My Quora session
Thursday, May 19th, 2016Here it is. Enjoy! (But sorry, no new questions right now.)
Here it is. Enjoy! (But sorry, no new questions right now.)
(-3) Bonus Announcement of May 30: As a joint effort by Yuri Matiyasevich, Stefan O’Rear, and myself, and using the Not-Quite-Laconic language that Stefan adapted from Adam Yedidia’s Laconic, we now have a 744-state TM that halts iff there’s a counterexample to the Riemann Hypothesis.
(-2) Today’s Bonus Announcement: Stefan O’Rear says that his Turing machine to search for contradictions in ZFC is now down to 1919 states. If verified, this is an important milestone: our upper bound on the number of Busy Beaver values that are knowable in standard mathematics is now less than the number of years since the birth of Christ (indeed, even since the generally-accepted dates for the writing of the Gospels).
Stefan also says that his Not-Quite-Laconic system has yielded a 1008-state Turing machine to search for counterexamples to the Riemann Hypothesis, improving on our 5372 states.
(-1) Another Bonus Announcement: Great news, everyone! Using a modified version of Adam Yedidia’s Laconic language (which he calls NQL, for Not Quite Laconic), Stefan O’Rear has now constructed a 5349-state Turing machine that directly searches for contradictions in ZFC (or rather in Metamath, which is known to be equivalent to ZFC), and whose behavior is therefore unprovable in ZFC, assuming ZFC is consistent. This, of course, improves on my and Adam’s state count by 2561 states—but it also fixes the technical issue with needing to assume a large cardinal axiom (SRP) in order to prove that the TM runs forever. Stefan promises further state reductions in the near future.
In other news, Adam has now verified the 43-state Turing machine by Jared S that halts iff there’s a counterexample to Goldbach’s Conjecture. The 27-state machine by code golf addict is still being verified.
(0) Bonus Announcement: I’ve had half a dozen “Ask Me Anything” sessions on this blog, but today I’m trying something different: a Q&A session on Quora. The way it works is that you vote for your favorite questions; then on Tuesday, I’ll start with the top-voted questions and keep going down the list until I get tired. Fire away! (And thanks to Shreyes Seshasai at Quora for suggesting this.)
(1) When you announce a new result, the worst that can happen is that the result turns out to be wrong, trivial, or already known. The best that can happen is that the result quickly becomes obsolete, as other people race to improve it. With my and Adam Yedidia’s work on small Turing machines that elude set theory, we seem to be heading for that best case. Stefan O’Rear wrote a not-quite-Laconic program that just searches directly for contradictions in a system equivalent to ZFC. If we could get his program to compile, it would likely yield a Turing machine with somewhere around 6,000-7,000 states whose behavior was independent of ZFC, and would also fix the technical problem with my and Adam’s machine Z, where one needed to assume a large-cardinal axiom called SRP to prove that Z runs forever. While it would require a redesign from the ground up, a 1,000-state machine whose behavior eludes ZFC also seems potentially within reach using Stefan’s ideas. Meanwhile, our 4,888-state machine for Goldbach’s conjecture seems to have been completely blown out of the water: first, a commenter named Jared S says he’s directly built a 73-state machine for Goldbach (now down to 43 states); second, a commenter named “code golf addict” claims to have improved on that with a mere 31 states (now down to 27 states). These machines are now publicly posted, but still await detailed verification.
(2) My good friend Jonah Sinick cofounded Signal Data Science, a data-science summer school that will be running for the second time this summer. They operate on an extremely interesting model, which I’m guessing might spread more widely: tuition is free, but you pay 10% of your first year’s salary after finding a job in the tech sector. He asked me to advertise them, so—here!
(3) I was sad to read the news that Uber and Lyft will be suspending all service in Austin, because the city passed an ordinance requiring their drivers to get fingerprint background checks, and imposing other regulations that Uber and Lyft argue are incompatible with their model of part-time drivers. The companies, of course, are also trying to send a clear message to other cities about what will happen if they don’t get the regulatory environment they want. To me, the truth of the matter is that Uber/Lyft are like the web, Google, or smartphones: clear, once-per-decade quality-of-life advances that you try once, and then no longer understand how you survived without. So if Austin wants to maintain a reputation as a serious, modern city, it has no choice but to figure out some way to bring these companies back to the negotiating table. On the other hand, I’d also say to Uber and Lyft that, even if they needed to raise fares to taxi levels to comply with the new regulations, I expect they’d still do a brisk business!
