SIROCCO 2023

Some quotes from SIROCCO 2023, a CS conference earlier this month.

"Everything's computation. Any process evolving and solving problems over time is computation." - Yuval Emek. An expression of the "computation is a metaphor for everything" that I hadn't heard framed this way before.

"Artificial neural nets use all kinds of swanky functions." - Frederik Mallman-Trenn

"... so first we need to understand what a human is. I'm oversimplifying a lot, but a human has a head and a body..." - Frederik Mallman-Trenn taking the physicists' approach to framing a problem

"This is the tapas version of the talk... I'm going to be going over a lot of topics to give you a taste of them." - Joshua Daymude

"There's so many dimensions here that we have not yet found one algorithm to rule them all." - Joshua Daymude, expressing an unscripted pervasiveness of Tolkien in popular culture

"This is a crowd that lies proofs, so I don't want to show off my simulations and offend anybody." - Joshua Daymude

"Roger is going to be talking about 'swarm intelligence', if that even exists." - Andréa Richa, introducing the next speaker's talk about swarm intelligence with a devastating level of objectivity

"I don't know how to solve this and I'm afraid to go back and tell them this, so you should help me! I want to hear your ideas." - Roger Wattenhofer, with an earnest appeal to the crowd that I can only imagine comes from unassailable tenure-granted confidence and seniority

[many inscrutable self-notes about the legibility of different presenters' slides, diagrams, and presentation styles]

The mildest possible theory praise: "I think you're touching upon the challenges of this area." - Michael Schapira

"Deep learning is voodoo. We can't debug it, we don't know how it works..." - Michael Schapira

"How do I know in real time that my decisions are no longer sound?" - Michael Schapira, expressing a CS question but also a question which suits intoxicated people

"The nice thing about this algorithm is... but the very horrible thing is, it uses infinite memory." - Bernadette Charron-Bost, expressing a truly beautiful "theory person has found a tradeoff" provable fact

The phrase "we need to use more elbow oil" was mentioned and my brain took a minute to figure out that the non-native English-speaking presenter probably meant the idiom "elbow grease", which on further inspection is a bizarre and faintly disgusting idiom.

Discussing the topic of algorithmic recommendation systems: "It's maybe something questionable from the moral point of view. But it's certainly interesting and I will not talk about it." - Boaz Patt-Shamir

This post's theme word is: craic (n), "good times involving pleasant company, enjoyable conversation, etc." Many speakers reminded us that SIROCCO is the 'fun' conference and is full of craic and jollity.

If you could give Turing machines one bonus feature, what would it be?

I like to give students the opportunity to be creative. Partially this is because I am noble and committed to good pedagogy practices and all that; partly it is for my own amusement. (Students had to create Twitter bots for my amusement, too --- you can see them here, Twitter login required.)

If you could give Turing machines one bonus feature, what would it be?

Some people were practical:

• I would add a more streamlined ability to count.
• Ctrl-F find value
• Writing proofs for me.
• remember marked spot (instead of needing to mark)
• Its head can go to any position on the tape instantly without having to step through the middle steps.
• to be able to jump to a location in the tape not right next to where the head is

Others were silly:

• snack dispenser
• a cool spoiler (illustration below)

• funny hats (illustration below)

• dance a jig

Some were infeasible:

• Always halt for every problem. [editor's note: provably impossible, you impertinent youth!]

And some were downright haunting:

• Express feelings. It would be nice to know how TM feels when in runs forever. Would it be sad? happy? bored? [editor's note: how would the TM know it is running forever? it could just be churning and doing real computation...]
• consciousness

By far the one that struck the deepest chord and has followed me, hounding me into my dreams, was: "teeth." Yikes.

This post's theme word is stenophagous (adj), "feeding on a limited variety of food." The zookeeper found the dietary needs of TMs a challenge; they were among the most stenophagous charges at the mathematical-hypothetical petting zoo.

