Incompetence or Malice?

Against the exclusive or; against “Hanlon’s Razor”; for “Socrates’s Solvent” and for “Augustine’s Glue”

Welcome to the Blackthorn Hedge. This letter begins a new arc of 24 letters, Arc 3, and it should stand alone without prior familiarity with my earlier letters; it keeps references light until the end. If you would like a summary of what I’ve just wrapped up, please read last week’s “Laying the Hedge: Arc 2.”

As always, I welcome all respectful comments, messages, and other engagement.

---

Things go wrong, and there are people to blame for it. Someone has messed up the birthday cake on MY BIRTHDAY. Someone has let the cat out and now who knows if it’ll come home. Someone didn’t fill the tank and now the car is out of gas. Someone’s argument on the Internet is incorrect. Someone has been killing civilians under cover of war again. But how should they be blamed? Should they be blamed as if they meant to do wrong? Should they be blamed as if it was an honest mistake?

The dichotomy here, between honest failure and intentional harm, is too simple. And that has become a problem.

There are some things that are best explained by malice, and there are other things that are best explained by stupidity. Sometimes they do split. And if you’re a repairman or technician, the adage “never explain by malice what could be explained by stupidity,” called “Hanlon’s Razor,” can make sense. People really can be surprisingly stupid about taking care of their homes, cars, computers, plumbing, etc. It can also make sense if you’re training green employees, rookie colleagues: new people do honestly find mind-boggling numbers of ways to mess up new tasks before they have the hang of them.

In the general case, however, both the split and Hanlon’s Razor are absurd. First off, there are the categories of malicious stupidity and stupid malice. Malicious stupidity should be familiar to anyone from grade school – acting dumb to get a rise out of a teacher is something we’ve all seen. Stupid malice comes up just as early: cartoons and kids’ stories are full of dumb bullies who are intent on harming others but instead harm themselves.

You need both malice and stupidity to explain either. And you can keep nesting: there’s stupid malicious stupidity when a kid plays dumb to bait the teacher but the teacher outplays the kid; there’s maliciously stupid malice when an underling thug deliberately screws up his orders to hurt someone from a boss he likes even less than the intended victim.

Nesting is only one of the combinations of malice and stupidity. Another common mixture is simple redundancy: a person acts with both malice and stupidity at the same time. Both contribute to harm at the same time. More complex mixtures form historically: being asked to do things you can’t over and over, made to fail again and again, breeds resentment; that creates malice. Thereby simple incompetence develops an admixture of malice. Similarly, when people shun an adolescent’s company for acting maliciously, the shunned person loses opportunities to develop social skills, so they can fall behind and become socially incompetent, seeming stupid ever after.

The most famous philosopher of the West, in fact, argued that there was no malice at all without stupidity within it: “there is only one good, knowledge, and one evil, ignorance.” Against Hanlon’s Razor, “never attribute to malice what could be attributed to stupidity,” we could set Socrates’s Solvent: “attribute every malice to stupidity.”

At first this might seem to be an even more extreme way of denying malice: “nothing is truly malicious, just ignorant of what’s good.” That’s not implied, though: there might still be true malice, just formed from ignorance, just like there are truly rocks formed of atoms. Malice would not be fundamental, but it would still be real. This would ruin Hanlon’s Razor: since all malice can be explained by incompetence but malice still has an explanatory role, the adage would be nonsense. It’d be like a sophomoric “the rock I threw didn’t hit your head, a collection of atoms shaped like a rock I threw hit atoms shaped like your head.”

Socrates’s Solvent dissolves all malice into its simpler constituent ignorances like a geologist’s reference acids dissolve rocks into atoms. The geologist is not proving that rocks don’t exist. It can be a fun game to play a few times, pretending that only parts “really” exist and not composites, but it’s not more than a game.¹

In fact some pedants do maliciously abuse Hanlon’s Razor in exactly this manner – they find ways to explain their malices by their ignorances of the good as if that eliminated the malice rather than just confirming it. “I didn’t mean to neglect our anniversary, I just genuinely forgot” misses the point that the forgetting is the neglect in itself and “not meaning to” compounds the first neglect with another layer of neglect of intention rather than excusing the first neglect of memory.

