by Malcolm Murray

Oops. During most of 2024, all the talk was of deep learning hitting a wall. There were secret rumors coming out of OpenAI and Anthropic that their latest training runs were disappointing. People were confidently stating that AI progress had now hit a plateau. Importantly, this was not just a few pundits that have bet their career on the current AI path being the wrong one, such as the Gary Marcuses of the world. It was all of tech and even mass media jumping on AI-hitting-a-wall meme.

And then what happened just at the end of the year? OpenAI announced its o3 model. A new model, with an unknown architecture, that achieves benchmark scores hitherto unimaginable. o3 achieves 88% on the ARC-AGI challenge, a benchmark designed to measure the efficiency of AI on novel tasks. o3 gets past 2,700 on Codeforces, making it comparable to the best developers in the world. Maybe most impressively, o3 gets 25% on FrontierMath, a benchmark created by Epoch AI just months earlier with the most fiendishly difficult math questions, where other AI models have literally had 0% correct (except o1, which had 2%).

The point of this post is not to argue that these are mind-blowing scores and that AGI has been achieved, as some were quick to do. There are always question marks in how the scores were achieved. For example, some people question whether having the model trained on some of the public ARC-AGI training set unduly influenced the results. Or perhaps, it only solved the easier questions in FrontierMath. But that is beside the point. The bigger point here is that, when it comes to AI evolution, prediction seems dead. As noted by Zvi Mowshowitz in his post on o3, no one saw this coming. Outside of OpenAI, the whole tech industry seemed all bought in on the AI has hit a wall meme. So what this means is that we need to stop focusing on trying to predict AI capabilities and instead focus on building resilience to AI risk. Rather than try to determine exactly when we will get specific AI capabilities, we need to start preparing society for its effects, so that the impact can be mitigated. Read more »

Crystal-Bison

out of coolblue
a bison

out of bitter cold
out of stratospheric skyblue
a crystal bison

from stillness, sheer,
mute as speech in utter space
from deadstill blueice
steps a crystal, a bison
apparition

a bison frozen in veneer of chilled time
cloven hooves to heaped hump
its mere eyes, its cryo-olfactory nose
smelldeaf due to rime

a crystalized bison steps forth
deliberate articulating jewel
hung round the neck of stilled rhyme
speechless

a graceful parentheses of horns
frames a frozen skull
each as sharp as the point of
breathingeverything
………… —intent: survive

Jim Culleny
1/29/19

Photograph by Tom Murphy

Enjoying the content on 3QD? Help keep us going by donating now.

by Jeroen Bouterse

In 2023, I wrote what I reckon was a calm, analytical column for this website, about how my country had talked itself into giving the xenophobic, far-right ‘Freedom Party’ a strong plurality of seats in Parliament. An equally level-headed update seems warranted, as its leader Geert Wilders has since maneuvered his party into leading the Netherlands’ coalition government, which we have now been able to observe in action for six months.

I spent a healthy share of that time, possibly even more than that, yelling at the daily news. Here we see the beauty of the written word, however: the process of creating a text provides an excellent opportunity to take a step back and find a broader, more generous perspective on things. Thanks to the alchemy of prose composition, my temperament and my primary emotional responses do not need to prevent me from giving you, the deserving reader, a balanced and distilled account of the state of politics in the Netherlands, and of the nihilistic bastards that currently dominate it. Indeed, I would hate for this to be just a longer version of the rants I post on unfashionable social media platforms, for the benefit of fewer and fewer friends. Be assured, then, that sublimated ideas and broadly applicable wisdom lie ahead, and not just expressions of rage and frustration. I am almost certain of it.

The phenomenon this essay will examine is that of the reluctant enabler. Like a group of youthful arsonists, the political burn-it-all-down faction tends to consist not only of Steve Bannon-level pyromaniacs, but also of impressionable, frightened, or calculating joiners. Usually, they at least pretend to be on board with the whole ‘setting fire to democratic institutions’ project. Rare is the teenager-slash-politician that actively carries kindling and matches to the target structure, all the while professing loudly that open fires are very dangerous and should be avoided at all times; adding, moreover, that they have never liked their friends, and cannot wait for the fire department to put a stop to this ill-fated enterprise. But that’s coalition government for you. Read more »

by David Winner

Angela, my wife, and I are over-educated middle-class white New Yorkers of a certain age.

