" and with no way of resolving that other than by belief.
That is why my finger pointed there...
@masek I see your point - though I disagree on "win this points, all others are won". I can easily imagine a set up where the trust is there, but people are only reading/learning/reproducing not doing. And that would also be an end of science as such.
But this is of course colored by me having read the whole paper and thus having a different feeling for what the individual citations mean as opposed to when one reads them hear out of the context.
(And would love to learn more about your work!)
-
@vicgrinberg Underrated comment
@gtsadmin this one blew my mind - it's a concept I struggled with phrasing for a while, really captured!
-
I talked about David W. Hogg's "Why do we do astrophysics?" in our seminar today. Thank to @aleks who posted about it here first!
I can only encourage everyone working in or interested in science and/or fundamental research to read it, it has broad relevance and the crucial parts, to me, are not even the LLM ones, but the ones that define the basis on which LLM use in astro is discussed.
️ https://arxiv.org/abs/2602.10181v1 A few quotes:
1/6
@vicgrinberg Thanks for the pointer! My area is mathematics, but---as you suggested---many of his comments are broadly relevant. I found his left-edge v.s. right-edge concepts around "clinical applications" interesting as a different presentation of some very well-worn issues in math research.
Separately (but not as much as I would like), his comment "We beat ploughshares into swords" is important, even if uncomfortable (maybe especially *because* it's uncomfortable).
-
@vicgrinberg Thanks for the pointer! My area is mathematics, but---as you suggested---many of his comments are broadly relevant. I found his left-edge v.s. right-edge concepts around "clinical applications" interesting as a different presentation of some very well-worn issues in math research.
Separately (but not as much as I would like), his comment "We beat ploughshares into swords" is important, even if uncomfortable (maybe especially *because* it's uncomfortable).
@nilesjohnson glad it was useful and thanks for letting me know!
I found the right/left edge part least convincing from the paper, in particular for astronomy (because a major "application" / edge issue is the perspective change that astronomy offers), & had the feeling that he himself struggled with in the text. But it's a good thought-provoking discussion anyway and worth thinking about.
And yeah, the ploughshares into swords ... especially for work that can only be done from space like mine
-
I talked about David W. Hogg's "Why do we do astrophysics?" in our seminar today. Thank to @aleks who posted about it here first!
I can only encourage everyone working in or interested in science and/or fundamental research to read it, it has broad relevance and the crucial parts, to me, are not even the LLM ones, but the ones that define the basis on which LLM use in astro is discussed.
️ https://arxiv.org/abs/2602.10181v1 A few quotes:
1/6
@vicgrinberg I agree with most of David’s comments. On a different topic, I’m a bit annoyed that there seems to be a double standard with respect to what is acceptable as a paper on the arxiv and what not.
-
R relay@relay.mycrowd.ca shared this topic
-
@vicgrinberg I agree with most of David’s comments. On a different topic, I’m a bit annoyed that there seems to be a double standard with respect to what is acceptable as a paper on the arxiv and what not.
@hannorein I feel that - being on archive gives the paper a gravitas that a random text on someone's website would not have (and at least to me it's not clear that this is intended for submission to anywhere else in any form...).
-
@hannorein I feel that - being on archive gives the paper a gravitas that a random text on someone's website would not have (and at least to me it's not clear that this is intended for submission to anywhere else in any form...).
Hogg has done that several times:
Magnitudes, distance moduli, bolometric corrections, and so much more
This pedagogical document about stellar photometry - aimed at those for whom astronomical arcana seem arcane - endeavours to explain the concepts of magnitudes, color indices, absolute magnitudes, distance moduli, extinctions, attenuations, color excesses, K corrections, and bolometric corrections. I include some discussion of observational technique, and some discussion of epistemology, but the primary focus here is on the theoretical or interpretive connections between the observational astronomical quantities and the physical properties of the observational targets.
