CipherThrowaway

It's easy to say "document what you can" when you leave "document", "what", and "can" loosely undefined. The article gives us a why and then leaves out the who, what, when, where and how.

If you observe people in their daily work, you will see there are many reasons why people do not document things. Time constraints, incentive structures, liability, general trade-offs and ambiguities. This is true not just in software/engineering but in any line of professional work.

There's plenty of grandstanding about the importance of documentation, but not much interest in understanding why people don't document in the first place.

So where is the "case for documentation"? The author provides no vision and no actionable recommendations. Documentation is not even mentioned until the end of the article.

Anyone can pat themselves on the back for pointing to the worlds problems and saying "this could be better!"

If you had to give a percent figure, what do you believe your mother's chances were of discovering a radical curative treatment for SCC and curing her own cancer?

I notice you've watered down your terminology from "cure" to "improvement on clinical best practice" which are pretty different things in the context of cancer treatment. However, I can respect the switch and I'll treat it as a sign you're starting to treat the subject a bit more seriously. It's more accurate to the outcome in the case study, too, since treatment was locoregional and neoadjuvant only. Full remission was only achieved with the traditional interventions of surgery and adjuvant targeted therapy.

You're in the awkward position of arguing that an expert in a field doesn't understand what she is doing while citing evidence to support yourself that you (by construction) don't have.

No. I'm in the non-awkward position of arguing that non-experts should be careful about interpreting a single case study without context. Especially in a way that implies miracle cancer cures are sitting around in labs with no one paying any attention to them.

I don't think the average HN reader understands just how many wildly different treatments, drugs and therapies are being thrown at different cancers and how quickly medical oncology moves as a field. Cancers are an extremely complex family of diseases. Early results and case studies are correspondingly extremely difficult to interpret due to the variation in individual responses and disease course.

The existence of a "miracle" cancer treatment is almost ruled out from first principles. But if such a miracle treatment is sitting around in a lab, it would be non-trivial to tell it apart from the thousands of other promising candidate therapies that go on to pan out to nothing.

Exhausting all treatment options is not the same as having nothing to close.

There's no shortage of cancer patients at end-of-life stage undergoing aggressive treatments and/or experimental therapies in clinical trials for minimal to no survival benefit. For almost all of these patients, the best option for them and their loved ones will end up having been a palliative or best supportive care model.

This virologist did believe that her colleagues were sitting around not rolling out something that would cure her.

And because her outcome was so unexpected and unusual it got published as a case study. What you don't see are all the cases where the experimental miracle cure treatment did not work. What you also won't see in headlines are all the trials where putative miracle cures and other promising treatments failed to demonstrate survival benefits in larger cohorts than 1.

One of the counterintuitive things about cancer is how badly individual cases and responses to treatment generalize to the broader patient population. If you didn't know any better, you could easily read a story like this and think "wow, this breast cancer cure was just stuck in a lab somewhere!" But to put a story like this into context, you need to understand just how many individual miracle remission stories there are, and how varied individual cancers and responses to treatment are.

There are potential miracle cures almost everywhere, and a large number of them are being aggressively researched, tested on cancer patients at any given time - often as part of combination therapies. Some of these promising technologies do become breakthrough cancer treatments that create durable remissions, such as checkpoint inhibitors. The rest fizzle out.

The development of PD-L1 expression was also interesting.

What doesn't make its way into case studies and HN headlines is all the stories of people who did get access to uncertain treatments and died anyway. Sometimes faster than they would have without the experimental treatment at all.

This isn't a case study about a breast cancer cure. This is a story about a single individual's cancer's response to an experimental treatment. For comparison, there are case studies of spontaneous remission in refractory cancers triggered by seasonal flu.

Virologists aren't sitting around waiting to develop cancer before they decide to roll out the miracle cancer cures. Oncolytic viruses have been researched, studied and tested on cancer patients for almost a century now.

This is a big simplification.

1. The interaction between oncolytic virotherapy and host immunity is complex. Consider that oncolytic viruses are also targeted by the immune system.

