Apologies for my recent hiatus in publishing articles: the household has been, once again, struck down by a spectacular range of health issues, and while I will allow that there are significant advantages to having been hit by this after the launch of Arca rather than before it, it's still unpleasant. I'm hopeful that more regular writing will resume soon.
If you work or write in the social criticism of tech, you hear a wide range of views and opinions about tech, some of them startlingly hostile to the very idea that you might be doing what you're doing. Consequently, you very often find yourself experiencing a wide range of different criticism of your work, your ideas and your viewpoints, some of them extremely valid, others... less so.
A common refrain when critiquing specific technologies in this way is that engineers will say that a given technology is "just a tool". You hear it when talking about all kinds of things: shells, containerisation, reactive frameworks and most recently it comes up a lot in discussions of LLM technology. What this response seems to mean is the following:
- The technology being discussed has no bearing on the attitudes or ideologies of the person holding it: use or non-use is ideologically neutral.
- The technology being discussed can be effectively separated from the people using it: nothing can be inferred about a person from the technologies that they use.
- The technology is used fully consciously: you are always fully aware that you are using a technology and have complete volitional freedom to use a different technology at any time.
- The technology does not influence the kinds of things likely to be done with it: use of the technology doesn't tend to encourage one course of action over another ("guns don't kill people, people do").
The consequence of these statements being true is that the only valid criticism of a tool or technology becomes how well it performs the task that it's designed to perform. You cannot, for example, say that a tool teaches bad habits, because if it isn't explicitly a teaching tool what it teaches people becomes out-of-bounds for discussion. Similarly, you cannot discuss the fact that driving an SUV tends to encourage shitty driving behaviour on the road because, after all, the SUV is "just a tool" and there are some legitimate use cases for it: because this is the case, the fact that the SUV has ideological and behavioural consequences is something that we're not really allowed to discuss. This kind of folk understanding of tools is very common, but unfortunately it's almost all wrong.
To explain why saying that something's "just a tool" is an illegitimate move, we shall have to rely on the work of notable Nazi (and unfortunately also very important 20th century philosopher) Martin Heidegger. While Heidegger's major work, Being and Time mostly discusses ontology and human being as being being-towards-death, a significant part of Heidegger's work there and elsewhere has to do with the nature and being of tools.
Heidegger thinks about tools like so: a tool is something (a thing that has being in the Heideggerian sense, not necessarily an object) that represents an extension of human capabilities in some way. A tool is thus any kind of thing that lets you do something you wouldn't otherwise have been able to. The important (for us) conceptual leap here is that when a tool works well, or isn't broken, it develops what Heidegger describes as a "ready-to-hand" quality: the tool fades into the background as a kind of human-tool gestalt of a human with the additional capability afforded by a tool forms. As a simple example of this, consider eating with a fork. When the fork is well-designed and not broken, you don't explicitly have "I am using a fork" in your conscious mind while eating dinner. You simply eat, and the assumptions encoded into how you eat with a fork and what you might eat with a fork fade into the background and are taken as a given. You only explicitly notice the fork when it doesn't work: perhaps a tine is bent, or maybe you're trying to eat soup. Either way, the tool only becomes, in Heidegger's wording, present-at-hand when it becomes obtrusive, or an obstacle to doing what you want to do.
The human-tool gestalt that forms with a ready-to-hand tool, however, is not neutral. Rather, the tools that are ready-to-hand each create a different gestalt, with a different range of allowable and potential "things you could do" and, accordingly, different thought patterns, ideas and ideologies that go with that. Take the fork, for example: it is, notionally, "just a tool" and should thus be neutral. Except that it really, really isn't. If you need an example of this, go find some older or more culturally sheltered people from an Anglo-Saxon background (if these are difficult to find where you live, substitute people from a different food culture where you do) and take them to a Sichuan restaurant.
