March 2024 BenGoldhaber.com Newsletter
this month: appreciating green and launching my job board
BenGoldhaber.com Media Appearances: The New Yorker ran a profile of the Bay Area rationalist scene. I think it is the most accurate depiction of the community I’ve read in print; it anchors the narrative around a description of Katja Grace, who quite rightly is everyone’s favorite person. I make a very brief appearance in the article:
To be clear, I think we can in fact avoid arms races!
We should build the institutions and international relationships to create trust. My colleagues in FAR have done amazing work to promote international cooperation and demonstrate the cross-country consensus among the global scientific community that powerful AI can be dangerous.
We should create formal control systems for AI that can be verified by a diverse set of stakeholders, shifting actors from a prisoner’s dilemma to a stag hunt1. Below is an example analysis from davidad of targeted R&D to expand the solution space, allowing for more cooperation and mutual benefit. Even if you don’t agree with his proposed tech path, the goal to find AI architectures that promote cooperation over competition seems obviously great!
Anyways I’d encourage reporters to reach out to me primarily for quotes about Twilight Imperium 4 and how to structure your fleet composition (build more fighters than you’d naively think!)
#links
Rank Things - Partner Attributes: Stack rank what you’re looking for in a partner. Courtesy of a newsletter reader - who not only suggested it, they created it! - the app is a dataset of 1770 different traits, where you compare each against each other; the app suggests the highest information comparisons, eventually giving you your list of desired qualities in a romantic partner.
It was pretty informative; both to think about what I want and to force me to consider how my wants sit in relation to each other. A cool potential extension would be using this to power archetypes - aka after answering ~200 or so questions you get like a personality quiz style description of your ideal partner. “It’s clear you’re incredibly vain, you should find your Pretty Pete!”.
Suno.ai: It’s now possible to make pretty good music strictly with AI. This song where a sad girl playing piano sings the MIT copyright show the wonders.
Prediction Market News: Susquehanna trading on Kalshi seems like a big deal - liquidity is the biggest limiter of contracts; if this a signal for real volume (and I am a little skeptical) could radically improve the ecosystem. Related: I’ll be attending Manifest.
On Green: Joe Carlsmith has been, and I don’t say this lightly, *cooking* with his recent blog series on Otherness and Control in the age of AGI. My favorite post so far is On Green, where he describes an attitude and mindset embodied by the Green Mana aspect of the MTG color wheel (which is a great general taxonomy):
Green is an orientation that includes a deep respect for ‘the natural order’ , ‘Yin’, attuning to the ‘Way of the Things’: clearly, it’s the color of all that spiritual shit that is hard to understand or justify through spreadsheets of utility quantification.23
I mentioned above that green is the “conservative” color. It cares about the past; about lineage, and tradition. If something life-like has survived, gnarled and battered and weathered by the Way of Things, then green often grants it more authority. It has had more harmonies with the Way of Things infused into it; and more disharmonies stripped away.
Thus, consider the idea of casually cutting down a giant, ancient redwood tree for use as lumber—lifting the chainsaw, watching the metal bite into the living bark. Green, famously, protests at this sort of thing—and I feel the pull. When I stand in front of trees like this, they do, indeed, seem to have a kind of presence and dignity; they seem importantly alive.13 And the idea of casual violation seems, indeed, repugnant.
…But it remains, I think, notably unclear exactly how to fit the ethic at stake into the sorts of moral frameworks analytic ethicists are most comfortable with—including, the sort of rights-based deontology that analytic ethicists often use to talk about liberal and/or boundary-focused ethics.
I’ve come to appreciate this attitude, though its very nature makes is hard to name, which is why I deeply appreciate Joe’s winding, poetic attempt to capture it; or at least, provide illustrative examples of how there is something here that is missed in our blue-white-black systems of understanding:
I haven’t found a single word that I think captures green, but associations include: environmentalism, tradition, spirituality, hippies, stereotypes of Native Americans, Yoda, humility, wholesomeness, health, and yin. The essay tries to bring the vibe that underlies these associations into clearer view, and to point at some ways that attempts by other colors to reconstruct green can miss parts of it.
…
Or consider experiences of wonder, sublimity, beauty, curiosity. These are all, paradigmatically, experiences of encountering or receiving something outside yourself—something that draws you in, stuns you, provokes you, overwhelms you. They are, in this sense, a type of yin. They discover something, and take joy in the discovery. Reality, in such experiences, is presented as electric and wild and alive.
And I think this joy is at least one core thing going on with green. Contra green-according-to-black, green isn’t just resigned to yin, or “serene” in the face of the Other. Green loves the Other
I find myself nodding in agreement because, as humanity grows in power and is better able to organize the world according to our explicit models of what is Good - whether because of technology like AI or virtual reality, or just further progress in organizing every human activity according to an IRS tax code, I worry that there is something that will be lost, if there’s not this Wisdom.
To the extent green is the “wisdom” color, then, I think we should be pretty interested in making sure we’re staying in touch with green-on-its-own-terms. And indeed, when I think about the sort of wise that I want future people to be, I imagine them having the attunement thing in spades—some kind of intensity and tenderness and vastness of consciousness, some deep receptivity and responsiveness. If what one learns, from attunement, is “basic sanity,” I want the future to be sane.
so, all this to say, I’m sad to report that I now think the Lorax was onto something.
BenGoldhaber.com Job Board: Speaking of the wisdom of age, I’m becoming more like my Dad in that I’m up in everyone’s business and finding I enjoy playing corporate matchmaker.
A close friend of mine who has built a very profitable b2b SaaS business wants to hire a VP of Sales/CSO to supercharge the operation. I love the guy, and do think if you’re a Glengarry Glenn Ross aficionado this could be a good opportunity.
Also, I often hear about opportunities in the AI Alignment space, and now think I’ve got a reasonable sense of what organizations a person might want to join, and/or how to plummet down the rabbit hole in general4, and I’m always happy to chat with people about this.
#good-content
Dune: I haven’t seen Dune 2, but I finally watched the first Dune. Good!
Song of the month: Red Line.
xoxo,
Ben
Memeworthy: One of the great Milhouse line deliveries.
Also, I think it should be more well known that prisoner’s dilemma is solvable with read access to the other players ‘source code’, which could apply to both AIs and and maybe human/AI systems?
I sometimes feel like I’m reading and writing for 25 year old Ben, trying to explain what I think he missed in his worship of How to Measure Anything.
I recognize the irony of saying this while right before it including a link to rank what you want in a partner. something something man contains multitudes.
if that seems right to do on its own term and for you, not because someone tells you to, resist the coercive forces of order and white mana!