The Karp-Lipton Advice Column
Today, Shtetl-Optimized is extremely lucky to have the special guest blogger poly: the ‘adviser’ in the computational complexity class P/poly (P with polynomial-sized advice string), defined by Richard Karp and Richard Lipton in 1982.
As an adviser, poly is known for being infinitely wise and benevolent, but also for having a severe limitation: namely, she’s sensitive only to the length of her input, and not to any other information about it. Her name comes from the fact that her advice is polynomial-size, which is the problem that prevents her from simply listing the answers to every possible question in a gigantic lookup table, the way she’d like to.
Without further ado, let’s see what advice poly is able to offer her respondents.
Dear poly,
When my husband and I first started dating, we were going at it like rabbits! Lately, though, he seems to have no interest in sex. That’s not normal for a guy, is it? What can I do to spice things up in the bedroom?
Sincerely,
Frustrated Wife
Dear Frustrated Wife,
Unfortunately, I don’t know exactly what your question is. All I was told is that the question was 221 characters long. But here’s something that might help: whenever you’re stuck in a rut, sometimes you can “shake things up” with the use of randomness. So, please accept, free of charge, the following string of 221 random bits:
111010100100010010101111110010111101011010001
000111100101000111111011101110100110000110100
0010010010000010110101100100100111000010110
111001011001111111101110100010000010100111000
0111101111001101001111101000001010110101101
Well, it’s not really “random,” since everyone else with a 221-character question would’ve gotten the exact same string. But it’s random enough for many practical purposes. I hope it helps you somehow … good luck!
Sincerely,
poly
Dear poly,
I’m a 29-year-old autistic male: a former software entrepreneur currently worth about $400 million, who now spends his time donating to malaria prevention and women’s rights in the developing world. My issue is that I’ve never been on a date, or even kissed anyone. I’m terrified to make an advance. All I read in the news is an endless litany of male sexual misbehavior: Harvey Weinstein, Louis C. K., Leon Wieseltier, George H. W. Bush, Roy Moore, the current president (!), you name it. And I’m consumed by the urge not to be a pig like those guys. Like, obviously I’m no more likely to start stripping or masturbating or something in front of some woman I just met, than I am to morph into a koala bear. But from reading Slate, Salon, Twitter, my Facebook news feed, and so forth, I’ve gotten the clear sense that there’s nothing I could do that modern social mores would deem appropriate and non-creepy—at least, not a guy like me, who wasn’t lucky enough to be born instinctively understanding these matters. I’m grateful to society for enabling my success, and have no desire to break any of its written or unwritten rules. But here I genuinely don’t know what society wants me to do. I’m writing to you because I remember you from my undergrad CS classes—and you’re the only adviser I ever encountered whose advice could be trusted unconditionally.
Yours truly,
Sensitive Nerd
Dear Sensitive Nerd,
I see your that letter is 1369 characters long. Based on that, here are a few things I can tell you that might be helpful:
- The Riemann Hypothesis is true.
- ZFC set theory is consistent.
- The polynomial hierarchy is contained in PP.
Write me a 3592-character letter the next time, and I’ll give you an even longer list of true mathematical statements! (I actually know how to solve the halting problem—no joke!—but am condemned to drip, drip, drip out the solutions, a few per input length.)
But I confess: no sooner did I list these truths than I reflected that they, or even a longer list, might not help much with your problem, whatever it might have been. It’s even possible to have a problem for which no amount of truth helps in solving it. So, I dunno: maybe try not worrying so much, and write back to let me know if that helped? (Not that I expect to understand your reply, or would be able to change any of my advice at this point even if I did.)
Good luck!
–poly
Dear poly,
c34;c’y9v3x
Sincerely,
Unhappy in Unary
Dear Unhappy in Unary,
Finally, someone who writes to me in a language I can understand! Your question is 11 characters long. I understand that to be a code expressing that you’re bankrupt, and are filing for Chapter 11 bankruptcy protection. Financial insolvency isn’t easy for anyone. But here’s some advice: put everything you have into Bitcoin, and sell out a year from now. Unfortunately, I don’t know exactly when you’re writing to me, but at least at the time my responses were hardwired in, this was some damn good advice.