Mathematical property of friendship

As part of my lecture, I ask interactive live-polling multiple choice questions. Today, one of those questions was:

Let S be the set of Swarthmore students. Consider the binary relation "is a friend of" defined over S x S. This relation is symmetric.
(a) true
(b) false

The point of this question was twofold: first, I wanted to confirm that everyone was on board with the notation, vocabulary, and definitions involved in reading the question. Second, I wanted everyone to laugh --- the joke is that either answer can be correct, depending on your attitude about friendship. (The class was pretty evenly split across (a) and (b), with one outlier protest vote for (d).)

Imagine my complete and sheer delight when one student's defense of answer (a) was "This is a Quaker school, so we are all friends."

HAH

It is utterly, totally perfect as a response: it confirms that the student understood the concepts AND that they got the joke AND that they joked right back, indicating they're comfortable with the classroom environment already, in week 1.

Dear student: you made my day, 'twas a stellar reply, thank you very much for your in-classroom participation today.


This post's theme word is breviloquence (n), "speak briefly and concisely." English humor favors breviloquence.

Problems with self-reference and recursion (Aronson's sequence)

The most delicious, frolicksomely frustrating things to think about are the problems which reference themselves. Recursion is such a twisted mind-trap. Having just exposed my class to the joys of the halting problem (animated video explanation), and using it to show that all sorts of other problems cannot be solved --- one of the duties of professorhood is teaching students how to solve problems, but the peculiarities of my work are that I teach students which problems they can't solve --- I was delighted to read a snippet about Aronson's sequence:

‘T’ is the first, fourth, eleventh, sixteenth, twenty-fourth, twenty-ninth, thirty-third …

Here's the introduction on Futility Closet.

Here is Aronson's sequence on the Online Encyclopedia of Integer Sequences (one of my favorite sites!).

I want to know how the sentence ends, but of course the sentence can't end as long as I'm stuck thinking about the way I expect it to end. I'm sure that some sufficiently proficient linguist-mathematician team could come up with a satisfactory, and finite, end to the sentence. I'd buy that book!


This post's theme word is pabulum (noun), "bland intellectual fare: insipid or simplistic ideas, entertainment, writing, etc." Using the word "fare" makes me think of other food analogies. The collection of results stemming from Gödel's (In)Completeness Theorems are savory intellectual nuggets, with not a morsel of pabulum.

What your favorite number?

I take attendance by having the students answer a question, although that policy has come under [my] scrutiny and it will change a bit this semester. (I'm using clickers in-class so I already know who is there, approximately.)

What is your favorite number?

One student said i; everyone else picked an integer with < 5 digits. I'm not sure much valuable or amusing information can be gleaned from this, although I do have follow-up questions. If the question had been framed differently, it might have solicited this information in addition.

Favorite numbers that I think are explicable (or at least Google-able):

• 0
• 1
• 2
• 4
• 7
• 10
• 13
• 42
• 69
• 1995
• 2996
• 3301

Numbers that require additional justification:

• 29 (primality? permanent age of all famous people?)
• 32 (age? factorization? course number?)
• 37 (age? primality? course number? looks cool?)
• 816 (area code?)
• 20 (total # of toes? current age?)
• 71 (film title? pointy when written? year of something?)
• 77 (smallest possible integer requiring 5 syllables in English, apparently?)

This turned out not to be as interesting as "what is the largest number you have counted to out loud?" Question framing is so important.


This post's theme word is skint, "having no money; broke; poor". The question was imagination-skint.

Course message board

There is a homework due this morning.

The single most popular post on the course message board occurred last night.

1:10am: "do we have to finish the problem set even though Trump is winning? we have no will to continue."

