I found that the weakest part, tbh, along with the bit about chopsticks. They both feel like they have that classic Evolutionary Psychology problem of being vaguely plausible ideas without any real evidence or counterfactual examples that back them up. It's very easy to come up with these retrofitted explanations, but it's a lot harder to demonstrate they're accurate.
At least for me, the stuff in the article doesn't really match my experience of either React or Vue. The article describes a very traditional separation between content, presentation, and behaviour, but both React and Vue explicitly push for merging these and instead splitting along component lines. I very rarely see code used as the interface between different frontend disciplines these days, regardless of framework. I'm very sceptical of the idea that React specifically leads to worse outcomes for e.g. accessibility - in my experience, React "thought leaders" love to talk about accessibility, and it's a very important topic in that space. Rather, I suspect the ubiquity of React means that there's a lot more bad React out there. The author is observing those bad React codebases and attributing that to something unrelated. Also, I find it odd attributing Tailwind's success to React when one of the big advantages of Tailwind above most other similar tools is that it's completely framework (and indeed language) agnostic, which has lead to it being picked up more by developers working outside of React, unlike most of the more standard CSS-in-JS options which tend to be much more React-focused.
I think the author makes a lot of good points throughout the essay, and I think the broad thrust of their argument is good and correct. But I think these examples are quite weak.
Speaking only for myself as an engineer who spread through the stack from back to front, React was a lot more approachable that other web frameworks because Javascript had patterns that resembled the backend languages I was coming from, and React let me live entirely in JS. In many ways, React let me run (build functionally complex web apps) before I learned to walk (understand CSS well). When the author talked about “de-emphasizing” writing CSS, I interpreted it as coming from this angle, ie CSS in React is second-class to JS, in contrast to the traditional separation of content/presentation/behavior where it’s on equal footing with HTML and JS.
Re: the Tailwind connection, I suspect that a reason for Tailwind’s success is that it lets people like me get by without understanding how CSS actually works. Under the author’s argument, this further entrenches the cycle by enabling developers who don’t actually understand CSS to build FE apps.
I've never used React so didn't actually understand how much it reframes the web in terms of JS. It now makes much more sense to me how it pushes towrads re-implementing everything in JS, so you get bespoke half-broken components compared to the browser's native version.
editing a React component requires you to know JavaScript for almost everything. This means that when building a React team, you're going to focus on people with development skills: someone expert at HTML and CSS but not great with writing code is going to be closed out by the gestalt. React, in fact, tends to de-emphasise the work of design or even writing CSS in general, pushing developers towards solutions like Tailwind
Langston Winner extends this in the 1980s to argue that technologies inherently exploit and reinforce political systems.
Technologies can be used in ways that enhance the power, authority, and privilege of some over others. […] But we usually do not stop to inquire whether a given device might have been designed and built in such a way that it produces a set of consequences logically and temporally prior to any of its professed uses.
I was impressed by the Szechuan food example at first, but then it seemed a little off-topic. You can pick out the chicken just as well with a fork!
The bigger disconnect seems to me that in my culture I was raised to eat everything on the plate, so it wouldn’t occur to me that half the food is “garnish”. Same with Thai soups, by the way — you definitely want to leave the chilis at the bottom of the bowl, not regard them as vegetables to be consumed.
Is this truly just because the food evolved under the influence of chopsticks vs. forks?
Another cultural disconnect that might help your understanding is that it's a shared communal dish and you're expected to take a bit back to your solo bowl of rice. If you sat down to eat an entire plate of 辣子鸡, peppers and all, it would be quite unpleasant regardless of the utensil you chose to use.
I'm not sure I'd call it common sense. I only ever hear it from people squirming to avoid ethical questions about their source of income or policy opinions.
Unfortunately common, but more because common sense is asking the questions that it excuses one from answering.
This isn't to say that this is never the correct call to make, or that you should never use the tool... But well... if you're going to be using a tool like that, it's incumbent on you to know what you're doing and be very careful about it.
This is exactly the position I've been trying to get across. We need to be careful, and we need to carefully consider both how we use tools and when and why and also how to make them better tools (by which I don't mean better at doing what they are designed to do, but rather better for the user and society in general). If social media is bad as it exists today, but sees good use also, then we need to bend our effort towards increasing benefits (both through thoughtful use and though redesign of the tool itself) rather than simply abstaining and hoping someone else will solve the problem.
Based on a lecture he gave in 1949, which featured the following observation:
"Agriculture is now a motorized food industry, in essence the same as the production of corpses in the gas chambers and the extermination camps, the same as blockades and the reduction of countries to famine, the same as the manufacture of hydrogen bombs."
