jpeloquin

Pretty much yes. Critical analysis is a necessary skill that needs practice. It's also necessary to be aware of the intricacies of work in one's own topic area, defined narrowly, to clearly communicate how one's own methods are similar/different to others' methods.

Even in group 1, when I go back to a project that I haven't worked on in years, it would be helpful to be able to query the build system to list the dependencies of a particular artifact, including data dependencies. I.e., reverse dependency lookup. Also list which files could change as a consequence of changing another artifact. And return results based on what the build actually did, not just the rules as specified. I think make can't do this because it has no ability to hash & cache results. Newer build systems like Bazel, Please, and Pants should be able to do this but I haven't used them much yet.

Patents sort of work that way, except that even people who didn't look at your work owe you their future profits.

I think I'm hoping for a result that anyone can train any model on any content, regardless of that content's copyright status. Mostly because I want AI assistant tools to be as effective as possible, to be able to access the same information I can access. But however it turns out there will probably be some unintended consequences.

The problem with gold OA is the proliferation of low-quality spam, and the recommended remedy is to both restore exclusivity (raise the barrier to entry) without supporting excessive (monopolistic, exploitative) journal subscription fees. The article's recommendation to build up society journals has a decent chance of accomplishing this. It won't prevent spam from getting published, but the only essential aspect is to keep the spam easily identifiable (e.g., published by MDPI) so it can be safely ignored. The www seems to facilitate the spam business model so I don't think the spam will go away entirely.

> So IMO, the first reform to conduct is to turn clinical researchers into real scientists. That won't be easy.

Mostly, there are not enough hours in a life to be both a good doctor and a good scientist. The few MD-PhDs who give equal weight to both areas are both brilliant and extremely driven. Multidisciplinary teams seem to have a better chance at success.

I guess if the scientific education was done early (undergraduate at the latest) dual training could work. Once medical training and practice starts there isn't a lot of time left over. And early / non-practicing science education, for whatever reason, doesn't seem to be very effective.

It often is arbitrary. Also, the rules as enforced by the awardee institution often differ from the funding agency's written rules. The people who are responsible for reviewing purchasing are only responsible for making sure no rules are broken, not with producing anything, so they tend to interpret the rules conservatively and just say "no" if they're not sure. The IT department also may try to block anything they didn't pick even if the funding agency would allow it. (Speaking from experience with university bureaucracy.)

It's not out of concern for cost-benefit ratio (overhead from parts selection and assembly) or anything like that. Getting something useful done is solely the researcher's problem.

Yes, the general spirit is good but the clinical trial stage is the very end of a long process. The clinical trial the most visible part of getting a new cancer treatment to patients, but the bulk of delay in developing a new treatment is waiting for the basic science. More aggressive clinical trials could save ~ years. More aggressive education and fundamental research could save ~ decades. Can't run a trial on a treatment that doesn't exist yet.

https://en.wikipedia.org/wiki/Amdahl%27s_law

I also have XM4's and they worked fine on Arch after addressing two problems:

Do you dual boot? Different OS's on the same computer will generate different pairing keys even though they share the same MAC, and this will cause connection issues. Usually that's reported as having to re-pair every time you switch OS's though.

https://unix.stackexchange.com/questions/255509/bluetooth-pa...

I've also experienced audio skipping & popping using a dual WiFi/Bluetooth card that were eliminated by disabling WiFi. Apparently the Linux driver was faulty and allowed some interference; the card worked fine on Windows.

> If we already know what a baseline untreated group looks like

There isn't really a single baseline untreated group. For a comparison between groups to be valid, all groups must be obtained by unbiased random sampling of the same population. In a clinical trial, that population is the patients served by the participating clinical center. Patient characteristics differ by time and place.

You can try to retrospectively construct a control group using a case control study design, but then you're getting to pick what control group to use, so the results are less reliable (more opportunity for human bias).

Unless a treatment is both miraculous in effect and works for everyone, it's hard to figure out if it works.

> A basic problem I have, looking at an Excel spreadsheet, is I don't know which cells are calculated by a formula, which are constants.

Use Ctrl-` (show formulas).

> How does version control work with Labview?

Labview does have diff and merge tools. It feels kind of clunky in practice, kind of like diffing/merging MS Office files. In my experience people think of versions of LabView code as immutable snapshots along a linear timeline and don't really expect to have merge commits. Code versions may as well be stored as separate folders with revision numbers. The mindset is more hardware-centric; e.g., when rewiring a physical data acquisition system, reverting a change just means doing the work over again differently. So LabView's deficiencies in version control don't stand out as much as they would in pure software development.

https://www.ni.com/docs/en-US/bundle/labview/page/comparing-...

Grant applications are primarily judged on the perceived strength of the research plan. The reviewers work in the same field as the applicant and are expected to judge the proposal on its merits. Citation count is therefore already not a major factor in whether a grant is awarded or not, at least in engineering.