My six months on Avenue D gave me a leg up in my search for further teaching positions around New York.  I was now deemed capable of teaching non-white students because I had taught non-white students before as if the teaching of non-white students were a skill unto itself.

I don’t know how cynical to be about such a formulation, but given the rowdy students of color depicted in so many movies and TV shows, it’s hard not to imagine that non-white students are somehow considered more difficult, their white instructors more trainers than teachers.  For the record – please forgive the invidious comparison – my biggest classroom struggles have been with white students, the kid at NYU who wrote “David Winner is an asshole” on his student evaluations.

Non-white classrooms have often been identified in job descriptions by peculiar racial dog-whistling, terms such as “population” (as if white people somehow aren’t that), “community” (same thing), “inner city,” “urban,” as if no white people live in cities.  “Diverse” and “multicultural” often just mean non-white rather than a confluence of different cultures.

Are all non-white students the same?  Can I be a “Black expert” and say, Judy, my fellow white person, be a “Latinx expert” and, Frank, an “Asian and Pacific Islander expert?” Graduate education in both the fields of ESL (English as a Second Language) and Social Work has sometimes answered these questions by the encouraging of stereotypes. On a hiring committee for an ESL professorship, I once listened to recently minted job candidates explaining to us about Chinese, Egyptian, Dominican people using hair-raisingly broad labels whereas a friend (white non-Hispanic like myself) “learned” in NYU’s Masters in Social Work program that she needed to be careful while working with Latinas because they might have an attack of “ataque,” a sort of loud hysterical fit that women from all over Latin America were deemed to be prone to. Read more »

by Terese Svoboda

My brother would roll his eyes back, shout applesauce or give me your hair, and fall to the ground in front of, say, the cheerleading squad, only to return to normal, dazed and confused, pale and clammy, with a big blank in his brain five minutes later. One in a hundred Americans suffer from incurable epilepsy, one-third of those untreatable. None of the many drugs prescribed for my brother’s epilepsy consistently prevented his seizures. The medication was never quite right; he was growing and his chemistry constantly changing, along with the pharmacology. Despite the disease’s very public manifestations, my parents’ approach wasn’t very helpful: if you don’t acknowledge the problem, it doesn’t exist. While my father wrestled my brother away before he fell to the floor and “things got out of hand,” we were supposed to act both surprised and unconcerned.

We tried. In high school, my tack was to pretend he didn’t exist. My sister, twenty months younger, remembers me ignoring her in high school too, so maybe it wasn’t so personal. My brother had only one friend, a fellow acne-pocked boy who barrel-raced with him in rodeos, and wasn’t too cruel. Suga, my brother’s quarter horse, wound around the obstacles with equine annoyance, and frequently tried to toss my brother from the saddle. Even Sugar knew. By not acknowledging my brother’s disease, or at least telling his siblings what to do when he was thrashing around and foaming at the mouth, my parents left us terrorized by the possibility of his collapse, afraid of him and for him. We weren’t even told to prevent him from swallowing his tongue, although now I think that doesn’t happen, it’s just another old wives tale people like to tell witnesses in the helplessness of a seizure. That we pretended he was just fine branded us as collaborators in our small town – he might as well have been contagious. The cost of epilepsy in a family is always more than just the patient’s suffering. I, for one, felt angry not being able to do anything, guilty that I avoided him, embarrassed that he behaved so weirdly, and perversely envious that he received so much attention from my parents. Read more »

by Eric Feigenbaum

My great-grandparents were among the 12 million immigrants who passed through Ellis Island and equally a part of the wave of 20 million immigrants who entered the United States between 1880 and 1920. America’s fast-growing economy needed more manpower than its existing population had available, and the poorer classes of Europe were the beneficiaries including four million Italians (largely southern) and two million Jews.

