10 ms·
Signs of the Times
- basicplus2 7y ago<Some images are less likely than others and it is my job to design an algorithm that gives more weight to the images which are more likely>
- userbinator 7y ago..."based on what I think is more likely".
- yorwba 7y agoDon't forget the observations that determine the image content. "what I think is more likely" is about describing the relationship between observations and image, whereas the most likely image is produced from actual observations using that relationship to draw inferences.
- rafaelvasco 7y agoNot entirely related with the article, will probably be downvoted knowing HN but the title makes something come to mind: Science is being so hold back by scientists that are too afraid to go beyond their model. Someone in a book wrote that Science is a religion too in a sense because it is as dogmatic as any religion. I tend to agree.
- snovv_crash 7y agoScience progresses one funeral at a time?
- dsfyu404ed 7y ago> <subject> progresses one funeral at a time? One of my favorite quips along those same lines is "X is Y years of progress being held back by Z years of tradition."
- alexithym 7y agoIt's incorrect to compare science to religion. In science, there are no facts that are presented as being undeniably true. Rather, all scientific facts can be independently derived and understood to be true by any person. This emphasis on reproducibility means that no scientific lesson has to be taken on faith, eliminating any sense of dogma. Furthermore, as our understanding of the world improves, we often challenge dated scientific ideas and prove them wrong using empirical evidence. If science was truly dogmatic this would never happen, because dogma requires facts to be presented as being undeniably true, forever.
- SimplyUnknown 7y agoIf reproducibility is so important then why are we in the middle of a reproducibility crisis? [1] If reproducibility was really that important then the scientific body would be more focused on replicating results rather racing to being the first group to publish an idea or securing funding. Scientific foundations are seldom challenged or refuted. The last paradigm shift in physics or maths was over one hundred years ago. And as soon as someone challenges them (as the author does) the criticisms are refuted. Say what you will, but science is mostly a social affair. Works of prominent groups are more likely to be accepted than works of groups with less status in the scientific community. Belief in scientific work may not rely on faith, but it is a far cry from pure scientific method. I'm not trying to be cynical here and do think a good scientific body advances humanity, but the current state of the scientific community might be closer to religion than one hopes it should be. [1] https://en.wikipedia.org/wiki/Replication_crisis https://en.wikipedia.org/wiki/Replication_crisis
- alexithym 7y agoBecause we are human, science in practice will always fall short of perfect adherence of the scientific method. However, the fact remains that the two are fundamentally different, because one is existentially founded on dogma, and the other isn't.
- jaredgorski 7y ago
- eesmith 7y agoExcept for all those times that scientists went beyond their model and overturned the previous models. Overturned scientific models: Lyellian uniformitarianism, the pre-Agassiz model that the Earth was gradually cooling so Ice Ages couldn't have existed (see https://en.wikipedia.org/wiki/Ice_age#History_of_research https://en.wikipedia.org/wiki/Ice_age#History_of_research ), Newton's theory of gravitation, and classical statistical mechanics (the "ultraviolet catastrophe"). Plus, the wide-spread belief that "all models are wrong" - https://en.wikipedia.org/wiki/All_models_are_wrong https://en.wikipedia.org/wiki/All_models_are_wrong - or the modified "all models are wrong, but some are useful". Which doesn't match the dogmatism attributed to religion, does it?
- dnautics 7y agoWell yeah, but those are all rather old results.
- eesmith 7y agoAnd your username contains three vowels. That is, I fail to see what your comment adds to the topic of how dogmatism in science is equal to that of religion, causing scientists to be too afraid to go beyond their model. I'll cherry pick some titles from the 9000+ results of a search for "a new model" on Goggle Scholar for the last year, at https://scholar.google.com/scholar?as_ylo=2019&q=%22a+new+model+for%22&hl=en&as_sdt=0,5 https://scholar.google.com/scholar?as_ylo=2019&q=%22a+new+mo... : > A new model for the biochemistry of pectin synthesis: GAUTs synthesize diverse HG glycans in structurally and functionally distinct plant cell wall polymers > The needle model: A new model for the main hydration peak of alite > Barriers to Hydraulic Fracture Height Growth: A New Model for Sliding Interfaces > A new model for MR evaluation of liver function with gadoxetic acid, including both uptake and excretion > A new model for esker formation sheds light on the processes within subglacial tunnels > Theorizing Transgender Identity for Clinical Practice: A New Model for Understanding Gender > Mutation of Murine Rpl5 reveals a New Model for Diamond Blackfan Anemia Characterized By Defective Erythropoiesis How do I conclude that scientists are too afraid to go beyond their existing models?