Put that way, the point is fairly clear: the affordances that eating with knife and fork allow are pretty radically different from the affordances that you have when eating with chopsticks. As a first point, the assumptions encoded about how food is prepared are completely different: you cannot feed diners eating with chopsticks a steak in the European fashion, and food in general has to be carefully cut so that it can be easily grasped with chopsticks. This follows on to what kinds of dishes get prepared, as in the situation where a lot of Sichuan food has an unfair reputation for being unreasonably... well, just take a look at this photo of some delicious Chongqing Laziji:
.jpg)
For a human-fork gestalt, this is a prohibitive amount of chilli. Even if you have the heat tolerance to enjoy all of that, the chilli in such a presentation remains dry and somewhat unpleasant to eat. This is not a dish, in short, that a human-fork gestalt would make. A human-chopstick gestalt, however, can rummage through the chilli with their chopsticks, pick out bits of chicken that are lightly spiced and made fragrant and just... avoid most of the chilli. In that case, our Chongqing Laziji becomes a legitimate and somewhat impressive presentation of what's actually a really good dish. In the extreme, such different affordances can go so far as to actually lead those gestalts to have different ideas about what they consider food at all.
As far as food is concerned, this mostly leads to interesting differences in culinary culture (though of course this is a fertile ground for racism and other such unpleasantness). With other tools, however... well, we have problems. The human-car gestalt, for example, experiences life and has opinions that are quite different from the human-bicycle gestalt, the human-train gestalt or even the bare human with no mode of transportation. What's possible and what's desirable for a driver is pretty radically different even to what's possible or desirable for the same human outside of a car: in individual instances of this happening, we call it road rage, and on a policy scale it leads to things like people getting extremely angry about bike lanes even when said bike lanes would benefit them at all times when they're not in their car. The human-car gestalt has 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.
And then, of course, there's the human-gun gestalt, or the "gun-man". Putting a gun in a person's hand radically changes how they see the world and the range of potential actions that they have available to them and that might seem good. You can see that, most dramatically, in policing in the USA. When you have a gun, you're predisposed, as part of the human-gun gestalt, to look for reasons to shoot. Situations that might ordinarily be resolvable without violence escalate into shootings because of the affordance that the gun offers. Intimidation and the demand to get your way take priority over negotiation and compromise, potential threats become foregrounded in the world and other people, imperceptibly and by degrees cease to be fellow humans and become potential subjects of violence on your part.
Clearly, then, nothing is ever "just a tool", unless, perhaps, it's a broken or obstructive one. This is enough to throw the way that the tech industry talks about all this into question by itself, but that wouldn't be very interesting, so in the next section we'll discuss how two outwardly very similar tools end up creating very different gestalts and very different potentials for action.
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. This kind of overweening arrogance is, of course, why we find ourselves in so much of the shit that we're in, but nonetheless, to really get my point across it's probably a good idea to spell out how these dynamics operate in the tech world specifically.
For the purposes of this article, we'll look at Vue and React, two reactive frameworks commonly used to build web interfaces. These are ideal for our purposes: they aim to facilitate roughly the same task, but have enough differences in design choices and philosophy that it's possible to see how the differences in the tool create a different gestalt in the human space.
To demonstrate, let's have a look at how an identical component is built in React and Vue respectively (I've chosen a basic form that I found here):
import React, { useState } from "react";
function NameForm() {
const [value, setValue] = useState("");
return (
<form>
<input
value={value}
onChange={(e) => setValue(e.target.value)}
placeholder="Enter name"
/>
<p>You entered: {value}</p>
</form>
);
}
<template>
<form>
<input v-model="value" placeholder="Enter name" />
<p>You entered: {{ value }}</p>
</form>
</template>
<script>
export default {
data() {
return {
value: "",
};
},
};
</script>
These two components do exactly the same thing: create a form with an input and a paragraph tag below it that displays the value entered. The gestalts that the two frameworks create, though, is very different.
We can begin with looking at what kind of object each of the components is presented at. React presents us, immediately, with a function written in JSX: the component as we interact with it is a function. HTML, CSS and the like are returned by the function, and the expectation is that when writing React you think in JavaScript first and only really emit HTML as an output. This extends to the way binding works in React: it's presented explicitly as an imperative command to set the value of the template variable to the value of the form input whenever it changes, which is a very JavaScript way of thinking about it.