You’re welcome,
poly
poly’s polynomial-sized advice column is syndicated in newspapers nationwide, and can also be accessed by simply moving your tape head across your advice tape. You’re welcome to comment on this post, but I might respond only to the lengths of the comments, rather than anything else about them. –SA
Follow
Comment #1 November 15th, 2017 at 5:41 am
Ahahahaha. Though I do not really get the last part
Comment #2 November 15th, 2017 at 6:47 am
Dear poly,
I was reading Scott Aaroson’s most recent blog article, and something he wrote made a deep internal conflict bubble up to the surface. As a result, I am surely doomed to many sleepless nights once again, unless this conflict can finally be resolved by your advice.
poly, what is the one true way to denote the kth level of the polynomial hierarchy? Is it \Sigma_k P, or \Sigma_k^p? (resp. \Pi_k P or \Pi_k^p?)
This dilemma first arose recently when I was writing slides for a talk, in which, seeking as succinct a definition of the polynomial hierarchy as possible (both to prevent slide clutter and to ease understanding of the concepts), I chose to omit the bounds on the quantifiers, figuring, “eh, the verifier runs in polynomial time anyway, which in turn implies a bound on the amount of the string it can investigate.” But nay — it later struck me that the verifier runs in polynomial time in the length of its *entire input*. The quantifiers ensure that the entire input is only polynomially larger than the instance. In fact without polynomially bounded quantification, one ends up with the arithmetic hierarchy, loosely a consequence of Turing’s universal machine.
But this seems to have implications for the dilemma in question: \Sigma_k P seems to evoke the idea of placing k arbitrary alternating quantifiers before a polynomial-time verifier, whereas \Sigma_k^p seems to evoke the idea of bounding the quantifiers themselves. So whilst I have an aesthetic preference for \Sigma_k P, and note that no lesser luminary than Scott Aaronson himself favours it in his bestselling book Quantum Computing Since Democritus, \Sigma_k^p might have correctness on its side.
Actually, maybe they’re both wrong: whilst \Sigma_k P might “technically” be \Sigma_k, \Sigma_k^p might actually be R or RE, since the notation doesn’t imply a bound on the running time of the verifier at all. Might the truly correct denotation be \Sigma_k^p P? But then this reminds me of, in another serious and important dilemma of our time, tabs vs spaces, the “tabs for indentation, spaces for formatting” option, which is a chimeric abomination, about which the less said, the better.
This grave dilemma rears its head once again because, in describing your own graceful and benevolent manner, Scott said,
“Her name comes from the fact that her advice is polynomial-size, which is the problem that prevents her from simply listing the answers to every possible question in a gigantic lookup table, the way she’d like to.”
The thought immediately occurred to me, “but even if she did, wouldn’t her polynomially time-bounded questioner be unable to store such a gigantic offering, let alone query it?” But mere seconds later I corrected myself: in standard definitions of P/poly, the questioner’s input is the instance and the advice string *combined*. Indeed, your big sister exp seems to…uh, nevermind.
I figure that with your many storied interactions with PH, as well as the deep relevance of this matter to your very own identity, surely you should be able to provide helpful advice on this manner, or at least tell me something interesting about somebody else’s 3252-character long problem.
Thank you in advance,
Greg
Comment #3 November 15th, 2017 at 6:59 am
A
Comment #4 November 15th, 2017 at 8:27 am
Since I’m not restricted to just reading the length of a message, I think I’ll comment on the Sensitive Nerd post: If anything I think that recent events should make Sensitive Nerd feel better: if they just try to be a decent person there’s a pretty high probability they’ll be ok compared to apparently a large fraction of people who are engaging in behavior which I have no doubt that Sensitive Nerd will immediately recognize as bad and unacceptable.
Comment #5 November 15th, 2017 at 8:46 am
Is this trusted advice? If it tells me the Riemann Hypothesis is true, is that safe, or does it also need to give a proof? (Why wouldn’t it include a proof, since shouldn’t it fit?)
Comment #6 November 15th, 2017 at 9:06 am
Joshua Zelinsky, I agree! If anything, recent events reveal that those unwritten rules are difficult for a disturbingly large number of men with power who are otherwise not known for weakness in interpreting social cues.
Perhaps these recent events will move the culture into abandoning these unwritten rules in favor of clear affirmative consent. That should make things easier for those who have a hard time parsing the unwritten rules.