2. Oncolytic viruses that directly destroy cancer cells may not depend on a host immune response at all for therapeutic effectiveness.

3. There are many common chemotherapy agents that enhance anti-tumor immunity. For example, 5FU is understood to enhance anti-tumor response and activate the p53 pathway.

4. Immunosuppressive chemotherapies can still enhance anti-tumor immunity by changing the tumor microenvironment. This is one of the principles behind combined chemo-immunotherapy regimes in treatment of solid tumors.

5. Some immune cells promote tumor growth and suppress anti-tumor immunity. Tumor associated myeloid cells are one example of an immune cell suppressed by chemotherapy that promote tumor survival.

This is just scratching the surface of some of the complexities here. In general, cancer and cancer treatment are incredibly complex with massive variation not just between types of cancer but within individual cancers themselves. Oncology does not lend itself to simplistic thinking.

He's not wrong though.

It's more of a "not even wrong" statement. It's the kind of useless and reductive analysis poor leaders trot out from a position of personal frustration after failing to surmount the challenges involved in steering a complex, messy, human institution towards success.

It's incredible how consistently this maladaptive "everyone just work harder!!" mentality crops up among failed leadership in institutions of all kinds and sizes. In this respect, Schmidt is no different to the average frustrated restaurant owner, blaming his business failures on his staff's work ethic, "no one wants to work hard these days" and other copes.

Let's say we go ahead and assume that long grinds and 100% in-office attendance is the only way a successful and highly engaged team can look. Getting to that point would still require leadership to perform the actual hard work of creating the right conditions and incentives for that successful, engaged work to emerge. Shaking your fists at the air and saying "work harder people! we need to win!" doesn't cut it.

If Schmidt allowed himself to look more closely and reflect more deeply, he would realize that Google is and was full of extremely hard workers. But their "hard work" more likely took the form of navigating Google's political structures and chasing up internal promotions and prestige.

Moonshot or no, these incremental improvements in surgery do translate into cured patients and saved lives. Surgery is the initial intervention in many cancers and the course of disease is highly sensitive to it.

Dietary advice is one thing. Woo peddling and fringe medicine is another. The original question didn't even ask about rheumatoid arthritis.

I'm not overreacting. There are many people who will read my comment and know exactly what I'm talking about. This is simply a "you get it or you don't" situation where you're currently on the "doesn't get it" side.

I know this is well-intentioned, but please keep this stuff out of health threads about serious health conditions. This kind of material lacks rigor, dramatically misrepresents the state of scientific and medical thought, cherry picks studies, overstates the effect sizes and passes off speculation and easily digestible explanations (for laypeople) as emerging medical truths.

Long-term sufferers of RA - and the people in their support networks - know first-hand that RA is a complex and progressive condition that requires some pretty hardcore medical interventions to manage. Like other auto-immune diseases, different people will experience different disease courses. A very small few will be lucky enough that their disease goes into remission for no clear reason. Others will try everything under the sun only to see their disease become worse and worse. The reality for sufferers is that there aren't quick fixes and simple triggers.

It's reasonable to expect that general lifestyle interventions such as healthier diet and the right type of exercise regime may improve symptoms within the margins permitted by the underlying disease process. But promoting content that centers the role of "lifestyle" once RA has already developed only trivializes the disease and widens the empathy gap that sufferers already face.

"AI assistants are bad" isn't the take. The thing making people feel icky/uncomfortable about Her - and about the 4o demos - is not the idea of Samantha but the idea of Theodore. It's about nerds' idealized AI versions of womanhood. Tirelessly flirty, giggly servants who they can interrupt and talk over as they please, that come without the inconvenient features of personhood.

This is the thing that is so wild to me. Outside of tech, everyone I encounter in day-to-day life intuitively understands that tech dudes inventing AI voices to flirt with them is not awesome or cool.

