Lindsey Kuper

Made in Iowa. Nearly 100% biodegradable. PL, concurrency, distributed computing, and all that. Free as in interpretable. Informal in various ways. Assistant professor at UC Santa Cruz.

blog: https://decomposition.al
home: https://users.soe.ucsc.edu/~lkuper
pronouns: she/her/hers
cats: yes

Public Key

npub1j8ecq8rk6gcn7l9mfq2tj7wv0gseud23r4sa0afgxf83lrxs5nhsg0ncgu

NIP-05 Address

lindsey_at_recurse.social@momostr.pink

Profile Code

nprofile1qqsfruuqr3mdyvfl0ja5s99e08x85gv7x4g36cwh755ryncl3ng2fmcprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmntvn3p6r

Publishing to

relay.momostr.pink

Author Public Key

npub1j8ecq8rk6gcn7l9mfq2tj7wv0gseud23r4sa0afgxf83lrxs5nhsg0ncgu

Show more details

Last Notes

2026-04-09 23:03:41 UTC

"Carol's Causal Conundrum", the newest zine from my research group, is out today! 🎉 Read online and print your own free copies: https://decomposition.al/zines/
Thread! 👇
https://recurse.social/system/media_attachments/files/116/377/234/708/041/085/original/a615e7c2a194a35e.png

2026-04-09 16:16:50 UTC

Apparently the prospective students have changed their email approach from "write in a resoundingly hollow and stilted style to sound 'academic' and 'professional'" to "write to sound like a LinkedIn post made by an insufferable tool".

2026-03-21 16:13:43 UTC

The Languages, Systems, and Data Seminar (https://lsd.ucsc.edu/lsd-seminar/) is now looking for speakers for April, May, and June!
We meet at noon Pacific time each Friday, in person and on Zoom, to discuss interesting topics in the areas of programming languages, systems, databases, formal methods, security, software engineering, verification, architecture, and beyond. Most of our speakers are grad students.
Prospective speakers should register their interest here: https://forms.gle/H2TUv9cUPh29WeL79

2026-03-21 09:57:58 UTC

Paulette mentions in her lovely, heartfelt post (https://types.pl/@koronkebitch/116263964501595558) that she's in PL for the people. The same is true for me.
I'm teaching undergrad PL again starting in about a week, and I'm thinking that this time around, I want to use lecture time to feature PL people that I love. Maybe once a week, I could show someone's photo, say a bit about how I know them, and then read out loud to the class something written either by or about them that conveys a sense of who they are.

2026-03-11 04:55:15 UTC

- reply

Honestly, e-bikes have been the technology that has improved my life the most in the last five years. I live in a hilly place, and with my first e-bike a few years ago, I suddenly became able to commute by bike. Then, with the cargo bike last year, I suddenly became able to take my eight-year-old places by bike.
She sometimes actually hugs me while we're out on the bike. Best damn thing in the world, and it couldn't happen in a car.

2026-03-10 04:54:10 UTC

I forgot to mention this when it happened, but we recently hit 500 miles on the used cargo e-bike we got last summer! During the school year, we really only use it for short trips with our kid a couple times a week, but every one of those trips is one we didn't have to take in a car. 🎉 I love that sitting in traffic with my kid in the back seat is now a rare occurrence.

2026-03-07 20:28:17 UTC

another thing about this whole CS-professors-using-LLMs discussion is that I've been hearing some of my colleagues excitedly say things like, "now that LLMs can generate good code, I don't need grad students anymore!", which horrifies me because of what it implies about how they thought of grad students this whole time: as code generation machines, I guess?

2026-02-24 00:31:21 UTC

- reply

Are you talking about abusive, unethical managers? I suppose a lot of jobs would be more "fun" if you didn't have to worry about annoying things like ethics.

2026-02-24 00:25:54 UTC

just saw a take that was like "programming is fun and people management is not fun, so instead of automating programming with LLMs, why don't we automate people management???" and, sweet mother of god, people, don't make me tap the fucking sign

2026-02-12 00:27:11 UTC

- reply

sheesh, why didn't I think of this one?

2026-02-12 00:20:08 UTC

Question for distributed systems folks. Go with your gut and don't overthink it.
What word would you fill in the blank with?
"A distributed system consists of multiple independent _______ connected by a network."