For me, the “value proposition” of Uber has almost nothing to do with the lower fares, even though they’re lower. For me, it’s simply about being able to get from one place to another without needing to drive and park, and also without needing desperately to explain where you are, over and over, to a taxi dispatcher who sounds angry that you called and who doesn’t understand you because of a combination of language barriers and poor cellphone reception and your own inability to articulate your location. And then wondering when and if your taxi will ever show up, because the dispatcher couldn’t promise a specific time, or hung up on you before you could ask them. And then embarking on a second struggle, to explain to the driver where you’re going, or at least convince them to follow the Google Maps directions. And then dealing with the fact that the driver has no change, you only have twenties and fifties, and their little machine that prints receipts is out of paper so you can’t submit your trip for reimbursement either.
So yes, I really hope Uber, Lyft, and the city of Austin manage to sort this out before Dana and I move there! On the other hand, I should say that there’s another part of the new ordinance—namely, requiring Uber and Lyft cars to be labeled—that strikes me as an unalloyed good. For if there’s one way in which Uber is less convenient than taxis, it’s that you can never figure out which car is your Uber, among all the cars stopping or slowing down near you that look vaguely like the one in the app.
I’ve supervised a lot of great student projects in my nine years at MIT, but my inner nerdy teenager has never been as personally delighted by a project as it is right now. Today, I’m proud to announce that Adam Yedidia, a PhD student at MIT (but an MEng student when he did most of this work), has explicitly constructed a one-tape, two-symbol Turing machine with 7,918 states, whose behavior (when run on a blank tape) can never be proven from the usual axioms of set theory, under reasonable consistency hypotheses. Adam has also constructed a 4,888-state Turing machine that halts iff there’s a counterexample to Goldbach’s Conjecture, and a 5,372-state machine that halts iff there’s a counterexample to the Riemann Hypothesis. In all three cases, this is the first time we’ve had a reasonable explicit upper bound on how many states you need in a Turing machine before you can see the behavior in question.
Here’s our research paper, on which Adam generously included me as a coauthor, even though he did the heavy lifting. Also, here’s a github repository where you can download all the code Adam used to generate these Turing machines, and even use it to build your own small Turing machines that encode interesting mathematical statements. Finally, here’s a YouTube video where Adam walks you through how to use his tools.
A more precise statement of our main result is this: we give a 7,918-state Turing machine, called Z (and actually explicitly listed in our paper!), such that:
A bit of background: it follows, as an immediate consequence of Gödel’s Incompleteness Theorem, that there’s some computer program, of some length, that eludes the power of ordinary mathematics to prove what it does, when it’s run with an unlimited amount of memory. So for example, such a program could simply enumerate all the possible consequences of the ZFC axioms, one after another, and halt if it ever found a contradiction (e.g., a proof of 1+1=3). Assuming ZFC is consistent, this program must run forever. But again assuming ZFC is consistent, ZFC can’t prove that the program runs forever, since if it did, then it would prove its own consistency, thereby violating the Second Incompleteness Theorem!
Alas, this argument still leaves us in the dark about where, in space of computer programs, the “Gödelian gremlin” rears its undecidable head. A program that searches for an inconsistency in ZFC is a fairly complicated animal: it needs to encode not only the ZFC axiom schema, but also the language and inference rules of first-order logic. Such a program might be thousands of lines long if written in a standard programming language like C, or millions of instructions if compiled down to a bare-bones machine code. You’d certainly never run across such a program by chance—not even if you had a computer the size of the observable universe, trying one random program after another for billions of years in a “primordial soup”!
So the question stands—a question that strikes me as obviously important, even though as far as I know, only one or two people ever asked the question before us; see here for example. Namely: do the axioms of set theory suffice to analyze the behavior of every computer program that’s at most, let’s say, 50 machine instructions long? Or are there super-short programs that already exhibit “Gödelian behavior”?
Theoretical computer scientists might object that this is “merely a question of constants.” Well yes, OK, but the origin of life in our universe—a not entirely unrelated puzzle—is also “merely a question of constants”! In more detail, we know that it’s possible with our laws of physics to build a self-replicating machine: say, DNA or RNA and their associated paraphernalia. We also know that tiny molecules like H2O and CO2 are not self-replicating. But we don’t know how small the smallest self-replicating molecule can be—and that’s an issue that influences whether we should expect to find ourselves alone in the universe or find it teeming with life.
