fastaguy88

PG is an excellent writer, but this essay seems remarkably misleading. The unstated premise seems to be that well-educated adults already know everything they want/need to know about everything, which is silly. I'm older than PG, and pretty well educated, but I am constantly learning new things. I don't think it's because they are not important or I am obtuse. I think it is because I am (still) intellectually curious.

Sometimes I learn new things because they are new. And sometimes I learn new things (that are well known to people in other fields) because while I know a lot about some things, I know very little about others -- so little that I don't even know those things overlap with my interests.

Those of us who enjoy learning appreciate that we will never know everything we would like to, and in fact we will never know the boundaries of knowledge for topics we care a lot about. It's not that it is unimportant to us, it's just that we hadn't learned about it yet. That's why we read essays.

In the protein annotation world, which is largely driven by inferring common ancestry between a protein of unknown function and one of known function, common error thresholds range from FDR of 0.001 to 10^-6. Even a 1% error rate would be considered abysmal. This is in part because it is trivial to get 95% accuracy in prediction; the challenging problem is to get some large fraction of the non-trivial 5% correct.

"Acceptable" thresholds are problem specific. For AI to make a meaningful contribution to protein function prediction, it must do substantially better than current methods, not just better than some arbitrary threshold.

Not a lawyer, but there are a lot of crimes that are not felonies. Speeding 10 mph above the limit in a 65 mph zone - not a felony. Reading hacker news for an hour during work time and not being paid $800/hr - not a felony. Calling in sick when you are hung over - not a felony. There is no federal tax on gifts for the giftee. Indeed, I suspect there are a surprising number of crimes that could get you jail time that are not felonies. Insider trading - it’s a felony, which is why people in companies with insider trading information are told they cannot trade at certain times.

I’m pretty comfortable believing I have probably not committed more than two or three felonies in my life. (Don’t want to find out I am wrong.)

Let's see: C elegans -- the worm no computer scientist can crack S. cerevisiae -- the yeast no computer scientist can crack E. coli -- the bacterium no computer scientist can crack HIV -- the virus no computer scientist can crack

Has a computer scientist cracked any complex system that was not engineered?

If you read the Final Scientific Integrity Policy, included towards the bottom is the statement:

"and under “Protecting Scientific Processes,” a statement noting that early termination of extramural awards is prohibited except under certain specific circumstances."

Clearly, this is not a policy that the current administration commits to.

Your concept of designed is very different from mine. The iPad is capable of providing a shell interface, but it is clearly not designed to. It is designed to provide a secure media consumption experience. There is nothing arbitrary (from a mass consumer security perspective) about not providing a shell. Providing a shell makes it much easier for bad actors to dupe unsophisticated users.

I am very puzzled. Apple has locked you in a cage because you bought an iPad to replace a MacBook? What is the cage, and why weren’t you the jailer?

I think it is worth pointing out that Bell Labs was an Engineering research lab. Scientists there created new disciplines, but in the service of relatively well defined engineering goals.

Contrast this with the NIH, where the science also has a goal - improving human health - but the system to be improved was not engineered. Curing a disease, which has a natural origin, is quite different from improving communications channel capacity.

I suspect that managing engineering research is much more amenable to process analysis than research on biological systems.

> Every University’s purported mission is to educate students and advance our > collective knowledge together with its students.

> That’s it.

Not if the university has a medical school. Virtually all R1 universities with medical schools have a hospital, and a large clinical practice. Most of medical school is an apprenticeship where you treat patients. Medical schools need patients, which means a lot of additional staff.

Likewise, in most fields it is no longer possible to advance knowledge just by going to the library or writing on a white board. Knowledge is advanced through experimentation, and experimental equipment and reagents cost money, and need staff to use and maintain them.

No university (and certainly no medical school), makes enough money in tuition and fees to pay for the education provided, and I seriously doubt that many universities have supported themselves solely through tuition since the beginning of the universities in the middle ages.