- m4r35n357 7y agoI bet you can't back up a single one of your claims. Are you actually qualified to make them?
- CaptArmchair 7y agoWhat relevant measure would you use to support your claim that 'science is hold back'? And what arguments support your claim that scientists are by far the root cause to this? 'scientific progress' is a paradox. There is no absolute way of measuring the velocity at which science advances. You can't translate this into a value on a speed-o-meter and then make assertions based on what the dial says. The entire point of science is the discovery of new knowledge and insights about reality. It's the pursuit of trying to know the unknown. Inherently, it's not possible to know up front (a) how long it will take to validate your thesis and (b) whether validating your thesis will yield the result you hoped to find (i.e. a cure for cancer) Looking back at science in the past and then comparing that to today's state of affairs is hindsight bias. "But, it used to be that scientists got more funding to do research. Think about the Race to the Moon!" But then you forget that science isn't practiced in a vacuum. Never has. Scientists happen to work in earnest on particular questions because the external circumstances are aligned in a way that enables - or disenfranchises - them to work on those questions. Older research always predicates current research. For example, there would have been little point to assume that machine learning could have been a thing in the late 1800's as neither Von Neumann nor Turing had come around to set the stage. Science begets science, and you can't rush that retroactively. Then there's politics and economics. Full time science comes at a premium. Science happens where there's money. Either through public grants, in academic institutions or large institutions with a sizable R&D section. As such, the type of science that dominates an era depends largely on the zeitgeist. Reducing science to pure quantitative research or the empirical method is intellectual dishonest as well. Any research is based on a foundation of need, curiosity and imagination. As such, the number of papers published or conferences attended are very much skewed reflection of the generated value.
- rafaelvasco 7y agoSorry, 'science is held back' is the correct term. Thank you.
- Psyladine 7y agoSpecifically in mathematics, Grigori Perelman abandoned the field due to political favoritism by Shing-Tung Yau. In more *systemic terms, a cabal of researchers focused effort (and more importantly, limited funding) to only one aspect of alzheimers research: https://news.ycombinator.com/item?id=21911225 https://news.ycombinator.com/item?id=21911225
- jacknews 7y agoThe real sign of the times is how a novelist can seemingly claim that her opinion that black holes don't exist is just as legitimate as the thinking of the entire community of expert physicists, cosmologists, etc, and the results of this huge international collaborative project.
- dnautics 7y agoLet's not forget that this 'novelist' has a PhD and a decade of professional post-phd work in high energy physics. She has more real research experience than probably at least a standard deviation from the median physics funding program manager, for example.
- jacknews 7y agofair enough, but she should be making her arguments with papers.
- sneak 7y agoWebsites are fine too.
- latortuga 7y agoUncalled for, gatekeeping is a distraction. If you have quibbles with the argument then debate on its merits.
- dnautics 7y agoYou've never been a scientist, have you? Maybe she doesn't want to work as an extremely underpaid laborer. Maybe she doesn't want to take emotional abuse from incompetent Pi's. Maybe she just got tired of it. Maybe it just doesn't spark a joy anymore. Just because she stopped doing science professionally doesn't invalidate her voice; after all this work is mostly taxpayer funded so in the end it all ought to be accountable to everyone, PhD or no.
- jacknews 7y ago
- Psyladine 7y ago>I should mention pre-existing biases; my default setting is skeptical. I don’t really believe that black holes exist Wonder if dark matter/energy will turn out to be the Aether of the 21st century.
- pete762 7y agoWhat does dark matter have to do with black holes?
- davedx 7y agoThey are related topically in that yesterday new research cast the very existence of dark matter into doubt, and this blog post is kind of trying to do the same (albeit with less rigour than the actual scientific paper posted here yesterday).
- rob74 7y agoWasn't that dark energy in yesterday's article?
- m4r35n357 7y agoDark matter is NOT the same thing as dark energy.
