tsbischof

For the article on Euler's conjecture, I am aware of this paper and that it serves as a sufficient proof of work for publication. It includes context for its purpose by citation, and the methods for verification were well-established. There is a class of literature where this type of structure works.

For the Kekule paper [1] there is a significant amount of information about the context and reasoning for the claim. This is not an isolated concept and he wrote at length as to why the idea might be plausible given the current evidence. He also could have written solely about the dream without context, but that lacks a grounding in the reality he was attempting to describe.

If it is possible to write a paper where the result is possible to verify using already-known methods, then by all means write in that style. But this is a subset of the useful papers to be written, and in my experience a small one.

[1] https://gallica.bnf.fr/ark:/12148/bpt6k281952v/f102.item

What do you see as the fundamental point of scientific communication? In your counterpoints you narrow in on papers being a means of communicating concepts or proof of work. In this view, showing the process itself is pointless or at least irrelevant to the main axiom.

However, others (myself included) see the the communication of methods as a primary function of the literature, because this is what enables others to understand, critique, and build upon the idea.

Solving this problem means providing a workable solution to the people in power, which in this case is homeowners. Can you find a solution which lets them keep their financial gains but also shifts the status quo? If not, they will fight you and win.

One wonky approach might be to introduce some form of government seizure over time, like a scaling progressive rule based on the quantity of property owned that is actually a forced sale to the local housing authority. So over a period the extra housing equity would be owned by the authority and redistributed, while the existing homeowner gets cash but not the property. This particular concept feels flawed but something which produces wins all around is easier to get moving.

As for building the movement, a crisp rallying cry and lines of communication to local stakeholders is key. You probably need to search around for the location with the right mixture of local activism and receptive leadership, in the same way that many startups benefit from specific locations.

The governing body and related stakeholders care at varying levels. For government-sponsored research, the primary metrics are publications, impact factor, number and prestige of collaborations, diversity of staff, and velocity of money (see [1] for an example, pages 3-4). At the most cynical level it kind of reduces to welfare for educated people. The end result is that there are researchers producing good work, but you have to filter to find them.

For external industry funders, they have a strong and vested interest in identifying good output before it is public. The strategy here typically is a mixture of identifying good researchers and making them kings (consistent funding, direct lines of communication, access to nonpublic materials, etc), and of placing lots of small investments to get in the room for discussion. An example of this mixture is the BP deal at UC Berkeley [2].

As for shifting standard of care, this is a commonly-expressed goal but often the research is far enough away that the individuals doing the work lack close communication with the actual practitioners. Some institutes are trying to resolve this by close physical and social proximity (e.g. TU Dresden's medical campus) but this is far from the norm.

[1] https://www.helmholtz-muenchen.de/fileadmin/Jahresbericht/Ja... [2] https://www.berkeley.edu/news/media/releases/2007/02/01_ebi....

The institution as a whole may have the incentive, which plays out on a 5-10 year timescale. Funding agencies in principle care, but no individuals are rewarded based on the results on this timescale. Industry, venture capital, and the like do care because they need reliable-enough results to build on, but they can filter for the good ones and scale from there.

The lifecycle for a junior PI is about 5 years, which really means they have about 3 years to produce a result sufficient to advance. This is enough for a bit of validation but unlikely anything beyond their lab, so the result is likely to be untested. But who in that lab cares? They just need to get the next job, so a 3-4 year timescale is fine. The PI cares long-term because a lack of new productivity means a lack of ongoing funding, which means they either improve the quality of their processes (which means herding the junior researchers through that system) or simply continuing to hack their way through.

There is a reason why people moan about scientific theories advancing one death of an old professor at a time.

The equipment can be affordable depending on how you structure things. For example, there are opportunities like the Molecular Foundry user program [1] which provide grants for the costs of running experiments on expensive machinery. The main trick is having a specific enough question to answer that you can define how the grant would help you do so.

There are also community initiatives like BosLab [2] which are more similar to maker spaces in structure.

[1] https://foundry.lbl.gov/user-program/get-started-here/

[2] https://www.boslab.org/

I would be glad to have funding explicitly set towards appropriate and open documentation of the actual protocols, equipment, and materials, plus a combined effort to enhance these reporting requirements by the journal. There has to be a balance between the one-off interesting new findings and the large-scale work, but clinical trials definitely fall under the latter category.

Part of this is being addressed through increased reporting requirements for data, software, analysis pipelines, pre-registered protocols, and the like. This has led anecedotally to some improvements in quality, and at minimum it makes initial due diligence on the publication significantly more impactful.

It is worse than nothing, because incorrect positive results and non-published negative results lead people to waste years of their lives building off something which is not true.

> What can help make the literature more accessible?

Publishing code and data associated with figures. A significant amount of the confusion about the literature comes from the difference between the documentation laid out in a paper and the reality of the actual implementation. Plus any peer review of the actual code used is checking the reality of the paper, rather than a description of what we think reality is.