Citation counts are considered by promotion committees though. In this case, committee members are often from different fields and have less ability to judge the quality of the applicants' work for themselves.

None of the above matters in this particular case though—the problem is that Corchado was the only candidate on the ballot: "… taking advantage of the strange surprise resignation of the previous rector, and presenting himself as the only candidate. He received the support of 6.5% of the 33,000 university members who were called to vote, with half of the faculty voting blank as a sign of protest. "

I just tested saving https://browse.arxiv.org/html/2312.12451v1 to disk using Chrome, transferring it to my Android phone, and opening it on the phone. Results:

1. Saving as "Webpage, Single File" (.mhtml): Neither Firefox nor Chrome even showed up in the list of available apps to open it.

2. Saving as "Webpage, Complete": Opened in Chrome but images were broken. Also very difficult to open with the default file browser because it uses a flat folder view and the sidecar folder pollutes the file list.

I was hoping this would work, perhaps you will have different findings. I agree that HTML is the superior format in theory but usability in practice is often lacking. I'm resigned to using both depending on context.

Across species, tendon strength is strongly correlated with modulus (normalized stiffness). Greater stiffness has the risk of reducing range of motion. When a graft is used it is intended/hoped that the graft will remodel over time to become more similar to the original tendon's properties, which match what the entire organ system needs to function properly. Nor does a stronger graft mean that the connection between the graft and the bone will be 6× stronger. 6× stronger might also be optimistic; prior work suggests more like 2× stronger and stiffer.

https://pubmed.ncbi.nlm.nih.gov/23599401/

I realized why this being based on simulation bothered me: this is a machine learning classifier that classifies viral genomes as synthetic or natural. The training set n = 72 (all negative, which is justifiable if you're ok with null hypothesis significance testing) the validation set n = 6 (only synthetic examples, which is less fine), and there's no test set. No effort was made to estimate true positive rate, false positive rate, etc. If this was published as a machine learning paper instead of a biology paper it would probably be held to a higher standard.

> The GP comment is sort of misleading...you wouldn't just pick enzymes at random to do this analysis.

I said pick blind, not at random. I recommend reading https://info.umkc.edu/drbanderson/p-hacking-and-the-problem-...

"To avoid losing power with multiple comparisons, we focus our analysis on the BsaI/BsmBI sites in SARS-CoV-2 and compare the BsaI/BsmBI map in SARS-CoV-2 to all other restriction maps of all other CoVs used in our analysis."

It's important to know whether or not the authors picked BsaI & BmBI blind, before looking at the genome. If they picked BsaI & BmBI with knowledge of the SARS-CoV-2 genome, that doesn't dodge the multiple comparisons problem and the p-values aren't reliable. I guess it depends on how many other commonly used type IIS endonucleases there are. The authors use 214 to generate their null hypothesis distribution for the CoV restriction maps but say only 6 are specifically amenable to BAC cloning.

The "wild type distribution" null distribution for fragment length (Figure 3C) being a simulation (permutation of known CoV genomes, split at randomly selected restriction sites) bothers me. On the first read I thought it was a distribution of fragment lengths in real viruses. Does synthesizing virtual genomes by permutation produce a realistic distribution of fragment lengths?

Getting everyone to agree what "gain of function" means is a lost cause.

"The term GOF didn’t have much to do with virology until the past decade. Then, the ferret influenza studies came along. In trying to advise the federal government on the nature of such research, the US National Science Advisory Board for Biosecurity (NSABB) borrowed the term … From that usage, it came to mean any research that improves a pathogen’s abilities to cause disease or spread from host to host."

https://www.nature.com/articles/d41586-021-02903-x

The actual regulations sensibly avoid the term, targeting "enhanced potentially pandemic pathogens".

https://www.nih.gov/news-events/research-involving-potential...

Your article actually argues in favor of the type of research BU was doing because it's being done post-outbreak: "Once an emerging outbreak virus has been identified, it needs to be analysed quickly to establish what type it is; which molecular mechanisms (such as receptor type) enable it to jump between individuals; how it spreads through human populations; and how it affects those infected In other words, at least four kinds of analysis are needed: genomic, virological, epidemiological and clinical.". BU's work is in the genomic and virological categories and addresses the kind of questions the article wants addressed as part of a surveillance strategy.

The article strongly opposes making predictions of whether a pathogen will be pandemic because predictions are a rigged game. There are enough factors that you are very unlikely to be correct, and even then a true positive combined with successful mitigations erodes public trust because "the severity of the virus had been overblown". Geologists avoid making predictions of volcanic eruptions or earthquakes for similar reasons.

It is not directly opposed to study of pre-pandemic animal pathogens: "Surveys of animals will undoubtedly result in the discovery of many thousands of new viruses. These data will benefit studies of diversity and evolution, and could tell us whether and why some pathogens might jump species boundaries more frequently than others." This makes sense, because study of pre-pandemic pathogens is needed to set a foundation such that the post-outbreak analysis the article wants us to focus on can be completed in a reasonable time frame.

