Contextuals are formatted like this (hover over me)congratulations!, and offer small bits of context around phrasing or word use.
Hovertips are formatted like this (hover over me)$S$Quantum electrodynamic action functional $= \displaystyle\int \psi^\dagger \gamma^0(i$$\gamma^\mu$Dirac matrices$D_\mu$gauge covariant derivative$- m)\psi$ $\displaystyle-\frac{1}{4}$$\displaystyle F^{\mu\nu}$electromagnetic field tensor$\displaystyle F_{\mu\nu}\ d^4x $, and offer media elements or small bits of extraneous information.
Sidenotes are indexed as a, b, c, ..., and are visible to the right of their index.
Endnotes are indexed as 1, 2, 3, ..., which indices serve as bidirectional links to and from their location at the bottom of the page.
Extra endnotes are indexed as A, B, C, ..., and are revealed in the text by clicking the button at the bottom of the page.
Also: expandables, marginalia, addenda, leads...
Linked TOC
About Me
Identity/Contact
Mariven is a pseudonym, but as faithful and consistent a pointer to the identity of the person who created this site and wrote everything on it as you'll find anywhere elseI have a government-standardized 'real' name that goes on my government-standardized documents, but it is not something I identify with and actively use, and I don't share such information except with people who need it for government-standardized reasons (whether the government is playing the role of facilitator as in mailing/shipping or that of supervisor as in hiring/funding). If you have a good reason to want such information, contact me directly.. You can contact me at this username through Discord, (Libera) IRC, or email mariven at !c{proton.me}{(equiv. protonmail.com, pm.me; I just use proton.me as a matter of idiosyncrasy)}; the Twitter handle was taken, so there I go by @psychiel. Preferred order is Twitter = email > Discord > IRC (which I only sporadically use).
Mentality
Two things fill the mind with ever new and increasing admiration and reverence, the more frequently and persistently one's meditation deals with them:the starry sky above me and the moral law within me. Immanuel Kant's Critique of Practical Reason (tr. Pluhar), Conclusion.
My purpose is to (a) understand the world, and (b) make the world the best place it can beThis is another way of saying "do what is moral". The world is all that is the case, and the moral valence of an !c{action}{not a person, not an intent, not an ideology} depends solely on the changes to the world effected by it.. The two are bound together: to intentionally effect any given change on the world requires a model of what the world is and how it evolves, a vision of what successful change looks like in the world, and a plan to manipulate the world in a way that one's model predicts will cause it to evolve into a more successful state.
To reason about what the best world would look like has proven difficultThere are questions of !c{infinite}{or unbounded, in any case} ethics, of the criteria for moral saliency, of the ontology of the "better than" relation, and so on., but a direct answer isn't strictly necessary yet. You can only optimize the world insofar as you are empowered to take actions that optimize the world, whereas being disempowered renders questions about what actions you should take pointless. In the next decade or two, humanity will finally manage to automate optimization itself; this achievement, the creation of artificial general intelligence, will be the finale of human technological development, the final chance we have to control our fate. On our current trajectory, it seems we'll end up prevented, whether through extinction or lock-in of the current status quo or some other fate, from taking actions that significantly change the world. Thus, while my ultimate goal is simply to make the world the best place, this goal factors throughIn the category-theoretic sense: a state transition $f: A \to B_\lambda$ factors as $A \overset{f_A^X}{\to} X \overset{f^{B_\lambda}_X}{\to} B_\lambda$ with $f_A^X$ satisfying some desideratum (in this case, $A$ is the present worldstate, $\{B_\lambda\}_{\lambda \in \Lambda}$ is a family of 'good' worldstates, and $f_A^X$ is a process, or transition between worldstates, satisfying the desideratum of safe AGI having been developed). the subgoal of ensuring that AGI is safely developed and deployed.
I'm a mathematician at heart—more specifically, an algebraist. My ultimate aim when confronted with an abstract problem is to formulate a perspective in which the problem simply does not exist, rather than to crack the problem with a well-targeted strike at its core (as must be done for concrete, real-world problems, which are invariant under changes of perspective). [UNFINISHED]
Statistics
Age: Mid-20s
Alignment: Lawfully Chaotic (not Chaotically Lawful—that's just unreliability...)
Neuropsychological type: minmaxedExecutive function and charisma are my dump stats. If you gave me a year to solve quantum gravity and left me alone, I'd probably get 90% of the way in six months before deadlocking over some menial chores for the next six. I do better with light overhead management.