At this point, I have to assume that being out-of-touch is deliberate branding from OpenAI. Maybe to appeal to equally out-of-touch investors.

but I don't think she has any rights to voices that merely sound like hers, when they're not being used to fraudently imply that she's involved.

IANAL but it's actually not that simple. There are laws and precedents around soundalikes.

I think the cultural defense for this is really lame. Was there really cultural value in evoking "Her"? Because outside of the narcissistic male nerd culture of laughing at your own jokes, and fixating on your own cleverness, there doesn't seem to be. No one outside of this bubble actually thinks it's super cool and culturally valuable for tech companies to attempt reinvent - seemingly without a trace of irony or self-awareness - the cautionary tales and trappings of dystopian sci-fi.

From what I have seen, the demos struck the average person as cringey and left women in particular with a deep sense of revulsion.

There is no point to the article. It's another boring anti-React/Electron screed but with a conspiratorial, junior high polsci bent to it.

Having frameworks that allow more devs to do more things means correspondingly that more people can be hired for those things with less training. In this respect, there is an obvious overlap between employer and developer interests.

Take MongoDB from the article. IME the biggest advocates for MongoDB are not managers (who barely have any idea what it is). It's developers who just want an "easy" way to store data that corresponds closely to the JSON document format they are used to, and doesn't force them to think about "annoying" and "outdated" things like relational modeling, joins and schema migrations.

If fungibility and lowest common denominator hiring drove the popularity of dev tools, then .NET and Angular would be ruling the roost.

Quick check of the bio reveals his alternative is Svelte. Weird since the usual Svelte argument is that it's so simple anyone can pick it up and be effective the next day - unlike bloated old React where each team needs an elite React hooks whisperer to navigate the hundreds of footguns required to correctly render a list without tearing under React Suspense.

Feels a lot like a lengthier, polemical way of saying that frameworks become popular because they allow more people to do more stuff more easily. But this is bad because businesses benefit or something.

None of the ideas here stand up to scrutiny. For example, Angular with its more opinionated design and framework approach was a much better choice for regularizing developer effort and making devs fungible. But by all accounts it lost to React, which is frequently criticized in comparison for allowing too much individual variation in development style.

React itself has become increasingly more complex over the years as it's moved up the S curve and the design direction is driven towards more and more remote parts of the problem space. More than ever, it requires specialized knowledge to use correctly.

Svelte - of which the author is a fan - fits his thesis much more than React. Svelte is billed as simpler and easier to learn and use than React, without requiring developers to wrap their heads around concepts like hooks, reducers and suspense.

If medicine is just following a decision tree why would we need LLMs to do it? Computers have been able to follow decision trees for 70 years or something.

Most places I've worked in the last 6 or so years are obsessed with tooling and add so many lint rules that it's often impossible to merge your pull request if you decide to format your code in a way that violates the rules in some trivial way.

Symptom of nothing better to do, I have found ;)

Hard to picture someone who values their time blocking PRs on tiny stylistic nits.

I recovered, but it took a long time.

I don't know about RA. OPs blog post sounds a bit too optimistic for someone living with untreated RA for 3 years.

Yeah, I agree. In some sense, tendons remodel more easily than the brain. Somatic pain disorders can easily ruin someone's life.

The question here isn't about damage but about what mechanisms in the "pain pathway" are responsible for OP's specific experience of this pain. One day he has sudden, debilitating pain - in both wrists - in the absence of a traumatic event. He's conspicuously omitted any kind of medical diagnosis, and the whole thing is part of an existential crisis on the tail end of an insane, psychologically unsustainable workload. Not a typical presentation of RSI, that's for sure.

Seems like a bizarre omission from the article given the number of tangents and topics covered. My guess is that going down the traditional medicine route lead to him being told there was no organic basis to the pain.