2026-02-06 00:14:03 UTC

- reply

Do security researchers ever get confused as to whether they're the good guys or the bad guys?

2026-02-06 00:10:22 UTC

Just trying to understand the phrase "malware analysis evasion and counter-evasion" (https://dl.acm.org/doi/10.1145/3150376.3150378) is like evaluating a formula with nested negations. "malware" (bad!), "analysis" (good!), "evasion" (bad!) "and counter-evasion" (and also good!)

2026-01-24 17:39:52 UTC

- reply

I do plan to keep acquiring books, but let's be honest -- it's more space than I'm ever going to fill that way!
I was imagining when I got the philodendron that it would eventually grow to cover all the shelves. It's doing okay, but it really needs to pick up the pace.

2026-01-23 22:47:52 UTC

I've worked here for almost eight years and only filled up 2/9ths of the shelves. At this rate I need to work here for another, uh, 28 years before they're full. Or increase my rate of plant, tchotchke, and/or distinguished paper award acquisition.
https://recurse.social/system/media_attachments/files/115/946/826/561/249/567/original/8c46b5978454b45d.jpg

2025-11-28 17:06:07 UTC

- reply

"Clearly stated opinions, weakly held" makes more sense to me than "Strong opinions, weakly held". But I suppose it's not as catchy.

2025-11-28 09:05:36 UTC

- reply

In the contexts I'm seeing it, "Strong opinions, loosely held" is supposed to be a *good* thing! It really doesn't sound good at all to me!

2025-11-28 08:55:10 UTC

I've been hearing the phrase "strong opinions, weakly held" a lot lately, and it's been confusing me, because I always thought "strong opinion" just meant "strongly held opinion". What is a "strong opinion", if not that?
That led me to thinking about the phrase "strong reject". I guess these sentences *could* have distinct meanings:
(1) My *strongly held opinion* is that this submission should be rejected.
(2) My opinion is that this submission should be *strongly rejected*.
But do they?

2025-11-20 21:45:54 UTC

Here are two old handwritten papers by Robin Milner that [@jer_gib](https://functional.cafe/@jer_gib ) helped me track down, and a few thoughts on them. https://decomposition.al/blog/2025/11/20/where-simulation-came-from/

2025-10-27 18:11:33 UTC

I just donated US$150 to the Python Software Foundation. If 9999 other people do that, then there's their grant money. https://fosstodon.org/@ThePSF/115446659188615376

2025-10-20 03:30:56 UTC

did I ever tell y'all about @npub17wn…akt9's idea for a superhero called "Great Circle" whose superpower is that when he throws punches, they travel along great circle routes and therefore arrive faster than normal punches?

2025-10-14 22:22:12 UTC

- reply

okay, fine, it's something like "they're both stratified", but you should still watch the talk!

2025-10-14 21:50:08 UTC

- reply

To find out, read [@koronkebitch](https://types.pl/@koronkebitch )'s paper or check out the talk video!

2025-10-14 07:01:11 UTC

💖
https://recurse.social/system/media_attachments/files/115/371/225/082/734/573/original/acb793ab276fe961.jpg

2025-09-23 21:10:32 UTC

I think the talk accompanying a PACMPL paper ought to be fungible between all PACMPL venues in the twelve months following its acceptance. You should present it at whichever one that makes the most sense for you to attend, be it for visa reasons, climate reasons, calendar reasons, etc. Or don't present it at any of them, if you prefer. We already have a patchy, ad hoc implementation of this via "SIGPLAN tracks", but it should become policy, and there should be SIGPLAN-wide coordination on it.

2025-09-16 19:31:47 UTC

I wonder if there's some well-maintained LaTeX package you can import that has \hyphenation commands for lots of widely (and not-so-widely) used computer science terms?
This question brought to you by "of-fload" and "stan-dalone".

2025-09-16 16:59:43 UTC

What being married to @npub17wn…akt9 is like, exhibit #47358:
me: *takes apart a broken stud finder*
me: *shows Alex the innards of the stud finder*
Alex: That's really simple!
me: Yeah!
Alex: So, here's a question for you: is this "AI"?
me: ...
Alex: I'm going to argue that it is!
me: ...
Alex: Here, lemme take a picture. This is going on a slide.