Vue, by contrast, presents us first with an HTML document: the component as presented is HTML with whatever JavaScript is needed included organically in a <script> tag. Binding in this component is declarative and two-way: we tell the value of the paragraph text what it is rather than what to do on a change. In general, you're expected to think mostly in terms of HTML, with the amount of JavaScript being written mostly being kept to a minimum.
These two ways of looking at things result, if you'd believe it, in two very different tool gestalts, even with exactly the same engineer working with them. This begins with the basic question of "who should be working with you on this project?" React almost demands the answer to be "other JavaScript software developers", because 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.
Vue, by contrast, opens up more space for people with different skillsets: someone who knows HTML but no JavaScript or who's great with CSS can edit the two-thirds of a Vue component that they know about and not touch the <script> tag and thus contribute effectively without the demand to know JavaScript preventing that. This opens up the range of people you might want on your team: designers, accessibility experts, semantic HTML experts... there's a whole gamut of skills that a Vue-using team can make use of that a React-using team will simply lack. And importantly, while none of this is directly demanded by the tools as such (someone could hypothetically edit HTML blocks embedded in a React component if they were careful), the constraint emitted by the gestalt is much stronger than that emitted by the tool itself, and in fact creates an ideological belief about who should and who shouldn't be involved in the work of web development that's what does the real damage.
Looking beyond that element, React and Vue prioritise different things as being important in the development of the component. With HTML and CSS being first-class elements in Vue, the Vue gestalt encourages thinking about structure, layout and presentation: the app is thought of as an interactive document that does document things and can use JavaScript to manage that interactivity. React tends towards the opposite pattern, in that the app is something that does computation and uses HTML and CSS to present the results of the computation to the user. The gestalts, then, will develop attitudes and ideologies that support those positions, and importantly, those ideologies will then become, for better or worse, part of the ideology of tech more generally.
Neither of these approaches are inherently wrong or inherently better, but they clearly shape attitudes and worldviews quite drastically and in ways that are very much not neutral. If the expertise of people who can't produce JavaScript is not respected, that will have significant negative consequences on the usability and accessibility of the resultant app. Some people are going to be included, others excluded, and who is included and who is excluded is liable to include the people who already hold power in the social sphere in which these tools operate and exclude those whom these people consider, in one way or another, "lesser". And all this from a choice of tool.
Much of the tech industry, what it does and the attitudes and ideologies that it presents are shaped by precisely this kind of gestalt dynamic, I think. The tools we build, shaped by already-existing ideologies, proceed to reproduce those ideologies and attitudes within the gestalts that they create. In particular, the React/Vue split is one that we see reproduced by a lot of tech tools, and it's the ideology of developer supremacy. The idea, repeated across tools as far apart from each other as React, Kubernetes, Docker, an awful lot of Linux distributions in general and most cloud providers, is that the way a software developer would do things is the way they should be done, and that other approaches to the task are necessarily inferior. This means that when the choice between writing more HTML on one hand and more JavaScript on the other comes up, the choice will always lean more towards JavaScript, whether or not that's actually a good idiom for whatever you're trying to do (Tailwind CSS is quite directly an attempt to resolve issues that primarily stemmed from React and JSX not playing nicely with standard CSS). This tends to create tools that are easy for developers to half-ass, but that might be difficult for say, designers or UX specialists to work with. The sinister thing here is that this tends to reproduce itself: tools that are easier for devs to work with tend to subtly reinforce that, for example, writing HTML or CSS is low-status development work (in some circumstances they can be, but they're as much design tools as development tools). This then subtly pushes designers, UX specialists, accessibility and SEO experts and all those kinds of people into a subordinate position, because the tools are designed for developers to do the job badly so they kinda bounce off them. One imagines that tools written for these people would create a more useful gestalt and get better status for them and better outcomes overall, but existing practices (created by the tools) push against that hard enough that it takes real effort to change the situation, so most people don't. And thus the cycle reproduces itself.
Obviously, I think that given such circumstances, where even if tools didn't directly create any of the immense number of social and cultural problems present in tech, they're heavily involved in reproducing and justifying said problems by virtue of their ready-to-handedness being directly pointed at the reproduction of those problems, maybe saying that something's "just" a tool isn't the best idea.