Some people might also object that what we’re asking about has already been studied, in the half-century quest to design the smallest universal Turing machine (the subject of Stephen Wolfram’s $25,000 prize in 2007, to which I responded with my own $25.00 prize). But I see that as fundamentally different, for the following reason. A universal Turing machine—that is, a machine that simulates any other machine that’s described to it on its input tape—has the privilege of offloading almost all of its complexity onto the description format for the input machine. So indeed, that’s exactly what all known tiny universal machines do! But a program that checks (say) Goldbach’s Conjecture, or the Riemann Hypothesis, or the consistency of set theory, on an initially blank tape, has no such liberty. For such machines, the number of states really does seem like an intrinsic measure of complexity, because the complexity can’t be shoehorned anywhere else.
One can also phrase what we’re asking in terms of the infamous Busy Beaver function. Recall that BB(n), or the nth Busy Beaver number, is defined to be the maximum number of steps that any n-state Turing machine takes when run on an initially blank tape, assuming that the machine eventually halts. The Busy Beaver function was the centerpiece of my 1998 essay Who Can Name the Bigger Number?, which might still attract more readers than anything else I’ve written since. As I stressed there, if you’re in a biggest-number-naming contest, and you write “BB(10000),” you’ll destroy any opponent—however otherwise mathematically literate they are—who’s innocent of computability theory. For BB(n) grows faster than any computable sequence of integers: indeed, if it didn’t, then one could use that fact to solve the halting problem, contradicting Turing’s theorem.
But the BB function has a second amazing property: namely, it’s a perfectly well-defined integer function, and yet once you fix the axioms of mathematics, only finitely many values of the function can ever be proved, even in principle. To see why, consider again a Turing machine M that halts if and only if there’s a contradiction in ZF set theory. Clearly such a machine could be built, with some finite number of states k. But then ZF set theory can’t possibly determine the value of BB(k) (or BB(k+1), BB(k+2), etc.), unless ZF is inconsistent! For to do so, ZF would need to prove that M ran forever, and therefore prove its own consistency, and therefore be inconsistent by Gödel’s Theorem.
OK, but we can now ask a quantitative question: how many values of the BB function is it possible for us to know? Where exactly is the precipice at which this function “departs the realm of mortals and enters the realm of God”: is it closer to n=10 or to n=10,000,000? In practice, four values of BB have been determined so far:
We also know some lower bounds:
See Heiner Marxen’s page or the Googology Wiki (which somehow I only learned about today) for more information.
Some Busy Beaver enthusiasts have opined that even BB(6) will never be known exactly. On the other hand, the abstract argument from before tells us only that, if we confine ourselves to (say) ZF set theory, then there’s some k—possibly in the tens of millions or higher—such that the values of BB(k), BB(k+1), BB(k+2), and so on can never be proven. So again: is the number of knowable values of the BB function more like 10, or more like a million?
This is the question that Adam and I (but mostly Adam) have finally addressed.
It’s hopeless to design a Turing machine by hand for all but the simplest tasks, so as a first step, Adam created a new programming language, called Laconic, specifically for writing programs that compile down to small Turing machines. Laconic programs actually compile to an intermediary language called TMD (Turing Machine Descriptor), and from there to Turing machines.
Even then, we estimate that a direct attempt to write a Laconic program that searched for a contradiction in ZFC would lead to a Turing machine with millions of states. There were three ideas needed to get the state count down to something reasonable.
The first was to take advantage of the work of Harvey Friedman, who’s one of the one or two people I mentioned earlier who’s written about these problems before. In particular, Friedman has been laboring since the 1960s to find “natural” arithmetical statements that are provably independent of ZFC or other strong set theories. (See this AMS Notices piece by Martin Davis for a discussion of Friedman’s progress as of 2006.) Not only does Friedman’s quest continue, but some of his most important progress has come only within the last year. His statements—typically involving objects called “order-invariant graphs”—strike me as alien, and as far removed from anything I’d personally have independent reasons to think about (but is that just a sign of my limited perspective?). Be that as it may, Friedman’s statements still seem a lot easier to encode as short computer programs than the full apparatus of first-order logic and set theory! So that’s what we started with; our work wouldn’t have been possible without Friedman (who we consulted by email throughout the project).