Nevertheless, in social settings where moral thought is shallow, people can often get away with the trick.

A cub scout uses hydrochloric acid to test if a rock will dissolve, via Wikimedia Commons

“We didn’t know our drug would have so many side effects, we just couldn’t delay sales to test it” would never fly with the Food and Drug Administration,² but its equivalents for political projects and tech companies are pervasive. For the social media companies, at least, it’s good to see internal emails coming out in public to show there were definitely-malicious cover-ups of harms on top of the probably-malicious neglect of testing or monitoring for harms. The more public knowledge, the better people can protect themselves.

One of the ways that they’ll try to protect themselves is by federal regulation. Activist projects are organizing support for various social media legal reform programs and their competition as I write, and some have my support. Unfortunately, a lot of that is going to look like malice to tech libertarians, who will find in any new regulations both (a) do-gooder idealism that’s naive to the point of malicious neglect and (b) cynical self-enrichments of bureaucrats at the cost of economic progress that benefits all. In what these anti-regulatory libertarians will see as good faith but many pro-regulatory activists will see as a malicious naïveté of its own kind, they’ll campaign against the new regulation.

It’ll be a mess, and that’s typical. One of the main appeals of Hanlon’s Razor is not that it’s true, not even that it’s a good guide – it’s said as a joke – but simply that it’s a quick way to cut oneself free from thinking about the messes. Taking true inventory of all the complicated back and forth of mixed malice, ignorance, and incompetence gets exhausting. Is it malice to retaliate honorably against malice? Well, what if they’re stupid enough at recognizing malice that what they think is retaliation is actually an unprovoked first strike based on nothing more than a projection of malice? And when is that projection of malice itself malicious (e.g., in anti-semitism) versus simply dumb (e.g., in an espionage farce)?

Not only is processing these messes exhausting, it’s also often corrupting. For various different and distinctive reasons, people who spend their time thinking about malice often become more malicious themselves. Some become convinced that malice is simply the way of the world and they must adapt. Some become convinced that malice is the way of everyone but themselves and they must purify the others with fire or sword. Most often, thinking of specific acts of malice inspires specific malicious responses – the effect doesn’t happen on the level of “thinking about malice in general,” but rather on a specific level of some immediate retaliatory logic; the person newly drawn to new malice may not even realize how badly they’re acting until well after the deeds are done (another typical case of how malice and ignorance mix).

Processing the mess of public interaction into clean understanding is a dangerous challenge. It often requires long rumination, a word and metaphor I’ve written about in my prior Hedge letter “Digesting Tradition.” For many animals, the answer to the problem of “grass being difficult to digest” is “don’t bother eating it” – be a wolf or a fox or an eagle; let others deal with the hard stuff in its original forms and then prey on their finished work. Similarly, for many people, the obvious answer to the difficulty of digesting the mess of malice and incompetence is simply to avoid doing it! Leave it to the courts, leave it to preachers, leave it to journalists, leave it to artists.

This isn’t unreasonable or malicious in itself, though it does depend on those complements as completely as the carnivores depend on herbivores. Leaving it to journalists and artists but then refusing to pay for journalism or art, for instance, won’t work for long. Still, not everyone needs to be processing these messes directly and Hanlon’s Razor has become a convenient way to signal solidarity with the groups that don’t. If you use that Razor, you mark your place in the social system effectively.

Herbivore does not imply “easy prey,” via Wikimedia Commons

In criticizing Hanlon’s Razor, I mark my place as well – not that it was ever unclear that I’d be among the ruminants, in that metaphor above! I don’t use the Razor, I often oppose its use, and I often avoid social settings where it is common.³ Instead, I use an exactly opposite heuristic: “never explain by pure malice or pure stupidity what can be explained by an incoherent mixture of both.”