The article is arguing for rapid response and against predictions. It's not opposed to basic research of pathogens, and actually supports this specific type of research.

> Nope, it only happens when something outside changed Firefox's files … This never happens* in Windows and macOS (it might nag, but you can definitely dismiss it).

Happens with multiple Firefox instances too, `firefox -P -no-remote` (I think in recent versions the -no-remote is redundant), even on Windows with official installers. In that case you don't even get the error message; new tabs just remain blank. There might be a delay between process A's update and symptoms appearing in process B; not sure.

Regarding the waiver—"The maximum discount is 15 percent of total monthly spending on AWS services". Was very excited at first.

As for leaving data in AWS, data is often (not always) revisited repeatedly for years after the fact. If new questions are raised about the results it's often much easier to check the output than rerun the analysis. And cloud storage is not cheap. But yes it sometimes makes sense to egress only summary statistics and discard the raw data.

> the ability to reserve the server on specific days

In an environment where there are not too many users and everyone is cooperative, using Google Calendar to reserve time slots works very well and is very low maintenance. Technical restrictions are needed only when the users can't be trusted to stay out of each other's way.

It seems like people in the US still agree that free speech is a good thing. 91% of Americans say that free speech is very or extremely important to them, and the rest say it's moderately important or just skipped the question (like below). No significant fraction of the respondents were against free speech.

When split by subgroups, a common theme is that each group believes _other_ people have an easier time exercising free speech. So maybe there's more perception of a change than a real change.

https://knightfoundation.org/reports/free-expression-in-amer...

Edit: Reading more, I guess there is some change in attitude, using respondent age as a proxy for time. Teens prefer feeling safe online over being able to speak their minds online, moreso than adults. https://www.pewresearch.org/fact-tank/2022/08/30/more-so-tha...

Yeah, changing the details is fine. Tolkien wrote in his letters, "I would draw some of the great tales in fullness, and leave many only placed in the scheme, and sketched. The cycles should be linked to a majestic whole, and yet leave scope for other minds and hands, wielding paint and music and drama." We're getting work from other minds and hands now. It's normal for a tale to change in the retelling. Given Tolkien never stopped revising and retconning his own work—the Lord of the Rings itself starts with a major retcon to the Hobbit—there's no single source of truth anyway. Points of difference are often interesting to discuss, but their mere existence is not automatically a bad thing.

> Science bodies (like the NIH) explicitly forbid reuse or reinterpretation of data.

I think your main point is that collecting new data is necessary to test existing ideas. But reuse and reinterpretation of data is routine, e.g., in meta-analyses. It's not forbidden. You do have to disclose where the data came from.

I agree that increasing public research funding is a good idea, but it can only fix the payline & hype problem (increase efficiency) temporarily, until universities once again expand their departments. For example, the NIH budget increased by ~ 2.5× between 98 and 03, which prompted many universities to form new biomedical departments. In a way, that's exactly the intended effect. The point of increasing funding was to do more science, and to a degree more researchers = more science. But hiring out-competed the funding increase and ultimately created the problems highlighted in the posted article. The capacity to consume money can always grow faster than funding.

On Hacker News we often talk about bad incentives in science from an individual's perspective, but the main incentive problem is at the organizational level. From an organizational perspective, every additional grant-seeking researcher is still a marginal benefit, no matter how little time they actually spend on research. E.g., even a 99% marketing 1% research split (an exaggerated ratio, only for illustrative purposes) means adding another "researcher" is still a marginal benefit to the organization even though it wastes most of public's money on institutional growth rather than science. The equilibrium state is a near-total focus on grant-seeking and marketing. To ensure the public's money is spent effectively, funding agencies need to pay more attention to designing their policies to counteract this tendency.

You might get your wish regarding the NSF specifically though. An 18% increase will make a difference for a couple years: https://nsf.gov/about/budget/fy2023/index.jsp

Reduce the number of PIs until grant paylines go back to ~ 30%. Institutions that don't play ball—that hire grant-seeking PIs at a greater than sustainable rate—should be penalized. First, by reduction of indirect costs, then by making applicants affiliated with that institution progressively ineligible for various funding mechanisms. Once the applicant pool is back in balance with the money pool, there can be more focus on science and less on salesmanship.

Sounds like journal clubs but organized through a specialized tech platform rather than normal communication channels.

(Journal clubs are groups of people, usually working in the same field, who pick & present a paper to the rest of the group on an rotating basis.)

Ah, so the social science reproducibility problems are hypothesized to be caused by bias (or, less provocatively, filtering), moreso than e.g. bad experimental design. Yes, that would nicely reconcile both perspectives.

Are they really? If social science is unreliable, then it is reasonable to conclude that (a) limiting its scope of investigation doesn't matter or even (b) limiting its scope of investigation is actually good, so that it at least does less harm. In what perspective are the issues fully orthogonal?