Comment #7 November 15th, 2017 at 9:10 am
Peter Morgan #3:
B
Comment #8 November 15th, 2017 at 9:15 am
John #5: Karp-Lipton advice is so completely trustable that a proof would be superfluous.
Comment #9 November 15th, 2017 at 10:05 am
Atret #6, have you considered that they might have been optimizing for a different objective, and were actually quite adept at reading the cues? (Sad as it is.)
Comment #10 November 15th, 2017 at 10:08 am
This comment is 35 characters long.
Comment #11 November 15th, 2017 at 10:12 am
Joshua #4 and Atreat #6: As a theoretical computer scientist, my working assumption is that in almost every case, whatever happens will be the worst that could happen. That way I’m rarely wrong … and better yet, I’m pleasantly surprised whenever I am! 🙂
I remember when, on the eve of Trump’s ghastly ascent to power, Scott Alexander presciently warned those of his readers who were tempted to see a silver lining in it, who said: “sure, Trump will destroy the US’s standing in the world, trash the environment, maybe even launch a nuclear war, but at least he’ll beat back the insane excesses of political correctness.” Other Scott explained why even that hope was incredibly naïve: because the forces of political correctness, correctly realizing they were under siege, would respond by simply tightening their grip over whatever they still did control (academia, Silicon Valley, many media outlets…) with a vehemence never before seen.
Likewise, I think it would be broad-minded and wise to respond to the latest revelations of predatory male behavior exactly as Joshua suggests: “just try to be a decent person, and there’s a pretty good chance you’ll be OK compared to apparently a large fraction of people…” But if my own reading of editorials and social media is any guide, this is not the response we’re going to see. Instead, poor schmucks like “Sensitive Nerd” should expect a level of heavenly fury that dwarfs anything that came before.
Straining for an analogy that’s equally likely to offend all sides: 🙂 as I see it, the powerful men who abuse women are like Al Qaeda or ISIS. The social justice warriors are like an American bomber squadron, responding to the latest terrorist provocation by raining down righteous vengeance from the skies. And the “Sensitive Nerds” are like the peaceful Afghan villagers who will wind up as collateral damage. (In both cases, incidentally, I put the lion’s share of moral responsibility on the terrorists who initiate the chain of events.)
Again, I hope I’m wrong about this.
Comment #12 November 15th, 2017 at 10:13 am
Does poly respond to (maybe obsene) cultural associations of numbers?
Comment #13 November 15th, 2017 at 11:06 am
Scott says, “Karp-Lipton advice is so completely trustable that a proof would be superfluous.”
Whoaaa! I thought that was only said of Carl Gauss…
Comment #14 November 15th, 2017 at 11:16 am
what are you smoking man?
Comment #15 November 15th, 2017 at 11:33 am
#14: I assure you, nothing but complexity theory lately, more BQP than THC, more PCP than … oh nevermind. And no more 25-character questions allowed.
Comment #16 November 15th, 2017 at 12:00 pm
Dear Sensitive Nerd,
I’m reminded of a (true) story. In 209 BC, two Qin Dynasty army officers, Chen Sheng and Wu Guang, were ordered to lead their troops on a march to provide reinforcements. They were delayed by heavy rain and flooding and couldn’t make their rendezvous time.
“What is the penalty for being late?” Chen asked.
“Death,” said Wu.
“And what is the penalty for rebellion?” Chen asked.
“Death,” said Wu.
“Well, we’re late,” said Chen.
And thus began the Dazexiang uprising.
You seem like a basically decent person. Treat other people like basically decent people, and if you get into trouble, take refuge in the fact that someone would find a way to be upset at you anyway.
—
Dear poly,
a68594c638077842
Comment #17 November 15th, 2017 at 3:18 pm
The “sensitive nerd” story is ridiculous. The people you are talking about committed various forms of sexual assault. Comparing it to the social missteps of a socially unaware nerd (which I, and many of your readership, are quite familiar with) is absurd.
When a well-meaning socially awkward person is accused of sexual misconduct and run out of town despite not harming anyone, we can start talking about political correctness crossing the line. But I don’t see any circumstance in which there’s another side (or a clever point to make) about a case like Roy Moore’s. You’re doing yourself (and even your point) a disservice with this joke/satire/whatever this article is.