In many ways America was a pioneer in pioneering. In the 19^th Century the United States ambitious sought to expand across a continent while developing a robust industrial economy. As the country entered the Industrial Age, it’s 1870 population of roughly 40 million just wouldn’t suffice. Therefore, America became the first Western country to undertake not just a loose immigration policy, but one at large scale – with the sheer audacity to believe it would grow its economy at unprecedented rates.

Not only did the gamble work for America, but it fundamentally changed American culture. Sure, by nature of being colonized and then in many parts staffed with abducted Africans forced into slavery, the United States began as a land of immigrants. But the 1880-1920 wave pushed the country into the Melting Pot we know today. Somehow, America decided anyone could become American if they were willing to assimilate and in turn, the country’s culture became deft at integrating arrivals.

This was not lost on the founders of Singapore. In 1921, a Republican Congress and President slammed the doors on forty years of open immigration. More than sixty years later, on the other side of the world, the government of a new, small post-colonial country grappled with how to power their economy as they moved from third world to first.

Lee Kuan Yew, the founding and longtime Prime Minister of Singapore described America as “a society that attracts talent from around the world and assimilates them comfortably as Americans.” Read more »

by Rafaël Newman

An empire, threatened on its flank, vents spleen
Upon the would-be sovereign state between
Its borders and a surly host beyond—
Which wavers not with weapons to respond;
The fray’s then joined by nations treaty bound
To emperors (one more, one less self-crowned):
And thus the world comes closer to a war
That will destroy both empires, and much more!

Now, do you think I mean our present plane,
Where Russia, versus NATO in Ukraine,
Once freed of its remaining moral shackle,
May trigger yet our terminal debacle?
Mistaken. I recall instead a time
Long past: which does, however, with ours rhyme.

It’s 1859 and Lombardy,
To Habsburg lately sworn in fealty,
Raring Risorgiment-ally to rise
Italian and autonomous, allies
Itself with Piedmont and Cavour, il Conte—
And with Napoleon the Third (less jaunty
Than the First, though uncle and his nephew both
To none but to themselves will pledge their troth).

These grandees fling their gauntlets at the feet
Of Francis Joseph, on whose balance sheet
The loss of Lombardy were scarce redressed
By assets in Galizien, or Budapest;
And who is thus obliged to quit his court
For Solferino, where the kingly sport
Of war awaits him, and his regents rival,
To prove which prince is worthy of survival.

But here’s the place where parallels part ways
And leave us to our later, lesser days.
For though the Kremlin calques the Kaiser and
The Pentagon’s as potent, if no wiser than
L. Bonaparte, while Brussels is the farce
To V.E.R.D.I.’s tragedy: yet try to parse
The rest—the battle proper, where each chief
Championed in person his own proper fief;
The last engagement on this bloody ball
Whose each host heeded its own monarch’s call
To arms; a skirmish where the issue was decided
In combat riskily by mounted royals guided—
And you’ll agree our progress technological
Comes at the cost of glory demagogical.

For where’s the majesty in leading men
By proxy into war, and out again?
What is a Situation Room compared
With Solferino, where Franz Joseph dared
To meet head on Victor Emmanuel
And Bonaparte—although it sound the knell,
If distant, for his coupling Kaisers’ clan
And for their Holy Roman Bantustans?
Where are such doughty captains in our day,
Who blanch not at the fracas and the fray
But swing their booted form upon a steed
And with a battle-cry their soldiers lead?

Or better yet—for valor in a crowd
Nor shines as bright, nor halloos half as loud
As on its own—let’s send our current masters
In single combat off to their disasters:
Let Vladimir Vladimirovich meet
Mark Rutte in a dead and final heat—
The latter, if he likes, with von der Leyen
For balance, and for bank, and local buy-in;
As for the former, should he need a boon
Companion, he can cozen Kim Jong Un.
Let Netanyahu call out Khamenei
(With Assad seconding, by Zoom relay);
Let Trump lead Xi and Trudeau, Musk and Vance
A merry MAGA trade-war Totentanz…
But give them, like Montcalm and Wolfe, the grace
To expunge each other from the human race
(Or Polynices and Eteocles:
Plus we would grant a grave to both of these)…
And leave their peoples, quondam cannon fodder,
To enjoy the mêlée as intact applauders.