The second idea was something we called “on-tape processing.” Basically, instead of compiling directly from Laconic down to Turing machine, Adam wrote an interpreter in Turing machine (which took about 4000 states—a single, fixed cost), and then had the final Turing machine first write a higher-level program onto its tape and then interpret that program. Instead of the compilation process producing a huge multiplicative overhead in the number of Turing machine states (and a repetitive machine), this approach gives us only an additive overhead. We found that this one idea decreased the number of states by roughly an order of magnitude.
The third idea was first suggested in 2002 by Ben-Amram and Petersen (and refined for us by Luke Schaeffer); we call it “introspective encoding.” When we write the program to be interpreted onto the Turing machine tape, the naïve approach would use one Turing machine state per bit. But that’s clearly wasteful, since in an n-state Turing machine, every state contains ~log(n) bits of information (because of the other states it needs to point to). A better approach tries to exploit as many of those bits as it can; doing that gave us up to a factor-of-5 additional savings in the number of states.
For Goldbach’s Conjecture and the Riemann Hypothesis, we paid the same 4000-state overhead for the interpreter, but then the program to be interpreted was simpler, giving a smaller overall machine. Incidentally, it’s not intuitively obvious that the Riemann Hypothesis is equivalent to the statement that some particular computer program runs forever, but it is—that follows, for example, from work by Lagarias and by Davis, Matijasevich, and Robinson (we used the latter; an earlier version of this post incorrectly stated that we used the Lagarias result).
To preempt the inevitable question in the comments section: yes, we did run these Turing machines for a while, and no, none of them had halted after a day or so. But before you interpret that as evidence in favor of Goldbach, Riemann, and the consistency of ZFC, you should probably know that a Turing machine to test whether all perfect squares are less than 5, produced using Laconic, needed to run for more than an hour before it found the first counterexample (namely, 32=9) and halted. Laconic Turing machines are optimized only for the number of states, not for speed, to put it mildly.
Of course, three orders of magnitude still remain between the largest value of n (namely, 4) for which BB(n) is known to be knowable in ZFC-based mathematics, and the smallest value of n (namely, 7,918) for which BB(n) is known to be unknowable. I’m optimistic that further improvements are possible to the machine Z—whether that means simplifications to Friedman’s statement, a redesigned interpreter (possibly using lambda calculus?), or a “multi-stage rocket model” where a bare-bones interpreter would be used to unpack a second, richer interpreter which would be used to unpack a third, etc., until you got to the actual program you cared about. But I’d be shocked if anyone in my lifetime determined the value of BB(10), for example, or proved the value independent of set theory. Even after the Singularity happens, I imagine that our robot overlords would find the determination of BB(10) quite a challenge.
In an early Shtetl-Optimized post, I described theoretical computer science as “quantitative epistemology.” Constructing small Turing machines whose behavior eludes set theory is not conventional theoretical computer science by any stretch of the imagination: it’s closer in practice to programming languages or computer architecture, or even the recreational practice known as code-golfing. On the other hand, I’ve never been involved with any other project that was so clearly, explicitly about pinning down the quantitative boundary between the knowable and the unknowable.
Comments on our paper are welcome.
Addendum: Some people might wonder “why Turing machines,” as opposed to a more reasonable programming language like C or Python. Well, first of all, we needed a language that could address an unlimited amount of memory. Also, the BB function is traditionally defined in terms of Turing machines. But the most important issue is that we wanted there to be no suspicion whatsoever that our choice of programming language was artificially helping to make our machine small. And hopefully everyone can agree that one-tape, two-symbol Turing machines aren’t designed for anyone’s convenience!
[Warning: This movie review contains spoilers, as well as a continued fraction expansion.]
These days, it takes an extraordinary occasion for me and Dana to arrange the complicated, rocket-launch-like babysitting logistics involved in going out for a night at the movies. One such an occasion was an opening-weekend screening of The Man Who Knew Infinity—the new movie about Srinivasa Ramanujan and his relationship with G. H. Hardy—followed by a Q&A with Matthew Brown (who wrote and directed the film), Robert Kanigel (who wrote the biography on which the film was based), and Fields Medalist Manjul Bhargava (who consulted on the film).