2025-08-28 17:36:20 UTC

- reply

Nice, thanks!
Incidentally: I knew that somewhere I had a public GitHub gist listing ICFP keynotes over the years, and I wanted to see if this talk was on it, so I searched for `icfp keynotes lkuper gist github`.
DuckDuckGo: first result was the correct link (https://gist.github.com/lkuper/00e26e57d442cfe40f01afa5bb09cadf).
Google: no relevant results on the first page, and this delightful bullshit "AI Overview":
https://recurse.social/system/media_attachments/files/115/107/571/509/730/191/original/7a95c412032b1df3.png

2025-08-28 17:02:10 UTC

There's an aphorism that goes something like: "Program optimizations fall into two camps: (1) do something now so you don't have to do it later, and (2) do something later so you don't have to do it now." The first person I heard articulate this was Dave Herman, circa 2011, but maybe it goes back further than that. Anyone know if there's a source I can cite, even if it's just a mailing list or something? (cc @npub1sqx…y49c, [@graydon](https://types.pl/@graydon ), etc.)

2025-08-20 22:34:30 UTC

- reply

From the Conference Publishing author kit for OOPSLA: "The title should use headline-style capitalization (e.g., "The Best POPL Paper of All Time"). All levels of headings use headline-style capitalization."
We've tried to ignore this in the past, and they complained. So they really mean it.

2025-08-20 22:10:16 UTC

Email from a division director at the US National Science Foundation: We're having trouble getting the video from last week's webinar posted on the NSF website, so here it is on YouTube, posted on *checks notes* the personal YouTube account of our IT person.
I weep for the NSF, the US government website situation in general, and the web platform in general

2025-08-20 21:30:00 UTC

ah yes, the part of camera-readying a paper where they make you title-case all the section and subsection headings, making them look pompous and annoying as hell

2025-08-17 05:01:22 UTC

Sylvia said such a lovely thing as we were heading home from dinner out just now (with her on the back of the cargo ebike). She pointed out Orion's belt, and then she said, "One of the best experiences you can have is letting your eyes adjust to the dark and looking at the stars at night, while you're on a bike with your family." I agree, that's gotta be one of the best experiences you can have

2025-08-16 00:35:31 UTC

- reply

Another question for people who are clued in about Singapore: do you have suggestions for hotels that cost less than USD $215/night and are reasonably convenient for getting to the Marina Bay Sands conference center, where I'll need to be during the day? (That's what the cheapest of the "official" hotels costs, even considering the conference discount. Note that all of these hotels are still much, much cheaper than the actual Marina Bay Sands hotel.)

2025-08-16 00:29:27 UTC

- reply

I have students with papers at both ICFP and OOPSLA. People in the know: how reasonable is it to imagine that my student with an ICFP paper will be able to attend the OOPSLA talk even if they're only registered for the ICFP days, and vice versa? I've never known other SIGPLAN conferences to actually check badges before letting people into *talks* (lunch is a different story), but idk, this is Singapore and an infamously fancy hotel we're talking about.

2025-08-16 00:05:25 UTC

Hotel $1930.39 SGD
Flight $942.61 USD
Conference registration $2,310 SGD
will someone who is good at the economy please help me budget this. my funding is dying

2025-08-06 06:19:48 UTC

I just received an email about a web hosting issue that is within the scope of responsibility of a volunteer position that I stepped down from *in 2018*, but for which someone somewhere evidently still has me listed as the point of contact!
(I forwarded it to the current person in that role, who is the person who took over from the person who took over from the person who took over from me.)

2025-07-21 08:35:21 UTC

noted
https://recurse.social/system/media_attachments/files/114/890/295/248/042/109/original/6bfe0c6a84bc5934.jpg

2025-07-19 16:47:41 UTC

we got matching shirts
https://recurse.social/system/media_attachments/files/114/880/910/744/298/630/original/be85f96fa52f650c.jpg

2025-07-18 03:54:15 UTC

- reply

Alex: You will develop the powerful Rudnick legs that are your birthright. The legs that are good for running and jumping and...
Sylvia: ...doing squats.
Alex: And doing squats.

2025-07-18 03:51:58 UTC

me: Sylvia, ready to go take your bath?
Sylvia (age 7): Yeah, I just have to do a hundred jumping jacks to get buff. Let's get swole! *starts doing jumping jacks in sync with @npub17wn…akt9*

2025-05-31 04:40:25 UTC

A student asked me if I know of any research (perhaps experience-report-style papers?) about starting with lightweight formal methods (say, bounded model checking) and transitioning to more heavyweight formal methods (say, mechanized proofs in a proof assistant) for a given software project. I can't think of any offhand. Any FM or FM-adjacent folks have pointers?