Myers-Briggs type: !c{INTp}{95-99% I, 80-90% N, 60-80% T, 40-70% P}
$p($doom$)$: 70%50%. Most of the $\uparrow$ comes from multiagent dynamics exacerbating peoples' inability to specify good goals and fundamental ontological advantages, most of the $\downarrow$ comes from the always-tangible possibility of weird things that are not thereby hopefule.g., a global nuclear war after which more tech-conscientious civilizations arise. This would be very bad, but not classifiable as doom.. See AI Safety is Team ∀. Update: increased credence in catastrophes neither directly related to AI nor irreversible. Much gloom leaves no room for doom.
QM interpretation: Many Worlds-esqueTo be specific, my best bet is that there's no such thing as collapse, and observation is just entanglement; further, consciousness is just a weird kind of way that physical matter sometimes gets. In fact, I'd bet that there's "no such thing as physics", just the inherent nature of the zero-variation plateau of the (presumably counting-like) action functional over the space of possible universal wavefunctions (consider how the principle of stationary action can be derived from the Feynman path integral formalism, and how the laws of physics as we know them can be derived from that principle).. Update: I find a bit more plausibility in epistemic interpretations of the quantum state nowadays, and the only QM interpretation I feel comfortable with is a sort of meta-interpretation: insofar as differing interpretations guide our perceptions of is-es and oughts in the same way—a hunch evinced by results like the decision-theoretic optimality of the Born Rule that Quantum Bayesians love to bring up—I'm fine equivocating, since I suspect that the "ultimate situation" might not be so easily conceptualizable as any existing interpretation.
Prior distribution: Boltzmann
Things I Like: optical illusions, dogs, higher-order functions, being helpful, the idea of fruit, digital personas, rhythm games, molecular biology, ravens, dependent types, self-reference, thunderstorms, public utilities, GMOs, winter, transcendental idealism, caffeine, earnestness, vanilla, nuclear power, bees, virtual reality, watermelons
Things I Dislike: complex analysis, sleeping, strongly typed languages, beaches, opioids, transcendental idealism, mushrooms (culinary), deceit, summer, warfare, onion rings, political identifications, version control, meat, gender, ideological commitments, parasites, formal dress, humidity
(Purple items are unambiguously true—at time of writing, at least).
For Cybersleuths
If you're browsing around this website trying to figure out who I am and what I/my work can offer you, or generally trying to piece together a mental image of me, consider looking at the poorly named thoughts page, which is just where I put all sorts of long-form writings that don't feel complete or thorough enough to stand under their own weight(and my home page should make clear how unreasonably high this bar is); it contains nearly a hundred thousand words covering all sorts of topics. If you're browsing my Twitter account, make sure to check out the replies tab, where most interesting, in-depth exchanges happen. And, if you have any questions or uncertainties, you can literally just ask. (If you don't want to read, or don't want me to write, several thousand words in response to a basic question about why I think X or seem to assume Y or whatever, just say as much, and I'll shorten my reply).
About this site
Logistics
If you're actually interested in the construction of anything here, just go to Inspect Element; excepting a few common libraries, I wrote the JS, (S)CSS, and, with the exception of a few macros like $!$c{abcd}{efgh} for the contextual !c{abcd}{efgh} which are regex-replaced by the aforementioned JS, write the raw HTML for each page in VS CodeThat's right, like an animal. It's what works for me, though, allowing me to create things like this sidenote in a clear, non-finicky way that WYSIWYG editors fail to provide. It also allows me to do things like turning an 800-page book into a single continuous webpage with a freely navigable table of contents. I feel like doing this in a "modern" content management system would result in a page that hangs for a whole second every time you try to scroll.. The site is served by nginx on a DigitalOcean server that costs like a few dollars a month, and the domain is managed by a blood oath with Cloudflare that costs like a few dollars a year.
Note: None of the code is written with the intention of impressing, or even being amenable to working with, other people. That means no tests, no coding conventions save my idiosyncratic preferences, little documentation save what is necessary for me and only me to understand what I was thinking earlier (let alone "self-documenting code"), and so on. It's not scalable or flexible except in the ways I personally want it to be scalable or flexible—there are definitely several smelly code patterns that any seasoned developer looking at this code would immediately notice and start fuming over, but they work for my style of thinking, for this website in particular, and that's what I care about. (Though, even then, there are some problematic issues that I want to fix, but have never really found the motivation or time to go back and rewrite my old code—so it's a palimpsest of novice oddities and later workarounds).