“I genuinely don’t know what society wants me to do”—I seriously do relate to this in a general sense, but the responses in this case are things like “don’t make advances towards underage girls” and “don’t expose yourself to women.” It’s very simple, even from the perspective of needing to learn societal norms by rote rather than picking them up naturally. I can’t imagine what legitimate point is being attempted here here.
Comment #18 November 15th, 2017 at 5:54 pm
I see most people in the comments are sensitive nerds.
Comrades, approach this topic as if you were approaching any other.
Find a good textbook, a good problem set, a tutor (desirably a former psychologist) some video lectures, and you know the rest. Just always remember treating people with respect even if you’re playing a harraser.
Sincerely, an Insensitive Nerd, Ph.D.
Comment #19 November 16th, 2017 at 12:41 am
anon #17: I hope Roy Moore goes down in ignominious defeat—well, there are maybe 500 reasons for that on top of his pursuit of underage girls—and I hope he’s then criminally prosecuted. I hope Trump gets impeached approximately yesterday (again, his sexual assaults are one of dozens of reasons). And as long as we’re hoping, I hope that a constitutional amendment then gets passed eliminating the Electoral College, forcing presidential candidates to disclose their tax returns, and other reforms aimed at ensuring that nothing like Trump ever happens again.
But the “Sensitive Nerd” story is anything but absurd. During the comment-171 affair, I got plaintive emails from literally hundreds of people in his position, a position I’m very thankful to have escaped myself more than a decade ago. And I suppose I have no choice but to continue giving voice from time to time to this vast underground ocean of human misery—so long as (to quote Unsong) “somebody has to and no one else will.”
It always flabbergasts me how blind my fellow progressives are to the problem, given how eagle-eyed they’d become about it if we changed the example. “Oh, boo-hoo, so your Facebook feed is entirely filled with florid stories of immigrants committing rapes and murders, one after another after another, along with the perps’ menacing mugshots? Well, if you’re an immigrant and you’re worried about being mistaken for ‘one of them,’ then that’s as easy as not committing any rapes or murders yourself, isn’t it? Or maybe you are ‘one of them’?”
If you’re a sensitive nerd, what you need is constant reassurances that you’re OK, you’re not one of the bad ones. Your working assumption, which was seared into your consciousness since childhood, is that all the normal people despise you—and if people patted you on the back a hundred times per day and told you that wasn’t the case, well, it would help, but two hundred times per day would help more.
And what you get instead, from everything you read, is the precise opposite: people you respect and admire explaining how you probably are one of the bad ones, how even raising the possibility that you’re not is derailing the conversation and therefore misogynist in itself. (If you think this characterization is unfair, spend a day reading social media and get back to me.)
To the feminists fighting the good fight against male piggishness, what I really want to say is this: you and I are comrades, just different divisions of the same army. You’re the ones planning bombing sorties against the fortresses of patriarchy and sexism. And I’m the one—there aren’t many of us—looking for ways to reduce the collateral damage when the bombs land, and even to administer emergency medicine to some of the innocent Aspergery civilians who we know will get maimed anyway. These two divisions can and should work together, because properly understood, their goals coincide.
I’ll probably close the comments on this post in a few hours, because the discussion isn’t breaking much new ground, I need to conserve my time for other things, and the original post was pretty silly anyway. Oh well, “somebody has to and no one else will.”
Comment #20 November 16th, 2017 at 5:19 am
Dear Scott — It’s my hope that this fine “Karp-Lipton Advice Column” thread might conclude with some advice — from a reliable complexity-theory advice-function, naturally 🙂 — regarding the following passage from Avi Wigderson’s Mathematics and Computation (see page 136 of the on-line PDF draft)
The Question Asked Does ‘P/poly’ = ‘P/polyQTN’, where ‘P/polyQTN’ is the complexity class of languages recognized by a polynomial-time Turing machine with a polynomial-bounded quantum tensor network (QTN) advice function?
Here “polynomial-bounded tensor network advice function” is an oracle function that computes, at zero imputed cost, any real-valued quantum expectation function that is given in terms of a tensor network whose operators and wave functions are specified by a polynomial-bounded advice string.