2025-05-19 15:53:57 UTC

this little potato that grew out of the compost is still bigger than any of the ones I've tried to grow on purpose
https://recurse.social/system/media_attachments/files/114/535/292/611/286/901/original/b8f87dffde676851.jpg

2025-05-19 02:28:55 UTC

- reply

littler cactus friend is now also strutting its stuff
https://recurse.social/system/media_attachments/files/114/532/125/795/153/511/original/145e640adfa34bc5.jpg

2025-05-08 04:51:44 UTC

Great post on the SIGARCH blog: "The Academic Pipeline Stall: Why Industry Must Stand for Academia" https://www.sigarch.org/the-academic-pipeline-stall-why-industry-must-stand-for-academia/
We need your help, industry folks.
Very concretely and immediately: I want to make more zines like https://decomposition.al/zines/#communicating-chorrectly-with-a-choreography, and I want to fund an undergrad to work with me on them over the summer. It doesn't look like the NSF REU funding we had for our first zine is happening this summer. Is there a company who'll step up with $10k to fund a student to make zines with me?

2025-05-04 00:46:05 UTC

unlike the Eiffel dude, I am definitely going to complain about it if you use citations as noun phrases and don't use `\citet`. come on, people, get it together

2025-04-29 22:34:18 UTC

cactus friend doing its spring thing
https://recurse.social/system/media_attachments/files/114/423/614/162/654/165/original/be04d98dd9aa6b5d.jpg
https://recurse.social/system/media_attachments/files/114/423/614/354/243/969/original/ae7b6d8034351b8a.jpg

2025-04-15 16:02:04 UTC

if you're in Santa Cruz County and you need a house cleaner, you ought to go with Maxima's Cleaners because not only are they the best, but they also have an Instagram featuring a guy who is extremely hype and concludes every video by pointing at the camera and yelling "MAXIMA'S!"

2025-04-11 19:26:54 UTC

- reply

+1, thanks, Sumana

2025-04-10 04:33:16 UTC

- reply

fucking hell

2025-04-03 22:00:28 UTC

- reply

Second, I want to mention that we UCSC folks tried to publish our part of this work on our own previously and got rejected, and the UVM folks tried to publish their part on their own, and got rejected. But teaming up worked. The lesson for me is: collaborate more! Collaborate across institutions, collaborate across time zones, collaborate across areas of expertise. Do it even though it's hard. Do it even though CSRankings discourages it. Get that PLDI paper! 😤

2025-04-03 21:48:34 UTC

- reply

Just a couple other things to mention here:
First, Mako Bates, the first author of this paper and Joe's student at UVM, is looking for a job for after he defends his dissertation in June. He's seeking remote-friendly software engineering or research work, ideally at a place with an interest in protecting consumer privacy, and would welcome any leads or suggestions on where to apply. His background includes applied cryptography, type systems, and language design. His CV: https://irvingstreet.me/cv/

2025-04-03 18:54:23 UTC

- reply

My amazing former MS student, Shun Kashiwa, came up with "choreographic enclaves" -- or "conclaves" for short -- as a way to sidestep the inefficiency of conditionals in HasChor-style library-level CP. When we talked to the UVM team, it turned out that they had had exactly the same idea and were pushing it even further. We joined forces, Shun quickly incorporated the UVM folks' ideas into our work, and we were off to the races. It's been a blast to collaborate with the UVM folks!

2025-04-03 18:48:36 UTC

- reply

This paper presents *three* new choreographic programming libraries -- for Haskell, Rust, and TypeScript!
Our original CP library for Haskell, called HasChor, came out in 2023. Since then, CP libraries for a variety of languages have appeared on the scene. This is all very exciting -- we think that integration of CP into general-purpose languages via libraries is essential for CP's widespread adoption. But so far, library-level CP has lagged behind what standalone CP languages can offer.

2025-04-03 18:23:53 UTC

Now it can be told! "Efficient, Portable, Census-Polymorphic Choreographic Programming" will appear at PLDI 2025. 🎉
The HasChor folks teamed up with Mako Bates, Syed Jafri, and Joe Near at the University of Vermont for an "our powers combined" paper after we realized that we were all working on closely related things.
There's a draft on arXiv at https://arxiv.org/abs/2412.02107 -- although we still have lots of changes to make to incorporate the very helpful reviewer feedback for the final version!