You are certainly correct that university deans and presidents have seen their mission shift with the increasing cost of education, and indeed faculty are writing many more grants than they did 75 years ago. So time commitments have shifted. But there is an implication that it could have been some other way -- that the money is there (or could have been there) if some other path were chosen. It is hard for me to imagine where the money might have come from.

It might be insane, if you believe that "staff" are all doing administrative duties. But, as was pointed out, "staff" are often anyone who is not a tenure track faculty. So librarians, research technicians, environmental health and safety, IT support, etc etc.

A more useful comparison would divide staff into "supported by tuition" (should be related to student count) and "supported by external grants and clinical income".

This idea that costs have increased because of administrative staff expansion is a popular one, but one that ignores what R1 universities spend money on, and where that money comes from. (Ironically, I suspect that the university may be spending more money on research, because of limits on indirect costs.)

I’m thinking you never took a logic class. Is there only one reason why teachers quit the profession? Do problems in one location mean there aren’t other problems elsewhere? I suspect not, which is why your comment makes no sense.

Non-scientists get excited and engaged by participating inthe scientific process. Including undergraduates on a paper annotating Drosophila genes is head and shoulders more exciting than rediscovering some genetic principal published 70 years ago. The purpose of outreach is engagement, not simply checking off the outreach box.

It is hard for me to understand how proposing some minor outreach goal (that would consume <5% of the budget) politicizes the science. Perhaps the outreach was politicized, but outreach is not science.

>IRBs are only REALLY required when you are intervening in patient care or pose some theoretical risk to a human.

Theoretical risk is a very broad category. IRB approval is required to download de-identified human genome sequences, in part because the genome sequence can be used to identify the individual. Drawing blood from an individual, or using blood that has been drawn for other purposes, requires IRB approval. And IRB approval is required for many sociology studies (or surveys), in part because publishing an individuals answers to a survey might put the subject at risk.

So the "theoretical risk" threshold can be very low. It's theoretical after all, so the harm need only be imaginable.

Early studies (1970's and earlier) used the antigen (hapten) di-nitro-phenol (DNP) to explore antibody diversity (how many different antibodies might be raised against a simple small molecule). DNP is very simple (non-chiral) structure that generates a diverse set of antibodies.

Whether an antibody raised against one structure interacts with a different structure is irrelevant to the question of whether an antibody can be raised to the other structure.

The claim that mirror molecules would not be subject to immune surveillance makes no sense to me. The immune system is happy to react to non-biological molecules. It reacts to shapes, not chirality. It is a separate question whether the immune response could break into mirror membranes and break down mirror molecules.

A meta comment: This paper gives an example of a "teaser abstract". It says what was done, but does not say anything about the actual results. This style is relatively common, but I find it very annoying. There was certainly enough room in the abstract to provide a concise summary of the actual results, which would both inform the reader and perhaps encourage more people to read the entire paper.

I'm sorry, but I could not get beyond the first sentence in the abstract: "Living cells are extremely well-organized autonomous systems, consisting of discrete interacting components." Living cells (and most other biological systems) are extremely poorly-organized. They work. Often very inefficiently, with lots of duplication. Any "organization" is purely accidental (in some sense, by definition, since it arose through evolution).

Reading a bit farther (RNA is lists, DNA doubly-linked lists) is embarrassing. DNA is double stranded, which does not make a doubly-linked list. And somehow we fail to recognize the difference between template-driven molecules (DNA,RNA,proteins), which have a genetic history, and lipids and carbohydrates, which do not.

The idea that biologists have difficulty with abstractions probably reflects the “solidity” of modern molecular biology. Biologists were doing genetics, working out the rules of heredity, mapping genes onto chromosomes, understanding meiotic cross-over, for decades before the physical nature of the gene was understood. Likewise immunology is full of terms and factors that today have a biochemical explanation, but again, for years only existed as abstractions. Biology is much easier to understand today, now that we have molecules for almost everything. But biologists did decades of very insightful and productive work that was completely based on abstractions.

I am a bit more comfortable saying "Biological systems also have 'solutions' that emerged through evolution." That is certainly true. But unlike designers, evolution is perfectly happy to re-invent the wheel (even if it is a less functional wheel). So different, evolutionarily independent, processes may provide the same solution, and of course solutions are constantly re-used to provide slightly different solutions. So I'm not sure that "complexity" is the hardest problem, though it certainly doesn't make things easy. The diversity of solutions for the same problem makes generalization/abstraction even more difficult.

I am always puzzled when biologists make analogies between living systems (or sub-systems) and human-designed devices (even complex devices). Human designed devices have an essential property that biological systems do not -- they were designed. While radios may seem complex, they are built from a relatively small number of component types with very well understood behaviors and typically limited numbers of interactions. Biological systems have no such constraints -- they were not designed, there can be many different ways to do the same thing (in the same cell), there are at least thousands, if not hundreds of thousands of different components, many of which have lots and lots of interactions. Considering how messy the systems are, it is quite remarkable that genetics and biochemistry discovered the basis of heredity (or at least some of it), the genetic code, hundreds of signaling pathways, etc etc. But there is no missing language, because there was no design that used that language.

One of the major breakthroughs in Bioinformatics was the recognition that local similarity scores (which can be thought of as runs of positive sequence similarity) are extreme-value distributed.[0] The logic of that discovery uses almost exactly the same mathematical argument as this paper [1], indeed I recognized some of the same equations.

It is difficult to overstate the importance of this discovery for biology, as today, the vast vast majority of protein functional inferences for newly sequenced genomes are based on the statistics of long runs of sequence similarity.

[0] https://www.ncbi.nlm.nih.gov/BLAST/tutorial/Altschul-1.html [1] https://www.pnas.org/doi/epdf/10.1073/pnas.87.6.2264

Just a minor correction. Most of the proteins encoded by the human genome are NOT enzymes. Only about 17% of the ~20,000 human proteins have known enzyme activities. Many proteins (probably the majority of protein mass in the cell) are structural, and others do signaling and metabolite transport.

If we are not doing Intel/ARM architectures, a lot can be said for learning PDP11 assembly. Among other things, is maps very cleanly into concepts central to higher level programming languages, like indirection.

One strategy for better pictures is not to worry so much about what the camera points at (or where the focus should be in the frame); instead focus on the edges of the picture. By shifting your focus from the "focus" to everything else, you end up with pictures that contain a lot more of what you want.

Absolutely. But DMD has a spectrum of severities, and it seems possible that for some patients, producing some normal transcript would make a real difference, so that until more people are treated, we cannot know when more normal transcripts make a difference.

The tricky business is defining effectiveness (and the target population). This drug was not effective at the specified clinical end-point for one set of patients. But it did show promise for other end-points (production of a normal dystrophin transcript).

For complex diseases with a range of severities, failing the specified clinical endpoint is not the same as selling dyed water.

Clearly written by someone who has never submitted an academic paper for publication. Latex (despite its many issues) does sevealthings well: (1) bibliographic references in many formats (2) internal references to figures and tables (3) figure placement (4) mathematical equations and special characters. Plus, many journals provide latex templates that correctly format your paper. I recently tried to move from Latex to Word because my co-authors preferred word, but getting the biographical references correct was a lot more pointless work. Almost all the arguments about layout the author make are irrelevant if you cite references and need to resubmit to a second (or third) journal.

Under-funded defined benefit pensions may be a Ponzi scheme. But properly funded (X% of salary per month, calculated to cover future costs) defined benefit pensions are significantly cheaper for a given level of benefits, because they only pay out only until the beneficiaries die. Employee owned accounts (IRAs, 503Bs) are passed on to the next generation. Good for the kids, but not for the parents or the company.

Curious -- how do we get information about the thickness/shape of the galactic plane from observations within the plane?