Observation Obviously ‘P/polyQTN ⊆ P/poly’; the question asked is whether the inclusion is in fact an ‘obvious’ equality — ‘obvious’ to complexity-theory experts, that is!
Motivation Motivation comes in part from an essay (a gently and hilariously satirical essay) by computer science grad student Zachary C. Lipton (any relation to Dick Lipton?), which appeared on November 9 in the weblog Approximately Correct, titled “ICML 2018 Registrations Sell Out Before Submission Deadline“.
Evidently there’s no shortage of machine-learning students who are deeply interested in the practical construction of ‘P/polyQTN’ advice-functions.
Avi Wigderson’s observations make it natural to ask “Why?”.
Specifically, Wigderson’s remarks authorize and encourage students to regard tensor networks from three perspectives: (1) as “computational devices”; (2) as mathematical “objects”; (3) as “useful heuristics”. Phrased broadly, therefore, the question asked is: how might a complexity class ‘P/polyQTN’ be formally defined, so as to usefully unify Avi Wigderson’s three perspectives on tensor network computions?
Comment #21 November 16th, 2017 at 6:01 am
You’ve got a sharp dividing line between evil combatants and innocent civilians in your analogy. It’s not that simple. Do the actions of Louis CK or Harvey Weinstein or Roy Moore make it sound like they are socially graceful with women?
One failure mode for Aspy guys is to withdraw into depression and isolation. Another failure mode is to get power and arrange situations that let them safely coerce women. There’s plenty of murky combinations of these modes and others.
Most Aspy guys of course instead find positive ways of growing and thriving. But even some of those (and I include myself in that group) end up discomforting, scaring, and hurting women on their path to finding a place in the social world.
Women are taking a risk when they give an Aspy guy one or ten or one hundred of those pats per day. Usually it’s fine. Sometimes it starts a virtuous circle. But it might get her a stalker or even a rapist. And it’s NOT HER JOB.
What I see as positive in your Comment #19 is the theme of reaching out to help other Aspy guys. We need to spend a lot more time reassuring each other and helping each other grow, and not expecting that women will do that for us.
We need to spend less time clawing for victim status. It’s a prize not worth having.
Comment #22 November 16th, 2017 at 6:05 am
I loved the idea of the column, by the way. I feel like I’ve got a much better grasp on P/poly now.
Comment #23 November 16th, 2017 at 7:48 am
Scott,
“Instead, poor schmucks like “Sensitive Nerd” should expect a level of heavenly fury that dwarfs anything that came before.”
I don’t know. We know that there exist Amanda Marcotte’s in this world. I don’t see how the current collective cultural trauma we are experiencing emboldens them. People like her will be screaming, “Burn all the radicals!!” no matter what. Colby’s friends will always be trying to “help” him not go too far.
When you see non-sensitive nerds getting legitimate just desserts for assualting and harassing women, I understand the instinct to flinch when you see the Marcotte’s feasting. Anger and rage can let loose fires that burn the whole forest even if it started as a “controlled” fire. Still, I think arsonists will set fires no matter what the situation is.
Comment #24 November 16th, 2017 at 8:14 am
While we’re discussing polynomial advice, a question about it. Has anyone thought that much about the computational complexity of generating the advice itself for a given task?
Here’s my very naive thoughts which can probably be improved (and probably already have been), but should make the question a little more clear and concrete:
For example, the advice needed to put ZPP or BPP into P/Poly is can be computed in PSPACE. But we can also bound the time it takes. For BPP, if we simply run through every advice string and check that it gives the right result for that output then the time it takes to find a good string is about (2^O(A+n))M where A is the length of working piece of advice, n is the length of our input to our BPP algorithim, and M is the amount of time it takes to actually compute with no advice or other tricks whether a given input for our problem is accept or reject.
But for ZPP we can do slightly better because we can keep a string as part of our advice if it works for a really long sequence of inputs: BPP can’t do this immediately because a BPP machine taking the advice can’t tell when a string is giving bad advice where as the ZPP machine will when given bad advice simply return I-don’t-know and then move on to the next part of the advice string. This means that if we want to make advice for a problem in ZPP we can get away with
(2^(O(A)+o(n)))M. Note that although A is the length of a working piece of advice, we’ve actually ended up giving advice that is slightly longer than our minimum advice length.