Making this part of the publication process is vital, because there is such a high barrier to actually requesting this information later.

I have been on both sides of this.

I worked for a time at a startup with decent salary in a generalist role, such that sometimes I did things directly related to my skillsets and interests, and other times got to pick up entirely new skillsets. This broadened my horizons significantly and put me in a decent position financially, though nowhere near FIRE.

After that I took a pay cut to pursue something fun and interesting (moving to a new country to help an old colleague establish a research group). This gave me a chance to work on my own projects and passions but also pulled me out of the fast-paced and wide-reaching environment that you find in a startup. The freedom is great but the cash compensation hurts a bit, so the rest of the package has to make up for it -- most days it does.

All things considered, you can find interesting things to do (and people to do them with) in most positions. Taking one with higher compensation and comparable commitment of time and energy is rarely a mistake. Especially if that compensation makes you financially independent with reasonable probability.

The challenge is that there are two directions for the knowledge retrieval: top-down and bottom-up. The "problem" here is that failure to learn from past, well-characterized solutions means we will be forced to revisit problems experimentally (expensive) that could be solved by examining the documentation (relatively cheap). Literate programming and inline documentation is great for understanding the bottom-up picture, because you colocate the components that are needed for understanding. For top-down, you are often summarizing several projects or proposals and trying to explain trajectory and architecture, and necessarily will not always have an obvious "best" location for retrieval.

For example, consider an assembly line in a factory. Having work instructions and debugging manuals next to a station is great, but where should we place the information explaining why this station must be configured in particular way to prevent an issue six steps downstream? Do both stations now need to have that information documented? Do we have higher-level documentation that needs to be synchronized? How would we find information about the changes made while debugging that process which ultimately did not make the cut, but would be useful to know as negative experiments?

I would be curious to learn more about the tool you work on. I have never found a good solution for architecting layered information for retrievablility, only least-worst hacks.

Pick a reputable company like Vanguard and sift through their categories of funds (index, sector, bonds, REITs, etc). Each will come with its own risk profile and past returns, so from there filter down into an investment thesis. Think Silicon Valley will keep doing well? Then consider a tech-heavy fund. Want to cash in on real estate plays? Then look into the holdings of their REITs.

Once you do that, find equivalent funds across the various companies, and pick the one which has the lowest fees.

For increased risk, put heavy allocation into a single sector. For reduced risk, pick multiple asset classes (e.g. bonds and international). For the most conventional options, look at the composition of target date funds, which balance risk geographically and temporally.

Once you have a picture for what funds you want, use a screener to explore from there. Iterate on the thesis as needed

It depends on your relationship to work, relative to your outside life. The PhD in a good institution is essentially semi-supervised playing, where you are given access to resources and people to come up with and answer interesting questions. The probability of the utility and monetary value of the knowledge varies across fields significantly, and so the main point is to learn how to learn and how to identify and answer questions.

This style of PhD is a lifestyle choice and a vow of "poverty". For context, at a school like MIT or Berkeley a typical fully-funded stipend is about $30k USD annually (gross), which you will make until you finish (at age 27-30, depending on when you start). Once you run through the cost of living in the area, you start to see that some compromises must be made to your quality of life. If you have independent money these compromises are less necessary, but if you are not careful it can alienate you from peers, in the same way that having money generally means you live in a different world than others.

If your goal in a career is to make enough money or have enough stability to support your actual life (family, community, hobbies, etc), doing a PhD is not a particularly reliable path. Having one opens doors in life, but you have to value that over the alternative for the decision to make sense.

(I have a chemistry PhD)

It could be as simple as checking tax records: in Germany, tithing is incorporated directly into tax declarations and payroll systems.

In practice it generally comes down to demonstratable utility more than direct concerns about animal well-being. At the most basic level, if you can learn a useful lesson from subjecting an animal to whatever conditions, then the work is generally approvable. But if the researcher has not done their due diligence in experimental design then the whole operation is pointless, or worse than pointless if they then go and p-hack their way through a publication.

For a physics analogue, think of animal experiments like a beamline proposal. In order to get time on the beamline you need to explain what you are doing, what your experiment is intended to accomplish, and why the experiment achieves those goals. With these basic checks you can hopefully ensure that the allocated time is not obviously going to be wasted. Animal experiments can easily be about as expensive in terms of money and time as a beamline-based project, including the cost for the beamline operation itself.

You do not cut out animal research but you push for a clearer path to safely start research in humans. Instead of repeated and overlapping pre-clinical studies, focus on designing studies to move as rapidly into humans as possible while retaining a reasonable risk profile. The longer people do work on animals the more they optimize for and are invested in the animal solution.

Much of the problem comes from the fact that there is insufficient incentive for the studies to be effective in the first place. The sample size issue is real and often justified due to a lack of resources, but this results in minimum publishable units which are anecdotal and an over-reliance on meta-studies which attempt to aggregate these positively-biased results. Increasing the sample size is not always possible, but sometimes there are alternative trial designs which can yield different but more reliable outcomes.