It is wrong and utterly foolish to say something like, "It's how we use ______(e.g., television, the Internet) that matters." As McLuhan put it in Understanding Media,
Our conventional response to all media, namely that it is how they are used that counts, is the numb stance of the technological idiot. For the 'content' of a medium is like the juicy piece of meat carried by the burglar to distract the watchdog of the mind...The effects of technology do not occur at the level of opinions or concepts, but alter sense ratios or patterns of perception steadily and without resistance.
I feel like this framing is true and useful. In particular, as a tool builder, I want to consciously shape my tools so that they produced the maximum goodness and minimum badness.
Something that worries me is that when people say "(not) just a tool" they are actually trying to say "should (not) be regulated by the government" by proxy. I find those indirect arguments unproductive, but representative of much of the modern public discourse.
[Meta] this article is amazing because it shows you can explicitly quote a “notable Nazi (and unfortunately also very important 20th century philosopher)“ and people on lobste.rs will upvote it if it says LLMs bad.
What should the author should have done differently when describing the human-tool gestalt idea written about by the Nazi philosopher? The word “unfortunately” in the author’s description of Heidegger makes me think that the author dislikes Nazis and is not endorsing them, so it’s not like this is a pro-Nazi blog post that people are failing to criticize.
Also, LLM hate isn’t the only explanation for why people are upvoting the article. When you wrote your comment, there were already other comments that praised some of the article’s non-LLM examples. I found the article upvote-worthy after reading the explanation of human-tool gestalts and the analysis of the ideas that React promotes.
It is an unfortunate fact that, when faced with such an analysis of technology, tech people will often think that for whatever reason the considerations that apply to literally the entire rest of the world of technology somehow don't apply to them.
Web tech is absolutely rife with this.
Favorite example: DHH insisting over and over again that fat controllers are the true Rails way, that Rails/web was somehow different enough that they could ignore software engineering wisdom around coupling, and the legions of Rails users swearing up and down that Rails Is Enough.
What followed was entirely predictable: the "slow tests" Rails meme, ungodly hacks abusing fork/exec to try to speed up said tests (with spectacular explosions when it hits code that doesn't expect that), a weird sort of collective acceptance that it sucks, and a protracted "discussion" about service objects that ended up becoming niche. Rails chose to ignore engineering history and hope for the best, and then refused to update their guidance even when huge users basically had to carve their own path to continue scaling.
I disagree with the thrust of this article, despite agreeing with most of its contents. I think the origin of disagreement is encapsulated in this paragraph:
Our industry, rather contrary to that idea, has tended to create tools destructive to both of these things. We've built social media networks that isolate people from each other and have become vectors for the massive spread of disinformation. We've built cryptocurrencies that have enabled fraud and deception on an enormous scale. We've built classification algorithms that reinforce the worst kinds of bigotry. And of course, there are the fucking LLMs, that do all of these things and a whole lot of other evil shit besides.
But before that, I want to point out that the article does not sufficiently distinguish the ethical considerations around the deleterious consequences of forming a gestalt with a tool to oneself, and the deleterious consequences to society. Here are examples of the former:
AI psychosis
Pushing attitudes and ideologies towards carelessness and slop
extended use will change how you see the world
interests and opinions that are separate from the human in the gestalt, often opposed to them and able to assert them over the human part of the gestalt
Now, it's not wrong to raise both sorts of consequence, but it's not clear to me whether the author believes that these are two separate sorts of consequence. I do, and that's probably one cause of the disagreement. I am perfectly willing to take risks on my own behalf! I don't own a car, and I am confident that my use of LLMs will not push me towards carelessness and slop. I think the author is saying (in my words and with apologies to both Nietzsche and Heidegger): be careful when you form a gestalt with the tool ready-to-hand, because the tool ready-to-hand is also forming a gestalt with you. Fine, but that's my choice, ultimately! The impact on society of my choice to own a car, or not, and the artifacts I produce with LLMs is something else.
But I don't agree with the author on that something else either. Coming back to the quoted paragraph above, I certainly would not work for a company producing, for example, addictive social media, gambling products, or tobacco products. But that's my own ethical choice. If I want
to develop a tool for my own use, because I see a personal benefit in it, and share it with others who would value it, perhaps receiving something in return, then I should be free to do so. Others are welcome to persuade me otherwise, but ultimately I'm not predisposed against developing a tool because some other members of society might choose to engage with my tool and subsequently suffer some harm. That's on them.
As much as I side with the author on « the fucking LLMs » (this change of tone really caught me off-guard), I'm skeptical about the point being made.
I understand (and agree) that tools definitely shape how we interact with the world. In essence, when you have a hammer, everything is a nail.