Or simply: to ignore the spectacle,
And sport themselves instead in dialectical
Disputes, in theater and thaumaturgy:
Nor any sort that calls for cross and clergy,
But rather wonders of the worldly kind,
Which calm the body and excite the mind;
Avoid all mongering, of hate or war—
Just maybe fish, to please the carnivore.
Leave them, who’ve borne their admirals’ abuses,
To guidance rather by nine gracious Muses;
Let them be spared the dreadful death-bound draft,
And turn their hand instead to vital craft.
And thus my prayer: deliver us, dear readers,
Out of the hands of all these bloody leaders!
_______________________________

In memory of journalists and media workers killed in Gaza

Tree in winter in Franzensfeste, South Tyrol.

Enjoying the content on 3QD? Help keep us going by donating now.

by William Benzon

On the one hand, nothing has changed since August 2020, when I wrote GPT-3: Waterloo or Rubicon? Here be Dragons. I argued that, yes, GPT-3 marks a major technological breakthrough, one that may transform the way we live. But this technology is not sufficient. It is not deep enough. If we’re not careful, we’re going to crash and burn, like machine translation did in the mid-1960s, and like classical AI did in the mid-1980s (the so-called AI Winter).

I still believe that, though the tech industry seems to have decided that we pretty much know what we need. We just need more of it. Lots more.

It’s not as though nothing has happened since 2020. ChatGPT blew us all away late in 2022. We’ve now got specialized engines for protein folding, predicting the weather 15 years out, and who knows what else. Machine learning won two Nobel Prizes in 2024, Geoffrey Hinton in physics and David Baker, Demis Hassabis, and John Jumper in chemistry. But the underlying technology is the same as it was four years ago. We’ve just become more adept at exploiting it.

In another sense, however, everything has changed. And least for me. When ChatGPT came out I dove in feet first, using it to analyze Steven Spielberg’s Jaws using the ideas of Rene Girard. I initiated a systematic research program designed to elicit clues about what’s going on under the hood, and I reported some initial results here in 3 Quarks Daily. I also laughed myself silly doing some crazy sh*t, like jamming with ChatGPT about the Jolly Green Giant and a cast of 1000s – well, maybe not 1000s, but you get the idea. I worked like the dickens and had fun.

Then in late November, 2024, I decided I needed to try a new chatbot. I’d heard that the cool kids liked Anthropic’s Claude so I decided to give it a try. First, I verified some of the work I’d done with ChatGPT. Then I decided to load Claude with the complete text of Heart of Darkness, from ancient Rome through London to Belgium over the Atlantic up and down the Congo back to London ending with the Buddha. In no time it was able to produce a good summary of the text. Sweet! Just before he dies, Kurtz (one of two central characters) utters: “My Intended, my ivory, my station, my river, my…” I asked Claude to explain how that one utterance linked to the whole story, all 39K words. No problemo. We… chatted? talked? interacted? What’s the word? Whatever the word is, Claude and I did it – Europe, ivory, Africa, death & murder, women, FORM, we talked about it all.

I was treating Claude as an intellectual partner, a junior partner to be sure – I took the lead, but a partner nonetheless, a partner who had “read,” for some strange sense of that word, far more than I ever had or would, and could bring it all to bear on this, this what? Conversation? Interaction? Dialog? Confab? We were in this together.

THAT’s what changed, and it changed everything. The fundamental technology is the same as it was in 2020. Bigger, faster, slicker, but under the hood, the same. However, I now had a new kind of relationship with it.

Whenever I’m working on something I’ll hold an imaginary dialog in my head. If it’s a technical issue, my interlocutor might be my old teacher, David Hays, or maybe Tim Perper. If I’m struggling with another piece for 3QD I’ll conjure up a generalized reader of 3QD. Since this conjuring is happening inside my head, sometimes we’ll speak in code rather than spelling things out. When Claude is my interlocutor, everything has to be spelled out. Sometimes Claude exhausts me. But we make it work. Read more »

by Brooks Riley

Enjoying the content on 3QD? Help keep us going by donating now.

by Monte Davis

What’s a blind spot? Physically, it might be an area obscured behind a clump of trees in an otherwise open landscape. Or a gap between the views in the side and rear-view mirrors. Lines of sight cut short, or nonexistent.