But we can actually with only a little overhead do the same for BPP by tagging our advice string so our advice essentially is of a slightly different form and reads:
From 0 to String A, use X
From String A+1 to String Y,
etc.
So in both cases we can do with time (2^(O(A)+o(n)))M. Can we do better?
Comment #25 November 16th, 2017 at 8:34 am
Alan #21: Of course it’s no one else’s job to help Aspy guys. What I think is reasonable to ask, and the bare minimum that other communities do ask, is that people don’t actively make the problem worse by libeling the sufferers, taking joy in their “tears,” etc. I challenge you to spend a day typing the relevant search terms into Google, then come back and tell me this is a request that the mainstream is currently meeting.
I also think you wildly underestimate the extent to which, if you’re a nerdy male, then absolutely anything you do or don’t do will nowadays get classified as Problematic. The example par excellence of this was a Telegraph writer—not some random blog—attacking Silicon Valley nerds like Mark Zuckerberg for donating to cancer research.
Likewise, if nerdy guys do what you endorse, try to help each other solve their problems, and “pat each other on the back” for not being jerks, do you not see how they’re creating a toxic and exclusionary bro culture that needs to be broken down?
Comment #26 November 16th, 2017 at 8:44 am
Joshua #24: Good question! No, I don’t think anyone does know how to generate the advice faster—if they did, it would certainly be of great interest in the context of the derandomization program, where of course the ultimate goal is to generate the advice in P.
Comment #27 November 16th, 2017 at 8:58 am
There’s no evident contradiction — that I can perceive anyway — between Scott’s perspective (#19 especially) and Alan Hardy’s perspective (#21-22 especially).
Quite the reverse, each of these two perspectives gains in scope and merit, when viewed in light of the other (AFAICT). Whence this mutual synergy?
Sociobiologist Ed Wilson’s books Consilience (1999) and The Meaning of Human Existence (2014) provide a provocatively acerbic starting-point:
More charitably, there’s no shortage of evidence in history to the effect that:
Accelerating advances in machine learning capability strikingly embody “new models of computation” — per the Avi Wigderson-reference of #20 — and these new models of computation are wonderfully (also fearfully) accelerating the rate-of-progress in many STEM/STEAM disciplines.
Scientifically speaking, recent decades have witnessed a growing appreciation that human social cognition neurotypically occurs by processes that are network-structured, both anatomically and computationally.
Which is not to say that social cognition does not very commonly occur by alternative cognitive processes — nerdily ratiocinative processes for example.
In view of real-world human cognitive diversity, it’s no simple matter to discern, or impute, or impose, any unifying structure whatsoever — any unifying mathematical, physical, economic, medical, or moral structure — upon humanity’s accelerating understanding of cognition and computation.
Speaking historically, what’s holding up pretty well (for me anyway) is old-fashioned Spinozism. Spinoza’s five-point program for Enlightened Modernity is as follows (from Edwin Curley’s translation of Spinoza’s Tractatus de Intellectus Emendatione, circa 1661):
Spinoza’s five Enlightened objectives are (of course) both technologically transformational and morally problematic.
Supposing that our global technological civilization does not revert to a pre-Enlightenment Dark Age, the next 50 years will witness more concrete progress toward a Spinozist Enlightenment than the previous 350 years — without much regard for whether, as individuals, we desire this progress or not.
This prospect is exciting of course, and scary too. It’s great to have Shtetl Optimized as a forum where, in regard to these fearfully tough issues, diverse points-of-view are respectfully and rationally set forth.
Comment #28 November 16th, 2017 at 9:19 am
Scott #25,
Just looked at that telegraph article and I don’t see how it is relevant to this discussion. First, if you steelman that article, the author’s objection is that we should not focus on curing diseases of old age because of population growth and demographic problems. That is of course highly problematic, but without knowing more and assuming bad faith I don’t see how to read that as hating on nerds.
Second, even assuming that the author is being disingenuous and really does just want to hate on Zuckerberg, what does that have to do with this?? The only connection I can see is that Zuckerberg is a nerd, and a rich and successful one at that. But he is his own person and maybe the author has reasons for hating on him that are completely orthogonal to his gender or his politics or anything to do with his superficial resemblance to your shy socially awkward nerd who can’t read social cues.