In part, this is based on my own experience in tech support, supposedly an ideal setting for Hanlon’s Razor, as a teenage summer job. Yes, users wouldn’t intentionally infect their work laptops with viruses just to make my life harder, that was never pure malice at me, but it was the case they’d most often infect them by looking for pornography that they were unmistakeably warned not to browse, and thus an indirect malice played a role. Similarly, if a computer came in soaked with tea, it wasn’t likely their own malice that spilled the tea, but it often was their cat’s! Or, actually more often, they’d have spilled it while frazzled because an angry superior or an inconsiderate deadline was putting more stress on them than they could handle.

In larger part, my heuristic is based on my experience as a chemist. If someone gives you some random powder or some pond water and they ask you what’s in it, the answer is “I don’t know, probably a lot of things.” If someone asked “is it made of negative ions or positive ions” you’d laugh because any ordinary substance would need either both or neither. The idea that pure types would be more natural and more common than mixtures makes little sense chemically, and that’s also been my experience morally and psychologically.

There’s an appropriate and well-known quote from a grave Russian regarding the moral mixture of good and evil,

Gradually it was disclosed to me that the line separating good and evil passes not through states, nor between classes, nor between political parties either – but right through every human heart – and through all human hearts. This line shifts. Inside us it oscillates with the years. And even within the hearts overwhelmed with evil, one small bridgehead of good is retained. And even in the best of all hearts, there remains…an un-uprooted small corner of evil. Since then I have come to understand the truth of all the religions on the world. They struggle with the evil inside a human being (inside every human being). (Aleksandr Solzhenitsyn, The Gulag Archipelago, Part IV, Chapter 1, via WikiQuote)

and personally, I find this rings true. It’s a good expression of a long Christian moral tradition that I seek out and admire. The problem of differentiating malice and stupidity is a problem of the finer structure of this line between good and evil, and my heuristic could be re-expressed to parallel Solzhenitsyn’s:

the line separating malice and stupidity passes not through acts, nor between thoughts, nor between plans either – but right through every human intention – and through all human intentions. This line shifts. Inside us it oscillates with the minutes.

I do think it moves much more quickly, on the scale of minutes (sometimes even seconds) rather than years; intentions change much more rapidly than the core of character symbolized by the heart. Malice can be very frivolous and flighty, even campy.

A canonical depiction of these sometimes-rapid changes of ignorance, malice, and failure is the pre-Medieval autobiographical classic Confessions by St. Augustine of Hippo, and his The City of God further provides a deeply influential philosophical discussion of how incapacity accounts for malice and vice versa.

In his Christian understanding, ignorance and sin are inseparably bound to each other, as in this passage which I’m sure will strike many contemporary readers as alien, and which they should feel welcome to skim lightly, but which is typical of past expression:

Let no one, therefore, look for an efficient cause of the evil will; for it is not efficient, but deficient, as the will itself is not an effecting of something, but a defect. For defection from that which supremely is, to that which has less of being—this is to begin to have an evil will. Now, to seek to discover the causes of these defections—causes, as I have said, not efficient, but deficient—is as if someone sought to see darkness, or hear silence. Yet, both of these are known by us, and the former by means of the eye, the latter only by the ear; but not by their positive actuality, but by their want of it. Let no one, then, seek to know from me what I know that I do not know; unless he perhaps wishes to learn to be ignorant of that of which all we know is, that it cannot be known.⁴ (Augustine, City of God, translated by Marcus Dods in the Modern Library series (1950), Book 12, Chapter 7)

In some sense, for Augustine all ill will is lack of will to know or love God, all lack of will to know or love God is or is due to lack of knowledge of God, and all lack of knowledge of God is or is due to lack of will to know and love God. They’re inseparable deficiencies: after applying Socrates’s Solvent to any foreground malice to dissolve it into fundamental harmful ignorances, he finds every fundamental harmful ignorance also still implying a fundamental ill will, i.e., another malice, and vice versa.