Follow-up comments throughout the night:

• "please, I second."
• "Well with the way things are going, will there really be a GPA left for this to ruin anyway? "
• "fourth'd. why is this happening."
• "Murica is now NP Complete"
• "Please Lila"
• "make my GPA great again"
• "Not gpa that doesn't matter right now as much as general health and sanity"
• "not that wall street doesn't matter as much as minority citizens"
• "please reply, we're crying"

My reply upon waking: "It is very important to be educated, as this election highlights." We'll see if  I can remain as anodyne in my not-on-the-permanent-and-searchable-record aloud delivery of lecture this morning.

I'm considering scrapping my network flow lecture to discuss social choice theory and voting (lecture tentatively titled: "Arrow's impossibility theorem, or how math is the reason why we can't have nice things"). On the other hand, the problems we face in the real world are usually not theoretical problems of design and feasibility, but implementation detail problems (e.g., how to check election results) and issues arising from grandfathered-in historical systems which had no rational design to begin with.

This post's theme word is peripeteia (noun), "a sudden or unexpected change of fortune, especially in a literary work. A classic example is Oedipus learning about his parentage." Too crushed by peripeteia for cleverness, the characters shouted "I want out of this novel!" in a fourth-wall-breakingly desperate plea for clemency.

Courtship tokens

My desk already had fun math puzzles, as decoration and to occupy my hands and mind while I spin my brain-wheels on research problems.

To these I now add recently-received wooing tokens, these two octopuses.

As far as romancing goes, I cannot think of a more attractive feature than access to a 3D printer and willingness to print tiny, creepy cephalopods.


This post's theme word is cumshaw, "a gift or tip." This cumshaw octopus collects scrimshaw.

Pi(e) day (observed)

Certain calendrical conventions mean that today is not any particular day. But on the other side of a large body of water, there's a continent where dates are written hodgepodge, in no significant order. In recognition of this dubious date mismatch, I ventured forth and had a piece of pie.

This post's theme word is nugacity, "triviality, futility." The nugacity of the approximation does not dull the palate's enjoyment!

Valentine's Day words

A nice, depressing, abstracted rumination for Valentine's Day: "lovers cradling one another on the beach, murmuring the three words that are the highest expression of what they mean to each other".

It's not that I'm aromantic. It's just that I am romantically susceptible to cynicism, and intellectually vulnerable to torrents of well-punctuated words. I know my weaknesses.


This post's theme word is aleatory "depending on the throw of a die". Luck in romance is partially aleatory. (Of note: aléatoire means "at random", and is very useful for attending mathematical talks in French.)

Mirabelle pie

Not really on-schedule for π day (and in any case, the food/mathematical constant homonym doesn't work in French), but every day is a good day to make pie.

This photo shows 80% of my entire kitchen space, and was taken standing in the dining room space.

This one was an experiment with mirabelles, a type of yellow plum I had never heard of. I followed a recipe which instructed me to make a custard to fill the interstices between the mirabelles. It worked.

This post's theme word is cunctator, "one who hesitates, a procrastinator, a delayer." The cunctator celebrated Pi Day in July.

Pi day

Happy π day!*

Adapting to my locale, I used mirabelles (illegal to import to the US!) and a toaster oven to celebrate. The severe limitations of my micro-kitchen and single pie dish mean that I was limited to one celebratory dessert, and not the usual half-dozen-plus-board-game-party.

Happy dessert! Celebrated on nerd forums everywhere. You may also be interested in these resources: (1) Sad that pi day is over? (2) Rant against pi day: Vi day.


This post's theme word is diktat, "an order or decree imposed without popular consent or upon a defeated party." The Gregorian calendar's vagaries of calculation and representation compel the diktat: pi day occurs on a bleak, cold day in March, when pie fruit is least available in the northern hemisphere.

*It is π day for places that write the date US-style: month/day/year. For European-style dates (day/month/year), π day either doesn't exist or will take place the 3rd of February, next year (which, after all, will be the 14th month of 2015, in a manner of reckoning). I also like Vi Hart's "Engineer's Pi Day" on 3/14/16... see you next year!