I read Kanigel’s The Man Who Knew Infinity in the early nineties; it was a major influence on my life. There were equations in that book to stop a nerdy 13-year-old’s pulse, like
$$1+9\left( \frac{1}{4}\right) ^{4}+17\left( \frac{1\cdot5}{4\cdot8}\right)
^{4}+25\left( \frac{1\cdot5\cdot9}{4\cdot8\cdot12}\right) ^{4}+\cdots
=\frac{2^{3/2}}{\pi^{1/2}\Gamma\left( 3/4\right) ^{2}}$$
$$\frac{1}{1+\frac{e^{-2\pi}}{1+\frac{e^{-4\pi}}{1+\frac{e^{-6\pi}}{1+\cdots}}%
}}=\left( \sqrt{\frac{5+\sqrt{5}}{2}}-\frac{\sqrt{5}+1}{2}\right)
\sqrt[5]{e^{2\pi}}$$
A thousand pages of exposition about Ramanujan’s mysterious self-taught mathematical style, the effect his work had on Hardy and Littlewood, his impact on the later development of analysis, etc., could never replace the experience of just staring at these things! Popularizers are constantly trying to “explain” mathematical beauty by comparing it to art, music, or poetry, but I can best understand art, music, and poetry if I assume other people experience them like the above identities. Across all the years and cultures and continents, can’t you feel Ramanujan himself leaping off your screen, still trying to make you see this bizarre aspect of the architecture of reality that the goddess Namagiri showed him in a dream?
Reading Kanigel’s book, I was also entranced by the culture of early-twentieth-century Cambridge mathematics: the Tripos, Wranglers, High Table. I asked, why was I here and not there? And even though I was (and remain) at most 1729-1729 of a Ramanujan, I could strongly identify with his story, because I knew that I, too, was about to embark on the journey from total scientific nobody to someone who the experts might at least take seriously enough to try to prove him wrong.
Anyway, a couple years after reading Kanigel’s biography, I went to the wonderful Canada/USA MathCamp, and there met Richard K. Guy, who’d actually known Hardy. I couldn’t have been more impressed had Guy visited Platonic heaven and met π and e there. To put it mildly, no one in my high school had known G. H. Hardy.
I often fantasized—this was the nineties—about writing the screenplay myself for a Ramanujan movie, so that millions of moviegoers could experience the story as I did. Incidentally, I also fantasized about writing screenplays for Alan Turing and John Nash movies. I do have a few mathematical biopic ideas that haven’t yet been taken, and for which any potential buyers should get in touch with me:
But enough digressions; how was the Ramanujan movie?
Just as Ramanujan himself wasn’t an infallible oracle (many of his claims, e.g. his formula for the prime counting function, turned out to be wrong), so The Man Who Knew Infinity isn’t a perfect movie. Even so, there’s no question that this is one of the best and truest movies ever made about mathematics and mathematicians, if not the best and truest. If you’re the kind of person who reads this blog, go see it now. Don’t wait! As they stressed at the Q&A, the number of tickets sold in the first couple weeks is what determines whether or not the movie will see a wider release.
More than A Beautiful Mind or Good Will Hunting or The Imitation Game, or the play Proof, or the TV series NUMB3RS, the Ramanujan movie seems to me to respect math as a thing-in-itself, rather than just a tool or symbol for something else that interests the director much more. The background to the opening credits—and what better choice could there be?—is just page after page from Ramanujan’s notebooks. Later in the film, there’s a correct explanation of what the partition function P(n) is, and of one of Ramanujan’s and Hardy’s central achievements, which was to give an asymptotic formula for P(n), namely $$ P(n) \approx \frac{e^{π \sqrt{2n/3}}}{4\sqrt{3}n}, $$ and to prove the formula’s correctness.
The film also makes crystal-clear that pure mathematicians do what they do not because of applications to physics or anything else, but simply because they feel compelled to: for the devout Ramanujan, math was literally about writing down “the thoughts of God,” while for the atheist Hardy, math was a religion-substitute. Notably, the movie explores the tension between Ramanujan’s untrained intuition and Hardy’s demands for rigor in a way that does them both justice, resisting the Hollywood urge to make intuition 100% victorious and rigor just a stodgy punching bag to be defeated.
For my taste, the movie could’ve gone even further in the direction of “letting the math speak”: for example, it could’ve explained just one of Ramanujan’s infinite series. Audiences might even have liked some more T&A (theorems and asymptotic bounds). During the Q&A that I attended, I was impressed to see moviegoers repeatedly pressing a somewhat-coy Manjul Bhargava to explain Ramanujan’s actual mathematics (e.g., what exactly were the discoveries in his first letter to Hardy? what was in Ramanujan’s Lost Notebook that turned out to be so important?). Then again, this was Cambridge, MA, so the possibility should at least be entertained that what I witnessed was unrepresentative of American ticket-buyers.