This inseparability and inscrutability (“that of which all we know is, that it cannot be known”) are the essential features of my “never explain by pure malice or pure stupidity what can be explained by an incoherent mixture of both,” and so I call the heuristic Augustine’s Glue: it keeps ignorance, other incompetence, and malice stuck together where Hanlon’s Razor would claim to slice them apart. It does not depend on God directly, just a morality in which lacking and neglecting knowledge of the highest good implies some form of ill will.

It’s a funny heuristic to live with. It doesn’t allow for self-satisfaction as a totally good-hearted person without any malice or a fully intelligent and knowing person, and it doesn’t allow for the shortcut of treating others’ malice as fully self-aware, deliberate, and intentional and thus deserving of no mercy. It can grow demoralizing without compatible camaraderie or faith. Nonetheless, I think it’s actually true, not just “directionally useful in context” like so many popular heuristics, and I think that’s worth it.

---

With that, I’ve made my points and will set this down for later reference. I hope that it has been interesting as a stand-alone, and I hope that readers who have been keeping up with prior arcs were seeing how this fits in and builds on prior themes. I’ll close with a couple of points about why I prioritized this letter now.

First, one of the key near-term reasons this is important to me is that I think these simple common-sense, jokey-but-serious heuristics like Hanlon’s Razor are often surprisingly crucial for AI alignment debates, AI regulation debates, and also for personal safety using current-generation machine learning. For instance there is also a dark mirror of Hanlon’s razor, the mastermind fallacy, in which people frustrated or demoralized into thinking that goodness is weak begin to conflate evil with competence and intelligence – and then choose evil. This has been a central preoccupation of LessWrong and Effective Altruist conversation, for instance a primary theme of Yudkowsky’s epic movement recruitment fanfiction Harry Potter and the Methods of Rationality. This is a step towards participating in those conversations at the level of the actually effective rhetoric.

Second, in this arc of letters I’m planning to write more about fascism, and talk on that topic tends to get badly tangled on simple misconceptions about the separability of ignorance and intent if they’re not prepared for ahead of time. Many political allies of fascism do not know that is what they are until too late, for instance – and same with tyrannical pseudocommunism. Politics is often cutthroat and opaque, so this is hardly a unique property, but it’s especially salient and unavoidable in discussions of fascism. It takes full advantage of stupidity for the sake of its malice; its leaders deliberately cultivate weakly and equivocally malicious stupidity to increase the reach of their own determined, unequivocal malice. I’ll want this as a link to send people wondering about my deeper beliefs on these points when they read those upcoming letters.

Finally, this fact of pervasive and forgivable malice is, for me, a key piece of the background for why and how to understand Jürgen Habermas’s peculiarly exact claims about communicative action and deliberative politics. His jargon like “collective opinion and will formation” often throws people, and I think one reason it does is that he’s trying very hard not to poke the public in one of its blind eyes. Before I’m ready to explain exactly what I mean there, I’ll write another letter on blind public eyes using the metaphor of “social shadow,” which is not exactly either Jungian Shadow or the unintelligible deficient evil that Augustine could not know, like darkness, in the quote above, but it is a close relative to both.

Thank you for reading. As always, I welcome all respectful comments, messages, and other engagement. If you find yourself using Socrates’s Solvent or Augustine’s Glue when thinking or speaking, I’ll be curious how they work for you!

1. After all, they’re really just quarks and gluons shaped like nuclei surrounded by electrons rather than atoms, and there’s no bottom to that (superstrings in quark-looking excited states, spacetime foams with superstring-like defect states…), just renormalization theories telling us how to proceed without a bottom.

2. Not that companies don’t try to push the line on cases like experimental therapies for urgent need, “we couldn’t delay saving lives to test it” – that line between “saving lives” and “making sales” is actively contested for good, valid reasons.

3. It’s an “emetic” to me and has a bad “smell,” in language I introduced in earlier letters.

4. Note Augustine’s method of “teaching what he doesn’t know,” here; this is in harmony with Rancière’s The Ignorant Schoolmaster, and I’ll return to that in a future letter.