I would not be surprised at all if some idiot self-declared SJW somewhere had decided to take our current societal events to spew hatred indiscriminately and some innocents or socially awkward nerds were burned, but I don’t see the article cited as even remotely related. And, to be clear, I think regardless of current events that arsonists will burn. Those that practice anger and rage will have no problem finding targets.
Comment #29 November 16th, 2017 at 9:21 am
Joshua #24: Good question! When computational advice is structured as a tensor network (as contrasted with a Boolean Circuit), then the geometric properties of tensorial varieties facilitate efficient “sculpting” of the advice, via deep-learning methods that have no evident analog in Boolean circuits (as far as anyone knows, definitely including me).
Mike Nielsen’s free online book Neural Networks and Deep Learning is one starting-point, the deep-learning website Off the Convex Path is another, a review article by Haegeman, Mariën, Osborne, and Verstraete, titled “Geometry of matrix product states: Metric, parallel transport, and curvature” (2014) is a third starting point.
To quote the Roman poet Claudian:
which is widely translated as
Best wishes! 🙂
Comment #30 November 16th, 2017 at 11:49 am
It is not clear to me what Sensitive Nerd’s concern is. Is it about how he may be viewed by others or is he worried of unintentionally causing discomfort or harm to others?
Either way, I sympathize with the anxiety that can be caused by the news and social media. I suffered from it myself and maybe still do at times. It fuels my darker thoughts of self-harm. But I refuse to be taken over by this, because I do not think it is in any way rational.
It is not helpful to appeal to other people to adjust their rhetoric. At least not if your goal is that all of them will actually do it. There will always be people who are, or appear to you, unfair or overly vicious.
What you can do however, is give concrete practical advice to the Sensitive Nerd on what they can change to lessen their suffering. For example, if you have difficulty interpreting social cues and are worried about getting it wrong, stay away from trying to find romantic connections in public or at work. Don’t ever ask anybody out who you met in a work context. Just don’t. Don’t even consider or think about it. Don’t ever approach anybody you see in public with this intention. Just don’t. In fact, I don’t even think the latter works very well for anybody anyway.
Instead, date people where it is agreed, from the very start, that the purpose of the connection is of a romantic nature. Ask friends to set you up on dates. Use dating apps or websites. Be explicit about consent at every step. Don’t just kiss them. Ask them beforehand. Sure, many partners would say it is a turn off, but some may like it and for most others it will not be the sole reason not to date you. The ones that really really mind, are probably not a good fit for you anyway.
If you are genuine and do your best to respect the wishes and boundaries of others. You will be fine. If you mess up in some small way, apologize. You will be fine. Consider what women experience in interactions with the Weinsteins of the world. Do you think they can maintain a lifelong grudge against a Sensitive Nerd who is truly apologetic about a minor unintentional misstep?
Seek therapy if necessary to understand that it is not possible to please every single person in the world. There will always be men and there will always be women, some of whom will be vocal feminists, who don’t really know you and will find a reason not to like you or to disapprove of you or your actions. I don’t see any evidence of a widespread problem where these people have any significant impact on your life, if you simply ignore them. In contrast, I see a lot of evidence of a widespread problem of men assaulting and raping women and men so, understandably, this concern may be of higher priority to some people.
Comment #31 November 16th, 2017 at 1:08 pm
Parsely #30: Ultimately, you’re right. Ultimately, it’s not possible to please everyone; sometimes you have to take an action even though some people in your social circle (or SneerClub or Twitter or Tumblr or whatever) will denounce you for it. I.e., you don’t first need to refute the people and make them admit they misunderstood you; you can just … take the action (!).
But this is, by far, the most difficult realization I ever came to in my life, far harder than anything scientific I ever learned. It took me, basically, the first quarter-century of my life to learn it, and I still keep forgetting it and needing to remind myself.
So I have a great deal of sympathy for any sensitive nerds who haven’t yet come to the realization that they can do things that somebody will despise them for. And even while I help the nerds come to that realization, I’d also like to work a bit on the rest of the world, to try to stop it from mercilessly taunting nerds so much, or (the same thing, just with an intellectual veneer) writing thinkpieces about how “nerd culture encodes problematic and toxic assumptions.”
Again, none of this is for me: I have a pretty good life these days. But I owe it to my younger self to speak up on behalf of the nerdy kids I meet.