Now what ? We've been using tools as means to evolution, from the very first sylex, to wheels, planes, and now AIs.
But how do we decide whether a tool is good or bad for evolution ? Is that even a call to make ?
Is the point of the article that using a tool can and will change your political views ? Or is it the other way around ?
CobyPear | 13 hours ago
as someone who's used the phrase "this is your brain on react" numerous times, this piece really resonated with me.
Johz | 10 hours ago
I found that the weakest part, tbh, along with the bit about chopsticks. They both feel like they have that classic Evolutionary Psychology problem of being vaguely plausible ideas without any real evidence or counterfactual examples that back them up. It's very easy to come up with these retrofitted explanations, but it's a lot harder to demonstrate they're accurate.
At least for me, the stuff in the article doesn't really match my experience of either React or Vue. The article describes a very traditional separation between content, presentation, and behaviour, but both React and Vue explicitly push for merging these and instead splitting along component lines. I very rarely see code used as the interface between different frontend disciplines these days, regardless of framework. I'm very sceptical of the idea that React specifically leads to worse outcomes for e.g. accessibility - in my experience, React "thought leaders" love to talk about accessibility, and it's a very important topic in that space. Rather, I suspect the ubiquity of React means that there's a lot more bad React out there. The author is observing those bad React codebases and attributing that to something unrelated. Also, I find it odd attributing Tailwind's success to React when one of the big advantages of Tailwind above most other similar tools is that it's completely framework (and indeed language) agnostic, which has lead to it being picked up more by developers working outside of React, unlike most of the more standard CSS-in-JS options which tend to be much more React-focused.
I think the author makes a lot of good points throughout the essay, and I think the broad thrust of their argument is good and correct. But I think these examples are quite weak.
jkbach | 7 hours ago
Speaking only for myself as an engineer who spread through the stack from back to front, React was a lot more approachable that other web frameworks because Javascript had patterns that resembled the backend languages I was coming from, and React let me live entirely in JS. In many ways, React let me run (build functionally complex web apps) before I learned to walk (understand CSS well). When the author talked about “de-emphasizing” writing CSS, I interpreted it as coming from this angle, ie CSS in React is second-class to JS, in contrast to the traditional separation of content/presentation/behavior where it’s on equal footing with HTML and JS.
Re: the Tailwind connection, I suspect that a reason for Tailwind’s success is that it lets people like me get by without understanding how CSS actually works. Under the author’s argument, this further entrenches the cycle by enabling developers who don’t actually understand CSS to build FE apps.
ThinkChaos | 11 hours ago
Yes, I found that specific part very interesting!
I've never used React so didn't actually understand how much it reframes the web in terms of JS. It now makes much more sense to me how it pushes towrads re-implementing everything in JS, so you get bespoke half-broken components compared to the browser's native version.
schmudde | 15 hours ago
I had not read the chopstick example. Very nice.
Langston Winner extends this in the 1980s to argue that technologies inherently exploit and reinforce political systems.
More context here.
It's unfortunate that "just a tool" remains the common sense understanding.
wrs | 9 hours ago
I was impressed by the Szechuan food example at first, but then it seemed a little off-topic. You can pick out the chicken just as well with a fork!
The bigger disconnect seems to me that in my culture I was raised to eat everything on the plate, so it wouldn’t occur to me that half the food is “garnish”. Same with Thai soups, by the way — you definitely want to leave the chilis at the bottom of the bowl, not regard them as vegetables to be consumed.
Is this truly just because the food evolved under the influence of chopsticks vs. forks?
briankung | 6 hours ago
Another cultural disconnect that might help your understanding is that it's a shared communal dish and you're expected to take a bit back to your solo bowl of rice. If you sat down to eat an entire plate of 辣子鸡, peppers and all, it would be quite unpleasant regardless of the utensil you chose to use.
landon | 11 hours ago
I'm not sure I'd call it common sense. I only ever hear it from people squirming to avoid ethical questions about their source of income or policy opinions.