2025-03-07 18:28:18 UTC

Santa Cruz folks: I'll be at the Stand Up for Science rally at UCSC today, along with several colleagues. Join us!
https://recurse.social/system/media_attachments/files/114/122/545/438/395/402/original/df00e27c649092b4.jpg

2025-03-03 21:26:17 UTC

Before y'all put your stuff on arXiv, do you first clean up the sources to get rid of all the snarky asides, commented-out bits that weren't deemed important enough to go in the paper, and other stuff that wouldn't be obvious in the PDF? Or do you not bother?

2025-02-22 18:07:31 UTC

✅ Author response off to PLDI
✅ Succeeded in unclogging my toilet

2025-02-05 02:33:52 UTC

We should be asking NSF exactly what "DEI principles" are. Make them say it.

2025-01-18 20:08:32 UTC

I forgot that the Spanish word for jellyfish is medusa, that's sick as hell
jellyfish: goofy, sounds like the name of a YouTube channel for toddlers
medusa: mythical being with deeply worrisome superpowers

2025-01-17 21:12:22 UTC

just described someone as being from "the extended Andrew Myers cinematic universe"

2025-01-01 03:36:02 UTC

To close out the year, I wrote about my friend Alan Jeffrey, who we lost in 2024. https://decomposition.al/blog/2024/12/31/alan-jeffrey/
In 2025 and beyond, may we all remember Alan fondly and try to live up to his example. ❤️

2024-12-27 03:24:03 UTC

It feels a lot like the stuff ChatGPT emits that *we* can see and use is actually a by-product, not unlike excrement, and the actual thing of value that it produces is just more wealth for its investors. https://mastodon.world/@Mer__edith/113721099488667070

2024-12-11 03:39:34 UTC

- reply

most of the students seemed happy to get the zines, and one student told me that taking my class, with its structured discussion process, had improved their English ❤️

2024-12-10 23:31:28 UTC

just printing zines to hand out at the final exam, normal professor stuff

2024-12-05 15:19:05 UTC

It's finally done! 🎉 I am so excited to announce "Communicating Chorrectly with a Choreography", the first zine from my research group!
Read online and print your own free copies: https://decomposition.al/zines/
Thread! 👇
https://recurse.social/system/media_attachments/files/113/600/873/948/701/630/original/5f81a8ecae7bf94d.jpg

2024-11-14 18:32:31 UTC

- reply

Your choice of abstract domain, { +, -, 0, i }, gets used a lot in abstract interpretation, too. (Although I guess Cousot and Cousot just decided to ignore 0 in the intro to their paper?)
https://recurse.social/system/media_attachments/files/113/482/694/006/768/722/original/2bc5c04c7daed08c.png

2024-11-14 17:53:01 UTC

I think some folks here will appreciate this:
A couple weeks ago, it was parent-teacher conference time. When we met with Sylvia's second-grade teacher, she looked at us and asked, "...how many cats do you have?"
We have two cats. But apparently, cats feature *so* prominently in Sylvia's writing, drawing, etc., at school that her teacher thought we had, like, *many* cats. It turns out she's also been writing about her friends' cats, the neighbor cat, etc.
My kid just really likes cats!

2024-11-06 00:10:33 UTC

Interested in giving a Zoom talk at the UC Santa Cruz Languages, Systems, and Data Seminar? Nominate yourself 👉 https://forms.gle/H2TUv9cUPh29WeL79
The LSD Seminar meets weekly to discuss interesting topics in the areas of programming languages, systems, databases, and beyond. Browse our talk archives (https://lsd.ucsc.edu/lsd-seminar/) to get an idea of the mix of speakers we've hosted in the past. We especially encourage talks from early-stage grad students! We're currently booking speakers for January or later.