Confounding science with science

Scott Alexander is resplendent in this blog post about science, statistics, confirmation bias, control groups, and the study of whether psychic effects are real. If that list of keywords is not enough to hook you, I really knew the article attained a blazing level of reading delight when I reached this paragraph:

Then there’s Munder (2013), which is a meta-meta-analysis on whether meta-analyses of confounding by researcher allegiance effect were themselves meta-confounded by meta-researcher allegiance effect. He found that indeed, meta-researchers who believed in researcher allegiance effect were more likely to turn up positive results in their studies of researcher allegiance effect (p < .002).

Everything about it is a delight. The layers of meta-analysis. The English noun-phrase-constructing rules that permit the construction of a sentence in which the prefix "meta-" appears five times, variously modifying words which themselves are modifying other "meta-"-modified words.

I wonder if the same researcher bias/confounding exists in fields where the experiments are entirely done on computers. Can researchers' belief in the effectiveness of certain machine learning techniques affect their experiments? What about physics simulations? I don't see how, but of course I deeply believe in the inviolable sanctity of mathematics. This is an opinion founded in my acknowledged bias. Maybe coders would self-sabotage by writing bad code, so that experiments run slower? ... but in the end this wouldn't affect the actual outcome, just the agony and feasibility of running the experiment many times.

On a larger scale, I am supremely happy that scientists are using their scientific reasoning to criticize the very practice of science itself. In the same way that I frequently remind myself that the basis of the field studying privacy is "trust no one"*, it would be nice to have big science conferences where we all get together and just shake our heads at how unreliable the current practice of science is. Apparently. I mean, check out this conclusion:

But rather than speculate, I prefer to take it as a brute fact. Studies are going to be confounded by the allegiance of the researcher. When researchers who don’t believe something discover it, that’s when it’s worth looking into.

... which sounds convincing. 

But.

You know what?

I'm skeptical. 

This post's theme word is obverse, "the more conspicuous of two alternatives or cases or sides." The skeptic and his obverse performed a coordinated, randomized, double-blind study.

*Or, as I memorably put it during a job interview, "We've known for a long time that almost everything is impossible."

Reasoning about information transmission

Every so often, I am struck with the realization that other people do not think the way I think. (This recurs, so perhaps it is a rerealization?) Of course this is obvious, but it is also too difficult to simulate everyone's unknown mental process all the time, and so I slouch into the lazy thinking of assuming everyone thinks the way I do. Everyone does this. (Don't they? I don't know, but I assume so... because I do...)

Here is a simple fact: probabilities are often unintuitive. Probably. Unless you are a technical person who thinks about them all the time.

Framing mathematics

No, not framing it for a crime. I just read the article A Revolution in Mathematics? What Really Happened a Century Ago and Why It Matters Today by Frank Quinn (math professor). It establishes a surrounding explanation for how modern mathematics came to be structured as it is. In the process, it contrasts two ways of thinking about mathematics: the "old" style, and mathematical sciences, wherein math relates to observable facts and is intuitive (but often yields incorrect results), and the "new" style, which he calls the "core" of mathematical research, wherein math is the study of abstract rules which do not relate to reality (but whose results are provably, rigorously correct).

It is fascinating.

I had never paused to consider the evolution of my discipline. Yet Prof. Quinn highlights and summarizes my experience of grade-school math education: everything seemed much clearer and more reasonable when -- finally! -- worked as abstract symbols according to rules. This is how I learned geometry (my first proofs!), trigonometry, and calculus. I cannot imagine attempting to learn calculus through intuition. What terror! (Does this infinite series feel like it converges? What's your hunch about the derivative of f(x)?) And of course now in retrospect I think of the math I learned earlier -- multiplication, fractions, arithmetic -- in the more advanced terms I learned later.