Unfortunately common, but more because common sense is asking the questions that it excuses one from answering.
singpolyma | 13 hours ago
This is exactly the position I've been trying to get across. We need to be careful, and we need to carefully consider both how we use tools and when and why and also how to make them better tools (by which I don't mean better at doing what they are designed to do, but rather better for the user and society in general). If social media is bad as it exists today, but sees good use also, then we need to bend our effort towards increasing benefits (both through thoughtful use and though redesign of the tool itself) rather than simply abstaining and hoping someone else will solve the problem.
lematheux | 19 hours ago
Never thought I'd see something sensible coming from Heidegger.
regulator | 15 hours ago
I found The Question Concerning Technology to be worth a read.
hyperpape | 13 hours ago
Based on a lecture he gave in 1949, which featured the following observation:
To paraphrase Anscombe (https://crookedtimber.org/2007/01/11/blogging-avant-la-lettre/), while there might be much to be said for this analogy, still it does not seem quite that illuminating.
freddyb | 16 hours ago
Langsec makes more sense now. (And so much more does too). Such a great way to explain how and why tools influence what we do.
pzel | 13 hours ago
From this summary about Marshall McLuhan's philosophy of media:
hc | 11 hours ago
the irony of using heidegger in an article warning about things inadvertently shaping your worldview, is staggering
could have gone with plato, who made a similar point about writing and forgetfulness
benji | 10 hours ago
I feel like this framing is true and useful. In particular, as a tool builder, I want to consciously shape my tools so that they produced the maximum goodness and minimum badness.
Something that worries me is that when people say "(not) just a tool" they are actually trying to say "should (not) be regulated by the government" by proxy. I find those indirect arguments unproductive, but representative of much of the modern public discourse.
I wish it weren't so.
ohrv | 11 hours ago
[Meta] this article is amazing because it shows you can explicitly quote a “notable Nazi (and unfortunately also very important 20th century philosopher)“ and people on lobste.rs will upvote it if it says LLMs bad.
roryokane | an hour ago
What should the author should have done differently when describing the human-tool gestalt idea written about by the Nazi philosopher? The word “unfortunately” in the author’s description of Heidegger makes me think that the author dislikes Nazis and is not endorsing them, so it’s not like this is a pro-Nazi blog post that people are failing to criticize.
Also, LLM hate isn’t the only explanation for why people are upvoting the article. When you wrote your comment, there were already other comments that praised some of the article’s non-LLM examples. I found the article upvote-worthy after reading the explanation of human-tool gestalts and the analysis of the ideas that React promotes.
mattgreenrocks | 10 hours ago
Can't help myself from nodding along:
Web tech is absolutely rife with this.
Favorite example: DHH insisting over and over again that fat controllers are the true Rails way, that Rails/web was somehow different enough that they could ignore software engineering wisdom around coupling, and the legions of Rails users swearing up and down that Rails Is Enough.
What followed was entirely predictable: the "slow tests" Rails meme, ungodly hacks abusing fork/exec to try to speed up said tests (with spectacular explosions when it hits code that doesn't expect that), a weird sort of collective acceptance that it sucks, and a protracted "discussion" about service objects that ended up becoming niche. Rails chose to ignore engineering history and hope for the best, and then refused to update their guidance even when huge users basically had to carve their own path to continue scaling.
tome | 8 hours ago
I disagree with the thrust of this article, despite agreeing with most of its contents. I think the origin of disagreement is encapsulated in this paragraph:
But before that, I want to point out that the article does not sufficiently distinguish the ethical considerations around the deleterious consequences of forming a gestalt with a tool to oneself, and the deleterious consequences to society. Here are examples of the former:
Now, it's not wrong to raise both sorts of consequence, but it's not clear to me whether the author believes that these are two separate sorts of consequence. I do, and that's probably one cause of the disagreement. I am perfectly willing to take risks on my own behalf! I don't own a car, and I am confident that my use of LLMs will not push me towards carelessness and slop. I think the author is saying (in my words and with apologies to both Nietzsche and Heidegger): be careful when you form a gestalt with the tool ready-to-hand, because the tool ready-to-hand is also forming a gestalt with you. Fine, but that's my choice, ultimately! The impact on society of my choice to own a car, or not, and the artifacts I produce with LLMs is something else.
But I don't agree with the author on that something else either. Coming back to the quoted paragraph above, I certainly would not work for a company producing, for example, addictive social media, gambling products, or tobacco products. But that's my own ethical choice. If I want to develop a tool for my own use, because I see a personal benefit in it, and share it with others who would value it, perhaps receiving something in return, then I should be free to do so. Others are welcome to persuade me otherwise, but ultimately I'm not predisposed against developing a tool because some other members of society might choose to engage with my tool and subsequently suffer some harm. That's on them.
z3bra | 8 hours ago
As much as I side with the author on « the fucking LLMs » (this change of tone really caught me off-guard), I'm skeptical about the point being made.
I understand (and agree) that tools definitely shape how we interact with the world. In essence, when you have a hammer, everything is a nail. Now what ? We've been using tools as means to evolution, from the very first sylex, to wheels, planes, and now AIs. But how do we decide whether a tool is good or bad for evolution ? Is that even a call to make ?
Is the point of the article that using a tool can and will change your political views ? Or is it the other way around ?
tuxes | 22 hours ago