2024-11-05 17:52:28 UTC

typical UC Santa Cruz morning
https://recurse.social/system/media_attachments/files/113/431/598/913/743/073/original/c0524e4e4a15b775.mp4

2024-09-19 18:11:20 UTC

- reply

More generally, I would love for there to be lots of !!Con-like events in lots of places around the world, and there already have been several. I wrote about a few of the early !!Con-inspired events back in 2017 (https://decomposition.al/blog/2017/03/31/scaling-bangbangcon/). And these days, there's GambiConf in Brazil, which is now in its fourth year! https://gambiconf.dev/

2024-09-19 18:04:45 UTC

- reply

This is a really late follow-up, but: !!Con 2024 was a hybrid event, and we did have people attending remotely from all over the world. We had a free live stream, and some folks at the Underline Center (https://underline.center/) in Bangalore organized a watch party (https://insider.in/con-syncwatch-party-bangalore-aug26-2024/event). The Underline Center folks are contemplating starting their own !!Con-like event in Bangalore, and if you want to be involved, you should get in touch with them!

2024-09-18 15:57:20 UTC

- reply

It's not "plain text"! It's got formatting! I think it's fine like it is. Loads fast, looks good on mobile with a minimum of fuss. If you think it's a little boring, you could do what I did with mine and throw in some emoji on the section headings.

2024-09-17 02:51:59 UTC

had an epic weekend that will be hard to explain
https://recurse.social/system/media_attachments/files/113/150/591/208/775/245/original/b122baf0ae0aa75d.jpg

2024-09-13 01:00:31 UTC

saw this majestic fella on campus the other day
https://recurse.social/system/media_attachments/files/113/127/533/875/679/334/original/dee2286e441ba2ff.jpg

2024-08-30 21:08:36 UTC

CS academic (or academic-adjacent) folks who plan to be around the bay area in October or November: let me know if you want to visit UCSC some Wednesday to give a talk at our CSE department colloquium! We're hammering out a schedule.
(The CSE colloquium is specifically for more senior folks who want to give talks aimed at a general audience of computer scientists, unlike the LSD Seminar, which is specifically aimed at PL/systems people, and where we try to have mostly grad student speakers.)

2024-08-20 16:56:34 UTC

Question for people who've presented posters at academic conferences: do you enjoy it?
I have never presented a poster. Personally, I like giving talks (performance!), and I like talking to folks in the hall (conversation!), but I think I'd find presenting a poster to be an awkward in-between performance-conversation. Am I wrong?

2024-08-07 04:41:02 UTC

We're seeking speakers for this year's Languages, Systems, and Data Seminar! 🎉
We meet on Fridays in fall, winter, and spring, in person and on Zoom. Most of our speakers are PhD students working in the areas of programming languages, systems, databases, etc.
If you've never given an LSD talk and you'd like to, nominate yourself: https://forms.gle/jGN6qUqKrQbABJvo8
We try not to have repeats, so if you *have* given an LSD talk before, pass the form along to a friend who you think would give a good talk!

2024-08-01 22:12:11 UTC

it's happening
https://recurse.social/system/media_attachments/files/112/889/053/708/477/343/original/d4c19fb009ae9479.jpg

2024-07-29 03:38:24 UTC

- reply

Anyway, all this is to say that if you were thinking about just powering through it, whatever "it" is for you, consider instead doing this weird new thing I'm trying out, called "resting"

2024-07-28 19:18:25 UTC

- reply

I've spent most of the last 24 hours in bed. It would have been terrible if I had actually gone! I probably would have landed feeling much worse (likely worse than I feel now, due to the additional strain of travel). I'd have taken a test, tested positive, and then been sick with COVID *alone in a foreign country* instead of at home. Not to mention that I probably would have infected other people all along the journey.

2024-07-28 19:09:14 UTC

- reply

Yeah, so I was supposed to be giving a series of lectures at the Scottish Programming Languages and Verification Summer School this week, but instead I'm home with COVID.
My husband @npub17wn…akt9 was the one who urged me to take a test, hours before I would've left. My suitcase was already packed. I had cold symptoms, but had been preparing to just power through it.

2024-07-27 18:04:18 UTC

well, shit
https://recurse.social/system/media_attachments/files/112/859/770/226/819/302/original/340d95644676c4ad.jpg

2023-10-14 00:21:22 UTC

Whenever we discuss a paper in my distributed systems class, I randomly split people into six groups and they spread out around the building to talk, while I walk from group to group. The groups are different every time. But so far, every time, I'm able to find five groups, while the sixth group goes off somewhere and I can't find or communicate with them until they eventually wander back to the classroom five minutes late.
I feel this is rather appropriate for a distributed systems course.