ADS (ui.adsabs.harvard.edu)
A likelihood function for the Gaia Data
When we perform probabilistic inferences with the Gaia Mission data, we technically require a likelihood function, or a probability of the (raw-ish) data as a function of stellar (astrometric and photometric) properties. Unfortunately, we aren't (at present) given access to the Gaia data directly; we are only given a Catalog of derived astrometric properties for the stars. How do we perform probabilistic inferences in this context? The answer - implicit in many publications - is that we should look at the Gaia Catalog as containing the parameters of a likelihood function, or a probability of the Gaia data, conditioned on stellar properties, evaluated at the location of the data. Concretely, my recommendation is to assume (for, say, the parallax) that the Catalog-reported value and uncertainty are the mean and root-variance of a Gaussian function that can stand in for the true likelihood function. This is the implicit assumption in most Gaia literature to date; my only goal here is to make the assumption explicit. Certain technical choices by the Mission team slightly invalidate this assumption for DR1 (TGAS), but not seriously. Generalizing beyond Gaia, it is important to downstream users of any Catalog products that they deliver likelihood information about the fundamental data; this is a challenge for the probabilistic catalogs of the future.
ADS (ui.adsabs.harvard.edu)
Data Analysis Recipes: Products of multivariate Gaussians in Bayesian inferences
A product of two Gaussians (or normal distributions) is another Gaussian. That's a valuable and useful fact! Here we use it to derive a refactoring of a common product of multivariate Gaussians: The product of a Gaussian likelihood times a Gaussian prior, where some or all of those parameters enter the likelihood only in the mean and only linearly. That is, a linear, Gaussian, Bayesian model. This product of a likelihood times a prior pdf can be refactored into a product of a marginalized likelihood (or a Bayesian evidence) times a posterior pdf, where (in this case) both of these are also Gaussian. The means and variance tensors of the refactored Gaussians are straightforward to obtain as closed-form expressions; here we deliver these expressions, with discussion. The closed-form expressions can be used to speed up and improve the precision of inferences that contain linear parameters with Gaussian priors. We connect these methods to inferences that arise frequently in physics and astronomy. If all you want is the answer, the question is posed and answered at the beginning of Section 3. We show two toy examples, in the form of worked exercises, in Section 4. The solutions, discussion, and exercises in this Note are aimed at someone who is already familiar with the basic ideas of Bayesian inference and probability.
ADS (ui.adsabs.harvard.edu)
-
Hogg has done that several times:
Magnitudes, distance moduli, bolometric corrections, and so much more
This pedagogical document about stellar photometry - aimed at those for whom astronomical arcana seem arcane - endeavours to explain the concepts of magnitudes, color indices, absolute magnitudes, distance moduli, extinctions, attenuations, color excesses, K corrections, and bolometric corrections. I include some discussion of observational technique, and some discussion of epistemology, but the primary focus here is on the theoretical or interpretive connections between the observational astronomical quantities and the physical properties of the observational targets.
ADS (ui.adsabs.harvard.edu)
A likelihood function for the Gaia Data
When we perform probabilistic inferences with the Gaia Mission data, we technically require a likelihood function, or a probability of the (raw-ish) data as a function of stellar (astrometric and photometric) properties. Unfortunately, we aren't (at present) given access to the Gaia data directly; we are only given a Catalog of derived astrometric properties for the stars. How do we perform probabilistic inferences in this context? The answer - implicit in many publications - is that we should look at the Gaia Catalog as containing the parameters of a likelihood function, or a probability of the Gaia data, conditioned on stellar properties, evaluated at the location of the data. Concretely, my recommendation is to assume (for, say, the parallax) that the Catalog-reported value and uncertainty are the mean and root-variance of a Gaussian function that can stand in for the true likelihood function. This is the implicit assumption in most Gaia literature to date; my only goal here is to make the assumption explicit. Certain technical choices by the Mission team slightly invalidate this assumption for DR1 (TGAS), but not seriously. Generalizing beyond Gaia, it is important to downstream users of any Catalog products that they deliver likelihood information about the fundamental data; this is a challenge for the probabilistic catalogs of the future.