From what I’ve read, the movie is also true to South Indian dress, music, religion, and culture. Yes, the Indian characters speak to each other in English rather than Tamil, but Brown explained that as a necessary compromise (not only for the audience’s sake, but also because Dev Patel and the other Indian actors didn’t speak Tamil).
Some reviews have mentioned issues with casting and characterization. For example, Hardy is portrayed by Jeremy Irons, who’s superb but also decades older than Hardy was at the time he knew Ramanujan. Meanwhile Ramanujan’s wife, Janaki, is played by a fully-grown Devika Bhise; the real Janaki was nine (!) when she married Ramanujan, and fourteen when Ramanujan left for England. J. E. Littlewood is played as almost a comic-relief buffoon, so much so that it feels incongruous when, near the end of the film, Irons-as-Hardy utters the following real-life line:
I still say to myself when I am depressed and find myself forced to listen to pompous and tiresome people, “Well, I have done one thing you could never have done, and that is to have collaborated with Littlewood and Ramanujan on something like equal terms.”
Finally, a young, mustachioed Bertrand Russell is a recurring character. Russell and Hardy really were friends and fellow WWI pacifists, but Hardy seeking out Bertie’s advice about each Ramanujan-related development seems like almost certainly just an irresistible plot device.
But none of that matters. What bothered me more were the dramatizations of the prejudice Ramanujan endured in England. Ramanujan is shown getting knocked to the ground, punched, and kicked by British soldiers barking anti-Indian slurs at him; he then shows up for his next meeting with Hardy covered in bruises, which Hardy (being aloof) neglects to ask about. Ramanujan is also depicted getting shoved, screamed at, and told never to return by a math professor who he humiliates during a lecture. I understand why Brown made these cinematic choices: there’s no question that Ramanujan experienced prejudice and snobbery in Cambridge, and that he often felt lonely and unwelcome there. And it’s surely easier to show Ramanujan literally getting beaten up by racist bigots, than to depict his alienation from Cambridge society as the subtler matter that it most likely was. To me, though, that’s precisely why the latter choice would’ve been even more impressive, had the film managed to pull it off.
Similarly, during World War I, the film shows not only Trinity College converted into a military hospital, and many promising students marched off to their deaths (all true), but also a shell exploding on campus near Ramanujan, after which Ramanujan gazes in horror at the bleeding dead bodies. Like, isn’t the truth here dramatic enough?
One other thing: the movie leaves you with the impression that Ramanujan died of tuberculosis. More recent analysis concluded that it was probably hepatic amoebiasis that he brought with him from India—something that could’ve been cured with the medicine of the time, had anyone correctly diagnosed it. (Incidentally, the film completely omits Ramanujan’s final year, back in India, when he suffered a relapse of his illness and slowly withered away, yet with Janaki by his side, continued to do world-class research and exchanged letters with Hardy until the very last days. Everyone I read commented that this was “the right dramatic choice,” but … I dunno, I would’ve shown it!)
But enough! I fear that to harp on these defects is to hold the film to impossibly-high, Platonic standards, rather than standards that engage with the reality of Hollywood. An anecdote that Brown related at the end of the Q&A session brought this point home for me. Apparently, Brown struggled for an entire decade to attract funding for a film about a turn-of-the-century South Indian mathematician visiting Trinity College, Cambridge, whose work had no commercial or military value whatsoever. At one point, Brown was actually told that he could get the movie funded, if he’d agree to make Ramanujan fall in love with a white nurse, so that a British starlet who would sell tickets could be cast as his love interest. One can only imagine what a battle it must have been to get a correct explanation of the partition function onto the screen.
In the end, though, nothing made me appreciate The Man Who Knew Infinity more than reading negative reviews of it, like this one by Olly Richards:
Watching someone balancing algorithms or messing about with multivariate polynomials just isn’t conducive to urgently shovelling popcorn into your face. Difficult to dislike, given its unwavering affection for its subject, The Man Who Knew Infinity is nevertheless hamstrung by the dryness of its subject … Sturdy performances and lovely scenery abound, but it’s still largely just men doing sums; important sums as it turns out, but that isn’t conveyed to the audience until the coda [which mentions black holes] tells us of the major scientific advances they aided.
On behalf of mathematics, on behalf of my childhood self, I’m grateful that Brown fought this fight, and that he won as much as he did. Whether you walk, run, board a steamship, or take taxi #1729, go see this film.
Addendum: See also this review by Peter Woit, and this in Notices of the AMS by Ramanujan expert George Andrews.