!tab For instance, I would never use a meta-meta-regex-parser in a production codebase, if only for the sake of whoever else has to maintain it, but I'm happy to write one here because I'm inclined towards solutions involving regex parsing and meta-meta-anything; they come naturally to me, and I'll readily explain what I'm doing to any curious person who wants to write one themselves or modify/extend mine. But if I ever have to make the functionality it performs more widely accessible (or someone pays me to do it), I know it'll have to be redesigned from scratch.
Cruxes
Logic
To decide to take one action over another is equivalent to obtaining information about the world you're in by way of how beings structured like yourself react to it; "free will" is only the ability, from one's own point of view, to decide which of a set of incompatible facts one will learn about the world. Optimal action, then, involves deciding to learn that one is in an optimal world. For instance, in the Parfit's Hitchhiker problem, the optimal world is the one in which I accept the ride and then pay the driver, so I ought to decide to learn that I'm in the world where a structure like me responds to this problem in that way, which just amounts to actually accepting the ride and then, even after I'm saved, paying the driver.
The sole criterion by which an empirical observation can be used as evidence to distinguish between possible worldstates is: "to what extent would I have made a different observation if the world were in this state?"
A famous anecdotal conversation between Wittgenstein and a friend: "Tell me, why do people always say it was natural for man to assume that the sun went around the Earth rather than that the Earth was rotating?" "Obviously because it just looks as though the Sun is going around the Earth." "Well, what would it have looked like if it had looked as though the Earth was rotating?". An alternative way to phrase this criterion is: "would I still be seeing this if the claim it supposedly evinces was false?". For instance, there are people who have spent decades compiling archaeological and geological evidence for the Biblical flood, and they can make some pretty powerful cases. The issue is that their ability to make pretty powerful cases is only slightly increased by the Biblical flood actually having happened, since it's mostly downstream of spending decades searching for evidence from such phenomenon-rich fields as archaeology and geology. You should still increase your belief in the existence of a Biblical flood upon seeing that these powerful cases exist, but only by the same factor with which the existence of such a flood multiplies the chance of your seeing such cases. (This is literally just Bayes' Theorem, but from the point of view of the internal cognitive processes that enforce it upon subjective credences).
Reality
Existence is incomprehensibly large, impossible to capture via human conceptual architectures, and radically alien to human sensibilities. Any implicit ontological assumptions we make of reality will be shattered by its strangeness, and any implicit axiological demands we make of reality will be violated by its indifference. When confronted with the possibility of our living in an Everettian multiverse, of our being Boltzmann brains, of our being in a simulation, etc., the best we can do is investigate with openness and equanimity, searching for the actual truth in the knowledge that it was always already true.
Morality
Truly moral action—whether in the sense of reflective equilibrium, coherent extrapolated volition, or any other logically-constrained framework—is generically brutal and horrifying by our present sensibilities. This is just as much a reflection on the nature of logic as it is the nature of human sensibility.
All suffering is of a single qualitative kind; all suffering is moral injury to the homogeneous Perceiver of subjective experience of whom we as individual experiential threads are projections. Everyone's pain is just as terrible as my own. In the ultimate senseAs opposed to the "conventional" sense, in which it is only ever a means to an end (e.g. it's a deterrent, it's a teacher, it's... whatever people need it to be in order to keep the blood spilling)., the infliction of suffering upon another can never be "justified" or "deserved", for it's exactly the same as if that suffering were to be inflicted upon oneself. The fact that one doesn't themselves feel the suffering they inflict is nothing but the world's greatest moral hazard.
(Maybe people will want to call this "open individualism". They can call it what they want -- but the Wiki page contrasts that label with "empty individualism", which seems to me to be another way of saying the same thing. I'll have to reject both of those -isms, then, since they're just ways to force me into fake rhetorical bins. The point is homogeneity of utility—insofar as beings ought to make distinctions between other beings, it should only be as a means to the end of (timelessly) benefiting all beings).
Evolution is a mechanism for making dust feel pain. The suffering of non-humans is qualitatively the same as that of humans, and there's no good reason to believe that it quantitatively differs by more than an order of magnitude (say, for most complex vertebrates). The natural world is a realm of unparalleled agony and terror, and this is a problem.