The article was summarized for me by this: "the old dysfunction was invisible, whereas the new opacity is obvious." Yes, math is opaque; I've studied for years and this is the first thing I'd admit. And my topics are squarely in the "new/core" section: I've done research in precise definitions, logical proofs, completeness. Carefully justifying each step is a technique that I use in my dreams. Math for me has always been its own arena of knowledge, one of three (the humanities, sciences, and math), each with its own methods. Even though we use the science word "discovery," a mathematical discovery is nothing of the sort. And as a grad student, I am amazed at what other grad students do as "research."


This post's theme word is anemometer, "an instrument for measuring the speed of wind." This magical anemometer predicts the trends in sociology research!

Another dream

I dreamed last night about different approaches to try to compute this calculation I am working on. It's a real calculation, that I am working on in my waking hours; it made for a rather unexciting dream. In the dream, I was just as stymied and frustrated as I was in real life.

Sometimes my dreams are just a rehashing of the day.


This post's theme word is scoria, "in metallurgy, the refuse or slag left from smelting," or "porous cinderlike fragments of solidified lava." The scoria of my mind comprises calculations, symbols, and silly wordplay.

Confident daughters

So do I. Perhaps I'll try to make them tactful as well.


This post's theme word is asperity, "harshness or roughness."

VC-dimension?

Do you know anything about computing VC-dimension? Or about the Sauer-Shelah lemma?

Both Wikipedia and the CC blog state the lemma in a way that I do not understand. Namely, they define the VC-dimension as, at some point, determined by the cardinality of the intersection of the set {x₁,…,x_k} with some other set. And both sources upper bound this number by 2^k, which seems absurd: how can it possibly be more than k? We have indexed the set; it has exactly k elements; set intersection is a strictly nonincreasing function in terms of set size.

It also seems unlikely to me that both these sources would have typos. So I must be misunderstanding something.

If you know something about VC-dimension, the Sauer-Shelah lemma, or related combinatorics, please send me an email and I shall grill you with my questions (seasoned with confusion... and paprika, if you like).

Update: I figured it out with the help of these lecture notes. It was an issue of misunderstood notation. (My bad, as usual.)


This post's theme word: curtilage, "an area of land encompassing a dwelling and its surrounding yard, considered as enclosed whether fenced or not." In my present frame of mind, it relates to set theory.

Building Turing machines

Two recent items of internews (internet+news = stories that are popular for a few hours) regarding Turing machines, my second-favorite type of machine.*

Massive in-game Turing machine built inside a mechanical game (via BoingBoing).

Actual real-life Turing machine built inside this universe (via BoingBoing, HackADay, Slashdot, Gizmodo).


This post's theme word: sedulous, "involving great care, effort, and persistence." Any Turing machine construction, whether engineering or mathematical, is of necessity quite sedulous.

*Printing presses. Nothing can beat 'em.

Happy e day!

Happy e day!

Remember pi day, back on March 14 (3.14)? Well, allowing for some creative calendrical interpretation, today is e day, the 71st day of February (2.71). I suppose that if you were to round instead of truncating, e day would properly be tomorrow, the 72nd day of February.

"But it's April!" you protest.

That is true. However, it snowed last week. It still feels like February psychically. Why should we discriminate in favor of the small elite of constants with fewer than 32 hundredths?* You constant bigot.

"What should be done to celebrate?" you query.

I don't know of any e-themed foods, games, or ceremonies. So please, just take a moment today (or tomorrow, if you are a rounder) and think of how lovely e is. For some light introductory reading, here's the Wikipedia article and many ways to represent e.

*For that matter, why do we limit ourselves to celebrating constants <13, based on our 12-month calendar?


This post's featured comic comes from Three Panel Soul:

Ada Lovelace Day

Tor.com has a fantastic animation up for Ada Lovelace Day:I don't have anything to add; honestly, my role models have been male. There are many technical women I admire, and they have been duly written-up elsewhere for this created holiday. See for example this, this, this, this, this, or this. Or this entire webcomic devoted to Ada Lovelace.


This post's featured comic is from A Softer World: