I want to ask about the bureaucracy aspect. I have never written a science grant application, but expect that some of it comes about because the applications want to ensure good governance around the proposals. Do you agree? For the fluff that genuinely has no productive value, do you have any explanation for why it is there?
Could LLM participation be blowing holes in good-governance measures that were only weakly effective, and therefore a good thing in the long-term? Could the rise in the practice drive grants arrangements to better governance?
These are very good questions, and I only have vague answers because it's not easy to understand how bureaucratic systems come to be, grow and work (and not my speciality), but I'll try to do my best.
Indeed, some of the fluff is due to the first reason - for example, the data management plan (where you need to specify how you're going to handle data) has good intentions: it's there so that you explain how you will make your data findable, interoperable, etc. which is a legitimately positive thing; as opposed to e.g. never releasing the research software you produce and making your results unreproducible. But the result is still fluff: I (well, Gemini and I) wrote one last week, it's 6 pages, and what it says can be said in 2-3 lines: that we use a few standard data formats, we will publish all papers on arXiv and/or our institutional repository, software on GitHub, data on GitHub or data repositories, and all the relevant URLs or handles will be linked from the papers. That's pretty much all, but of course you have to put it into a document with various sections and all sorts of unnecessary detail. Why? I suppose in part due to requirements of some disciplines "leaking" into others (I can imagine for people who work with medical data, it's important to specify in fine detail how they're going to handle the data. But not for my projects where I never touch sensitive data at all). And in part due to the trend of bureaucracies to grow - someone adds something, and then it's difficult to remove it because "hey, what if for some project it's relevant?", etc.
Then there are things that are outright useless, like the Gantt chart. At least in my area (CS), you can't really Gantt chart what you're going to do in a 5-year project, because it's research. Any nontrivial research should be unexpected, so beyond the first year you don't know what you'll exactly be doing.
Why is that there? I suppose it can be a mix of several factors:
- Maybe again, spill from other disciplines: I suppose in some particular disciplines, a Gantt chart might be useful. Perhaps if you're a historian and you're going to spend one year at a given archive, another year at a second archive, etc... but in CS it's useless.
- Scientists who end up at bureaucratic roles are those that don't actually like doing science that much, so they tend to focus on the fluff rather than on actual research.
- Research is unpredictable but funding agencies want to believe they're funding something predictable. So they make you plan the project, and then write a final report on how everything turned just as planned (even if this requires contorting facts) to make them happy.
I came here to write - I think awk would fit in the list.
Awk is sold on pattern matching, and there are earlier technologies that do pattern-matching - ML, SNOBOL.
But awk's historic significance is something else: it was the embryonic scripting language. You could use it in an imperative manner, and in 1977 that showed a new path to interacting with a unix system. It allowed you to combine arithmetic, string manipulation, and limited forms of structured data processing in a single process without using a compiler.
Two language schools grew from imperative awk. (1) The scripting language that expose convenient access to filesystem and OS syscalls like perl/pike/python/ruby; (2) The tool control languages like tcl/lua/io.
It may also have influenced shell programming. Note that awk was released before the Bourne shell.
I would like to agree – I'm always surprised when I realise how old awk it. It feels like an incredibly modern language. It's also obvious that it inspired the likes of dtrace and bpftrace.
That said, I don't know how many other languages explicitly have cited awk as an inspiration, which was the criterion for this list.
My comment did not talk about where awk is useful today.
Unix gurus will recommend awk as a pattern matching and substitution tool.
But my comment was about awk the vanguard imperative scripting language. I don't know of anyone who recommends use of awk's imperative style over python in 2025.
As an exercise, I tried writing a simple roguelike in awk in an imperative style. Within twenty minutes, it felt obvious where perl came from.
You seem to have a chip on your shoulder about Christianity, and that's your right. But in the course of that you may overlook that faith-based treatment of problems is a powerful tool that serves some people well. Consider the culture of people taking a break from alcohol over Lent for several weeks each year.
It depends what you use it for. I worked on a interbank messaging platform that normalised everything into a series of standard xml formats, and then used xslt for representing data to the client. Common use case - we could rerender data to what a receiver’s risk system were expecting in config (not compiled code). You could have people trained in xslt doing that, they did not need to be more experienced developers. Fixes were fast. It was good for this. Another time i worked on a production pipeline for a publisher of education books. Again, data stored in normalised xml. Xslt is well suited to mangling in that scenario.
In the UK the government has somewhat gone after banks at a that level - money transfers. When someone falls for a scam and transfers money to a scammer, the government often makes the call that the bank is at fault for that. This is the absolute laziest thing the government could do, because it allocates all responsibility to the last line of defence rather than being an intelligent response to the problem.
This has led to the situation that doing a wire transfer regularly leads to intervention by the bank’s anti fraud team. This attitude has created a huge cost and risk overhead for all the banks, it forces inconvenience on consumers, and it hurts productivity of the economy.
A better way to combat fraud would be to drive improvements to the telephone network. Regulate to make the networks enforce accuracy of the phone numbers they are displaying, give the feature to reliably blacklist phone numbers, make the phone providers monitor for patterns of behaviour that look like scammer or mass marketing activity. There is no good reason that the phone companies should not have been expected to reach these standards decades ago. These fixes would assist with other law enforcement matters, such as tracing prank emergency services calls. But it requires meaningful work from policy makers, and it is not glamorous, so that never gets done.
UI concerns need to be in service to the full set of requirements and the data model.
UIs are easily accessible to end-users and product-managers, and can allow people to focus on a subset of the requirements. The trap is to allow the UI perspective to direct the development process.
It is vital to set an expectation with customers that allows discussion about UI matters as part of requirements discovery, but where they expect it to churn. During early development UI should be rough and should churn constantly in response to changes of more foundational matters: the business requirements, the data model, concurrency matters, interactions with other systems and the deployment.
I have in the region of 200 piano rolls in London. Classical emphasis, for example, lots of bach and beethoven. The piano player itself is long gone. The roll collection has been picked over, there will be no monetary high value items. If anyone wants the collection, we could coordinate, I am cratuki at the google service.
The Musical Museum in Brentford might take them off your hands. They have a large collection of fully operational and lovingly restored self-playing instruments.
Granted, domain or general knowledge is important in reading comprehension; I kinda know what a megalosaurus is but had no idea what Michaelmas Term was.
But, you can only gain that by having a broad interest and reading a lot in the first place. Call me grandpa but over the past 100 years or so of radio, TV, then internet, people have been reading less and less, so naturally reading comprehension has gone down too.
Flip it around and have an avid reader watch a modern gaming video and you'd see similar poor comprehension I suspect.
In the study, the subjects have access to both reference books and phones to allow them to look these things up. I didn't know Michaelmas Term either and looked it up while I was reading.
I am doubtful that this is about people not reading as hard and often. Alternate hypothesis: the problem is that people are not thinking as hard and often.
If someone gave this test to me, and I was volunteering my time and it didn’t matter to my grades/etc, no chance in hell I’m going to look up every single term I don’t know. I’m a busy person and I’ll just say “meh, it’s some old English thing that I’m sure will be explained through context in later paragraphs, and if I got through the whole first chapter of the book and it turned out to be an important word, I’d look it up. Next question please.”
According to the authors I’d be “functionally illiterate.”
> the problem is that people are not thinking as hard and often.
Modern language as understood by modern readers doesn’t require you to think as hard, and that’s a good thing. We shouldn’t strive to make language intentionally difficult just so we can feel smug and call those who don’t waste their time on it “illiterate.” Dickens’ prose is a sort of poetry and is more about creating a mental picture. The definition of “michaelmas term” and “lord chancellor” don’t seem like important parts of that paragraph at all. The important thing of the paragraph is for you to read it and get a picture in your minds eye about how dreary and muddy London was then. That’s all. Giving this to readers and saying “If you don’t choose to look up the word michaelmas, you are functionally illiterate” just shows that the authors need to look up what a fucking “study” is supposed to accomplish. Hint: it’s not about stoking the author’s ego.
There are complex reports that every European-regulated finance entity needs to submit to their regulator. They are always complicated, but they are only sometimes well-specified. The formats evolve over time.
There is a cottage industry of fintech firms that issue their clients with a generator for each of these reports. These generators will be (a) an excel template file and (b) an excel macro file.
The regulators are not technically sophisticated, but the federated technology solution allows each to own its regional turf, so this is the model rather than centralised systems.
If the regulator makes a mess of receiving one of your reports, they will probably suggest that you screwed up. But if you are using the same excel-generator as a lot of other firms, they will be getting the same feedback from other firms. If you did make a mistake, you can seek help from consulting firms who do not understand the underlying format, but know the excel templates.
There are people whose day-to-day work is updating and synchronising the sheets to internal documentation. It gets worse every year.
Sometimes the formats are defined as XBRL documents. Even then, in practice it is excel but one step removed. On the positive side - if you run a linux desktop you have decent odds to avoid these projects, due to the excel connection.
We use the word inheritance to refer to two concepts. There is implementation-inheritance. There is type-inheritance. These ideas are easily confused, which should be cause to have distinct words for them. Yet we don't. (Although Java does, effectively)
Could LLM participation be blowing holes in good-governance measures that were only weakly effective, and therefore a good thing in the long-term? Could the rise in the practice drive grants arrangements to better governance?