ADS (ui.adsabs.harvard.edu)
Data Analysis Recipes: Products of multivariate Gaussians in Bayesian inferences
A product of two Gaussians (or normal distributions) is another Gaussian. That's a valuable and useful fact! Here we use it to derive a refactoring of a common product of multivariate Gaussians: The product of a Gaussian likelihood times a Gaussian prior, where some or all of those parameters enter the likelihood only in the mean and only linearly. That is, a linear, Gaussian, Bayesian model. This product of a likelihood times a prior pdf can be refactored into a product of a marginalized likelihood (or a Bayesian evidence) times a posterior pdf, where (in this case) both of these are also Gaussian. The means and variance tensors of the refactored Gaussians are straightforward to obtain as closed-form expressions; here we deliver these expressions, with discussion. The closed-form expressions can be used to speed up and improve the precision of inferences that contain linear parameters with Gaussian priors. We connect these methods to inferences that arise frequently in physics and astronomy. If all you want is the answer, the question is posed and answered at the beginning of Section 3. We show two toy examples, in the form of worked exercises, in Section 4. The solutions, discussion, and exercises in this Note are aimed at someone who is already familiar with the basic ideas of Bayesian inference and probability.
ADS (ui.adsabs.harvard.edu)
@knud @vicgrinberg He can do that. Not many others would get past the moderators with these kind of "papers".
-
@knud @vicgrinberg He can do that. Not many others would get past the moderators with these kind of "papers".
@hannorein @knud yeah...
-
@knud @vicgrinberg He can do that. Not many others would get past the moderators with these kind of "papers".
Which is a shame. The 2022 writeup on distance moduli, magnitudes, k-correction and other things is better than anything I've seen in a textbook so far. So in my opinion it's a very valuable resource.
-
"People are always the ends, not merely the means. [...]
When we employ a graduate student to perform some work, it absolutely must be because the graduate student will benefit from that work, not merely because that work needs to get done."
3/6
@vicgrinberg So obvious, and yet this seems like a controversial statement in much of academia.
-
"You can’t do science if you don’t live within a network of trust. You have to trust your coauthors, you have to trust the literature, and you have to trust the machinery and tools that you use."
"A trusted partner is one that takes responsibility for their work."
5/6
@vicgrinberg This, 1000 times
.
"Why won't you work with xyz, are you not a team player?" neglects the lack of trust that had been demonstrated time and time before, and yet people don't seem to think it matters. -
"Practice of astrophysics cannot be learned from reading. [...] If you want to become an astrophysicist, it isn’t sufficient to read or take classes. You have to do it, and doing it requires doing novel things, that haven’t been done before, and which connect to important scientific questions in the literature."
2/6
@vicgrinberg In paleontology, there is (and has been) a small but noisy cadre of what have been referred to as "armchair paleontologists": people who "do" paleontology (mostly taxonomy, really) without ever going and studying actual specimens; only making pronouncements based on what others have published. Some have made positive contributions, but most have been decidedly detrimental to the science. They're often so loud that they are difficult to simply ignore, and many have done real damage.
-
@vicgrinberg IMHO if you win that point, all the others will be won at the same time.
I am not an astrophysicist or even a scientist (other than by education).
My job is mostly strategy. Foresee where and when battles will be fought and why... and to win them before anyone else knows they will happen.
We live in a time of historical crisis not seen for centuries.
José Ortega y Gasset defined such historical crisis as a change of fundamental ways of thinking.
The last crisis for him was the Renaissance where the belief-based mindset battled the knowledge-based mindset. The belief-based mindset lost.
Now we see the knowledge-based mindset being attacked. And the best way to defeat it would be to make a discourse about knowledge impossible.
As long as that discourse is alive, science cannot be vanquished.
But by destroying the network of trust, you make discourse impossible. You end up "my book says" vs. "the other book says" and with no way of resolving that other than by belief.
That is why my finger pointed there...
@vicgrinberg @masek This is why librarians and others teaching about the evaluation of information sources must clearly connect the format, process, etc. to the creators and reviewers of the products! Ultimately, 'do you trust the resource?' is, 'do you trust the people involved in the process of it existing?' #library
-
"People are always the ends, not merely the means. [...]
When we employ a graduate student to perform some work, it absolutely must be because the graduate student will benefit from that work, not merely because that work needs to get done."
3/6
@vicgrinberg
I don't really understand this one. Will work that 'needs to get done' in most cases not also benefit the student?