Metaphorically, it gains new connotation: a blind spot is something you don’t perceive or understand — and you are unaware of not perceiving or grasping it (while presumably the one who points that out does so.) So: no there there.

Anatomically, it’s the optic disc – the spot on each retina where neurons with news from all the light-sensitive rods and cones of the retina converge into the optic nerves. The optic disc itself, first described in 1666, has no sensors. So perceptually, your blind spots are two portions of the visual field that are inaccessible to vision. Even if you learned about them long ago, try this this simple, clever demo. (The link is a PDF file, and works poorly on a smartphone screen: view it on a desktop or laptop screen, or print it out.)

Are the blind spots big? SWAG it: your field of vision is roughly 210 degrees left to right, 150 degrees top to bottom. The blind spots are ovals 7-8 degrees wide and 5-6 degrees high, making their area less than one percent of the field’s total area. On the other hand, the full moon is just one-half degree across. So those two no-go zones, the blind spots you’re unaware of very nearly all your life, are each about a hundred times the area of the full moon. Are the blind spots small?

As a magazine science writer in 1978, I had an inch-deep acquaintance with neuroscience when I attended a talk by David Marr at MIT. At 32, Marr had been a rising star since his Cambridge doctoral thesis, A General Theory for Cerebral Cortex. (Yes, it was just as ambitious as that sounds.) Trained in mathematics as well as neurophysiology, Marr developed a schema for questions to be answered at different levels in any theory of an information-processing task. At MIT he concentrated on vision, and his talk was electric with ideas.

In an interview afterward, I asked Marr what he’d like to know about our non-perception of the blind spots. He replied that he hadn’t given it much thought, but the answer would probably involve interpolation from the surroundings of the spots, and integration over time as our gaze moves. Then he brightened:

“But that would be relatively simple sleight of hand. The real trick is keeping you unaware of just how narrow your macular and foveal vison is, and how sketchy all the rest is. I’d love to understand that!” Read more »

by Shadab Zeest Hashmi

You arrive in Makkah, a stranger. The skyline, a hotel-scape, all cold commerce, won’t let you forget it. But aren’t you here to come closest to your heart, finally, to behold the Ka’ba whose cosmic coordinates you would easily recognize, even with eyes closed, led by a primordial longing? Your eyes are closed this instance as you walk towards it; you are sublimely solitary and sublimely inseparable, conjoined with every manifestation of reality, all at once. The crowds around you are immense, as is the crowding of your consciousness, a jostling in every dimension. It takes some work to access the Makkah of the heart. There is an escalator and a few stairs you must descend, and yes, there is a hand to hold as you step down with your eyes closed. You are grateful for this hand, a new hand, but it belongs to no stranger. You trust your newly found teacher S. Sadiyya’s hand the way you earlier trusted S. Hamza’s words about sighting the Ka’ba with the energy of prayer alone at first, eyes shut. There is a verse by the poet Hafez e Shirazi about dipping your prayer rug in wine if your teacher says so, but you, educated in the West, accustomed to calibrating cultural distances, have not come to trust teachers without question. These teachers, however, share the culture of your very heart, guided and guiding with both ilm (knowledge) and ishq (love) in beautifully balanced measures.

What you are about to witness with your eyes for the first time, is the magnet whose force was your first turning as a newborn in your father’s arms when he made the Adhan/Azaan in your ear, also the direction you turned, or sought to turn to five times a day, every day. Today you embody that compass. Labbayk, “Here I am,” the compass says, “I answer your call.” When you open your eyes, the Ka’ba appears as magnificence and mercy, a simple cube structure covered in hand-embroidered black silken cloth; the words that appear on it are more than two-dimensional letters, they carry Divine revelation, “ayaat,” signifying each realm of creation. You are one among hundreds of thousands of roving bodies at any given time here, but you feel the outward stirrings cease. For a moment the air becomes cool and there is a palpable stillness and silence; could it be a breath of “Sakinah,” the quality of peace holding both tranquility and habitation— the peace you enter as an abode, “Maskan,” the peace that envelops every fiber of your being. It feels, without doubt, like your first homecoming. All other homes and arrivals were illusions.