All that being done, we can now return to our original question: what is so wrong about saying that something is "just a tool"? Considering how much of an impact tools can have on the gestalts that they become a part of, and thus the ideologies that the human parts of those gestalts hold, something might be a tool but as it happens, tools are actually really, really important. Given that we in tech are largely in the business of creating tools for both ourselves and for other people to do things, then, we ought to be very, very careful about what kinds of tools we build. We should build tools that tend to create gestalts that support individual flourishing and social harmony or, if those cannot be balanced, that carefully balance the need for both against each other and other considerations.
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.
LLMs are a particularly important case: we can quite directly see the use of the tool forming some very strange gestalts. In its most extreme case, we call this AI psychosis: the gestalt has attitudes, ideologies and even a perception of the world that diverges so much from what we consider reasonable that we label it an illness. Obviously it's hard to say when observing something like this that the tool has not, in fact, drastically shaped the person using it's attitudes, experiences and ways of being in the world. The phenomenon described in this article quite clearly happens, then, and if it can happen so drastically and obviously with this one tool, we would do well to expect to see more subtle and far-reaching consequences over a much wider range of technologies.
Now at this point people will point out many cases in which tools that are in some way "bad" have been used for things that are good and beneficial. Frameworks and languages that I'd consider harmful have been used to build tools that have done a lot of good. LLMs have been used, at this point, to build quite a few useful tools: hell, even I've had some success using them when I've tried. These are all valid arguments, but in the gestalt framing they don't actually refute the point: it's quite possible for a tool to be used for good things, and in fact for it to be necessary, but for the viewpoint that it fosters to nonetheless be one that we shouldn't encourage. After all, all kinds of fascinating weapons systems are being used in defence of freedom and democracy in Ukraine, and this is both good and necessary, but none of this makes the effects on leaders of having access to bombs and artillery, or on individual people of having access to firearms, not ceteris paribus deletrious. Carrying a gun or driving an SUV will, in general, make you more predisposed to being an asocial asshole who doesn't care much for human life, and while it's quite possible to fight this tendency and a lot of people do, the logic of the gestalt for these tools still pushes towards the bad.
This means that even when you, for example, use an LLM to build some useful software that helps your nonprofit do well, and even when you're very careful about what generated code ends up actually being deployed (even if you literally read every line of code), you're nonetheless fighting against the ready-to-hand nature of the tool that pushes your gestalt (and thus your attitudes and ideologies) towards carelessness and slop every step of the way. This isn't to say that this is never the correct call to make, or that you should never use the tool: as before, I am generally supportive of fighting back against an invading force even if I think that guns generally do bad things to your attitudes. 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. You need to be aware that extended use will change how you see the world and put protections in place to mitigate that (it's an interesting observation, by-the-by, that regular militaries tend to put quite a bit of effort into that kind of mitigation when it comes to firearms, while police forces often don't). You generally want to make that kind of compromise as infrequently as possible, and you only want to make them when the problem that you're facing is drastically bad enough that making that manner of compromise makes sense.
If we were simply using these tools (if, say, React and Vue were things we found in nature, unchanged by human hands), this might be an unfortunate fact about tools and about tech in general that we needed to be wary of. But we make these tools. When we develop any tooling in tech, we make decisions about what shape it's going to take, what kind of affordances it offers, what gestalts it forms. We not only have to choose what compromises we make, we as an industry largely dictate what compromises are available both to other people and to ourselves. And we have, on the whole, made rather bad ones, ones that are deeply detrimental to the world and ones that, in consequence, we're largely living with the consequences of today.
So, when I hear a tech person say that something's "just a tool", I can't help but read it as an unwillingness to take responsibility, as an industry, for what we put into the world and for what we do to ourselves. Tools, unfortunately, simply cannot be the neutral, separate thing that so much of the tech world would like them to be, and pretending otherwise, let alone telling people affected by the tools that they're simply using them wrong, is an abdication of our moral duty to avoid, inasmuch as is possible, doing harm. We really ought to do better.
You will notice a small advert beneath this article: Arca is, in some ways, my attempt to teach technology in ways that are mindful about tools and what they do to us. Have I succeeded yet? I don't know, but I certainly believe I've made significant progress towards it. Sign up and find out!