- AnIdiotOnTheNet 7y agoAs another reply mentioned, the paper in question cast doubt on one of the key pieces of evidence for Dark Energy, not Dark Matter. Despite the similar names, they are not fundamentally related concepts and there is significantly more evidence for the existence of Dark Matter.
- mumblemumble 7y agoWell, MACHOs are one hypothesis for explaining dark matter, and they're typically considered to include (but not be limited to) black holes.
- mumblemumble 7y ago
- abnry 7y agoTL;DR "I don't believe black holes exist and it is easy to p-hack your image processing algorithm, so I don't buy the media hype of the first image of a black hole."
- dnautics 7y agoI didn't believe the amyloid hypothesis of Alzheimer's and I complained it was easy to p-hack all sorts of observations, and then I quit science.
- ruarai 7y agoThis article read a little annoying. I don't know why this PhD holder failed to find the actual paper describing the algorithm: https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Bouman_Computational_Imaging_for_CVPR_2016_paper.pdf https://www.cv-foundation.org/openaccess/content_cvpr_2016/p... The author is making it sound like the work is just p-hacking when it's /very clearly/ not. This single paragraph in the Method section explains it best: > "Reconstructing an image using bispectrum measurements is an ill-posed problem, and as such there are an infinite number of possible images that explain the data [28]. The challenge is to find an explanation that respects our prior assumptions about the “visual” universe while still satisfying the observed data." This is literally all of physics! You can say that Newtonian physics is 'just curve fitting' just the same. I don't know enough about the philosophy of science to make a further argument here but clearly the author does not either.
- whatshisface 7y agoThe allegations are that the black hole image could have turned out to be anything, given different starting assumptions about what it would look like. The exact nature of this variation, along with the degree to which the other possibilities would be plausible, is up for debate. Unfortunately that debate is very technical and not suited for general-audience blog posts.
- CrazyStat 7y agoThe author of the post ignores all the work that went into validating that the methods they used give reasonably consistent results, including simulation studies and multiple blind analyses to make sure that they weren't just making the image they wanted to see. This is discussed in the Caltech talk which is embedded in the post. It seems the author cherrypicked all the quotes that sound bad and ignored the parts that disagree with her narrative.
- raverbashing 7y agoYeah this article is clickbait I would paraphrase that "saying" about teaching as thus: "those who can't do, nitpick" The TED talk was for a different algorithmic technique. Couple that with assertions like "An image with a hole or without a hole looks the same in the amplitude domain." and I see that this article has as much credibility as climate deniers. Same thing with the Ligo "naysayers" who tried to disprove the experiments without knowing the basics of signal processing using FFT, etc
- dzdt 7y agoIts ironic. The author formed a theory that the black hole image research was flawed, specifically that the researchers are filtering for images with the right shape and merging these to produce an image similar to predictions. The author then looks at various talks and presentations and filters for isolated statements that out-of-context support the idea of the research flaws. The author has a flawed approach exactly analagous to the accused flaw in the black hole research.
- dnautics 7y agoIn one case data directly about the object in question were thrown out, in the other case, data were ignored (analogously: throwing out images of stars when cataloging galaxies based on priors of galaxy spectral properties vs star spectral properties)
- iamthepieman 7y agoso if the data could represent any image (not sure if the term "any" is truly infinite or just very large) then at least Bouman's algorithm and work sets a stake in the ground and challenges other scientists to explain why it does or doesn't belong there. Reading this article makes it seem like the work is a scientific version of my kids artwork. Interesting to the right audience, tells you a lot about who created it but bearing only a slight semblance to reality.
- CrazyStat 7y agoIf you watch the Caltech talk (embedded in the blog post) or read the original published paper, they did quite a bit of work to validate that their method was actually giving them good results and not just what they wanted to see. The author of the post doesn't mention any of this, presumably because it doesn't fit her narrative.
- losvedir 7y agoI think this article is running off on a wild goose chase. I remember looking into this when it all came out and I don't think Prof Bouman's algorithm was actually used for that picture. Her TED talk was great, and I think that's why everyone latched onto it, but IIRC they ended up using a different reconstruction algorithm. Some people were unreasonably concerned with who actually did what with that image, and I wouldn't normally care about it. It's just that this article is trying to cast doubt on the existence of black holes by attacking an algorithm that wasn't even used to generate the picture they're showing on the page.
- raxxorrax 7y agoWithout phase information the cat probably looks like a black hole if transformed back into image space.