As someone who went through a similar experience, I would not be surprised if the author's pain is/was entirely psychosomatic (this doesn't diminish its severity or significance). Probably a direct result of burnout.

increase of supply lowers the price at equilibrium

The thing you're missing here is that "healthcare services" and "doctor's labor" aren't the same unit. Ceteris paribus, efficiency increases allow the price of healthcare to decrease while the price of doctor's labor increases. The thing that makes this non-contradictory is that a single doctor can now produce "more" healthcare. Economics says the opposite of what you think it does here. Increasing productivity drives expansion in market size, which drives up the ratio of value in the market to its labor inputs which drives up salaries.

Like I said, there might be important real-world reasons why these scenarios won't play out in medicine the way the theory predicts. But so far, you haven't provided any.

Manufacturing has also seen the opposite of what you are saying here. Global manufacturing production value has exploded over the last century, quite literally lifting billions of people out of abject poverty. In particular the last 3 decades of enormous per capita income increases in China have been driven by industrialization. I'm guessing you're taking a US-centric view that is exclusively focused on the local collapse of US manufacturing. This is to do with globalization and free trade, not improvements to labor efficiency.

1. There is a meaningful difference between replacing and assisting. Replacing implies being able to take over the entire function. Technology that assists doctors with one part of their function, or process, can improve their output but is not capable of producing like-for-like output on its own. So the question of whether AI can replace or only assist doctors is very relevant to determining its impact on the role. Power tools didn't replace tradesmen, for example. If AI was able to replace doctors, then your clinic would be able to scale down to 0 rather than 10.

2. If a clinic can use an AI tool to make doctors 10% more productive, doctors become worth more rather than less. Firms are incentivized to hire more rather than less in this scenario. What you're invoking here is the "lump of labor" fallacy. There are market conditions where increasing efficiency really does reduce quantity demanded, but it's not clear that medicine really is one. As far as I can tell, far from there being a fixed lump of medical work, the general population in most of the West is under-serviced and struggles to get reliable, timely, cost-effective access to medical expertise.

It definitely can be. The presence of uncontrolled interruptions is a common complaint in open offices. Individual employees tend to manage this issue with noise cancelling headphones. I've also seen eng managers create dedicated periods of focus time or designate a focus room for this reason.

But it doesn't matter either way, because in-person is a different medium to text with a different dynamic. If you'd read the article, you'd see this is directly called out:

You're trying to be polite by not jumping right into the request, like you would do in person or on the phone. But Chat is neither of those things.

A group message is still chat, but it is a different dynamic to sending out individual messages. Can you honestly say - in good faith - that you don't see the differences between directly and personally addressing someone, versus addressing a group? Try to put yourself in other people's place and imagine how these things might be different.

I also agree with the sibling comment that walking up to a bunch of people working and saying "Hello" with no context then waiting around for a response would be poor etiquette. If you can see that people aren't working - that they are taking a break or already idly chatting - then I would say it's fine. But these are the types of rich context clues that you get from in-person communication that you usually don't get from chat.

Your reply from elsewhere:

They don't need to poll a channel, most chat applications will pop up a notification when a message appears.

It was you that said the purpose of "hi" was to have them check back. If the notification were sufficient on its own, then you wouldn't need to say "hi" at all. They'll receive a notification once you ask your question either way. The selfishness is in the asymmetry of the interaction. The standalone "hi" is the lowest effort and highest ambiguity way of interrupting someone and asking for their attention.

What would be the purpose of providing more context before actually asking a question?

So that people have some idea of how important your question is going to be, how long it might take to answer, whether it is relevant to the work they're currently doing and so on. Maybe you see a "hi" in chat and it doesn't matter to you either way. If that's the case, you are probably a minority among technical workers.

That's even worse. You're sending out rounds of "Hi's" to multiple people, distracting them to prepare them for a question you might not even ask?

Candidly, you sound like an incredibly selfish communicator.

At the least, its a communication that you are about to ask them something, i.e. check back here in the next minute or so while I type out my question.

This is part of what makes it such poor etiquette. You've created this ambiguous, open-ended distraction for the other person. You expect them to divert focus away from what they're doing so they can poll the chat channel waiting for you to get your thoughts in order.

At least include some context so they have a better idea of what you're about to ask.