Another aspect of effectiveness comes from whether the animal studies are actually designed to advance anything at all. For example, if the goal is to develop an imaging technique in humans, then the purpose of animal experiments is primarily to confirm that the idea is not wrong, debug the machinery and protocols, and generally get things to the point that the system is no longer the point of failure when running experiments on humans. Instead it is common to see series of publications on new imaging methods and contrast agents which grind away at small animals with no visible attempts to progress upward, which pushes back the time at which you can actually learn how the system works. Publishing this way is far safer but ultimately pointless: drugs for Alzheimer's is a classic case of optimizing for animals and completely missing in humans.

This is at least partly why pre-registration of experiments is becoming normalized, as it is in human clinical trials. The goal here is to require basic statements of purpose and methods prior to execution, such that it is possible to improve the (often quite expensive and time-consuming) experiments beforehand and not during the manuscript review process. Many research groups lack this internal quality control and as a result some institutions are stepping in to require or provide it.

Independent verification is not always performed for 510k submissions, but yes that would be ideal

I would be curious to see anything which lays out the specific claims and the evidence or justification used. For example, do you claim to detect malignant moles and if so what tests have you performed to demonstrate this. The language varies somewhat on your website so the approval documents or the information you described would be helpful to understand the device better.

Could you provide a link to evidence obtained using your own device? Or perhaps a description of the tests you performed to determine substantial equivalence to the predicate devices?

Can you clarify what you mean by certificates? Do you intend to link to approval documentation from a 510k clearance?

This is an important question, and something we have considered when building our group.

My background is largely in experimental physical sciences, and one of the vanity metrics in my fields is publication in a Nature- or Science-family journal, or equivalent high-impact journal. In some cases publishing there has led to high-quality peer review and publication, but the pursuit of these papers drives many to over-sell or over-complicate the work. I tend to focus more on smaller studies which can be reasonably completely described, versus anthology publication.

But more directly, we measure output in three categories, in order of importance: people, validated concepts, and publication/dissemination.

For people, we want to develop a sense of integrity and curiosity, which means some mixture of freedom to roam and an ability to focus on defining and answering a concrete question. We want people to leave our group as good citizens and scientists.

Validated concepts take several forms. In the physical sciences it is often expensive to recreate an experiment, so performing and reporting the correct set of controls and validations is critical. A successful experiment is one which does not need to be repeated, because there is sufficient clarity and design to assure the next person that what was found is correct. Often this means obtaining a proper negative result which shuts down entire lines of inquiry. Especially since many projects require months or years of build-up, we must ensure that the foundation is solid.

As for publication and dissemination, our goal is that someone else would look at the work and consider building on top of it. In order for someone else to accept the work there needs to be a real trust or means for validation. We are open with our data, code, and methods, but much of the challenge lies in ensuring the dataset is valid and complete in the first place.

You are right to note that popularity drives academia, in the same way that charisma and a good pitch can get a startup off the ground. High citation counts, big-name grants, fancy titles, and time spent at big-name institutions are the standard metrics for success. I come from a somewhat outside perspective since I spent some time in start-up prior to this position, where our focus was more on actionable outcomes than positive results. But for the students and postdocs, these pressures for advancement are real and not easy to reckon with. I have no good answer, apart from attempting to impart personal ethics and aligning interests where possible.

The longevity and source of funding is a key factor in fighting the popularity contest. For our institute this is largely solved by solid political capital, which gives us some cover to pursue things without significant concern about short-term popularity. Long-term we want to be known for producing useful research, which others incorporate into their own work and become more productive.

In short, a given researcher is successful if they have identified a self-contained question and answered it sufficiently to be a building block for further work, either as a positive or negative result.

I am helping lead the first research group at a recently-founded institute in Germany.

What is the mission of your institute? Is this a collection of individuals doing their own research, or a group of collaborators working towards a shared goal? How do your ideal members measure their own performance? Is it through papers, blog posts, tech demos, products, or something else? With academics you need to be especially vigilant about the difference between vanity metrics and real output.

For funding, are you allocating cash on an annual basis, up-front, on-demand, or through some other scheme? What sort of timeline do you want for people to stay at the institute? And do you intend the budgets to be competitive, either locally or internationally?

What lab infrastructure will you build and maintain, versus having the researchers handle? As a lab-based researcher, setting things up can easily consume the first 6-12 months (plumbing, electrical, HVAC, optics, wetlab equipment, regulations...). Once someone goes through the effort to build this they can easily become quite protective, potentially locking out newcomers or creating an internal system of patronage.

Having answers for these questions and others will be essential for recruitment. Your first few hires will dictate the direction of the institute, since further recruitment will be driven largely by their networks.

Feel free to email me (username at gmail) if you would like to discuss further.