The first calls to prayer made from the noble sanctuary itself flood you with emotion; you recall the calls past, every day, every season, at times live from a nearby mosque, mostly a recorded voice under the roof of your exile. You have brought a collective ache here; the ache of a dispossessed people, haunted by war, genocide, displacement. You, who are all ache, are now becoming all prayer. Read more »

Sughra Raza. Rorschach Landscape, Guilin, China, January 2020.

Digital photograph.

Enjoying the content on 3QD? Help keep us going by donating now.

The Music of What Happens

The music of what happens
begins with the bottom line of drums
—percussion, the thumps of
bass in sync with a wind of horns,
baritone, bassoon, and in the
whispers of brushed snares, in the

round tones of tympany, and, in the
rests; the spaces between, those silent
shifts that may change everything:
a thunder-crash of cymbal —but then,
again, a rest, a void followed by
bells of glockenspiel;

the music of what happens comes by violin’s
delicately sweet squeals, comes by all
as expressed in its sum:

the orchestra, the choir, quartet,
the trio, duet, the band—

the music of what happens is in
the concerted song of time,
its sum

Jim Culleny
12/29/24

Enjoying the content on 3QD? Help keep us going by donating now.

by Jonathan Kujawa

Four years ago was peak COVID-19. I was thinking about moving furniture, which led to this essay here at 3QD. The topic was the problem of moving a sofa down a hallway with a 90° corner:

The problem is to find the “couch” of largest possible area that could be smoothly pushed down the hallway, around the corner, and then onwards. Leo Moser first asked this question in 1966 [1]. The image above is the couch found by Joseph Gerver in 1992. Gerver’s couch has an area of approximately 2.2195 square meters (where the hallway is exactly 1 meter wide).

As we talked about in that essay four years ago, Gerver used some clever math to find the shape of his couch. If you look very closely, you’ll notice that the corners are clipped off, and the inner semi-circle isn’t exactly a semi-circle. Without those tweaks, you have Hammersley’s couch from 1968, which has an area of π/2 + 2/π ≈ 2.2074 square meters. Several decades of time and Gerver’s hard work got us a whopping 0.0121 square meters.

Of course there was no guarantee that Gerver’s couch was the biggest possible. Dr. Gerver’s approach made no promises that it gave the best possible, after all. A little more convincing is the fact that in 30 years we haven’t been able to do any better. But mathematics is a game of centuries and millennia — a few decades is small potatoes. In 2018, Yoav Kallus and Dan Romik proved that the couch could be no larger than 2.37 square meters. But the gap in size between Gerver’s couch and the Kallus-Romik upper bound is an order of magnitude larger than that between the couches of Gerver and Hammersley.

I’m not an expert in this area, but if you asked me a month ago, I would have said it could be many years before we’d resolve Moser’s couch question. Indeed, this could easily be out of reach forever. Many easy-to-state math questions are practically unsolvable. One could make infinitely many tweaks to Gerver’s couch as possible improvements. Or, you could do some entirely different shape, maybe something fractally and totally unexpected.

Part of the problem is that there is no single formula that gives you the area of all possible couches. If you had such a formula, you could try to maximize its output like you learned to do in your Calculus class. But if different shapes have different area formulas, there is no effective way to tackle them all at once.

This is no doubt my own human biases, but I find it a big bummer that Hammersley’s couch with an area given by a beautiful formula using pi is not the maximum. Once you show me that tweaks can give you bigger couches with ugly areas like Gerver’s, my hope for a definitive answer pretty quickly disappears.

So I was gobsmacked to see two interesting papers about Moser’s couch problem in the last few weeks. Read more »