A Problem Squared - 140 = ABBABAAB and AI-Bah! 

Episode Date: August 10, 2026

⚽ What’s the fairest way to pick teams?🤖 How do you discourage colleagues from using AI to write everything?🚗 And this road trip is jeopardised by playing ‘Accelerator or Break?’Thanks to all of you who sent in answers to “what are you doing while you’re listening to A Problem Squared?” - we go through as many as we possibly can in this episode!Here is kram1032’s post on Maths Stack Exchange that helped Matt with his problem solving:https://math.stackexchange.com/questions/2078609/whats-the-fairest-turn-sequence-for-n-players Here’s more info on Open Sauce: https://www.opensauce.com/And here’s a kind stranger’s Insta reel of the Ratatouille robot: https://www.instagram.com/reels/DbGwSh4vHQq/ And here’s more info on UKIE: https://ukie.org.uk/ And the stream of Bec hosting the event: https://www.youtube.com/live/4lw_ypXroa4 Head to our BlueSky/Twitter/Instagram to see the U-Haul-Truck-Camera photo of Matt. If you’re at the Edinburgh Fringe and would like to see Bec’s Work In Progress show, tickets are available here: https://www.edfringe.com/tickets/whats-on/bec-hill-s-work-in-progress Matt has extended his tour - and he’s filming the show in London in October - all the dates, venues and tickets can be found here: http://standupmaths.com/shows Want to pose a problem for Bec and Matt to solve? Head over to https://www.aproblemsquared.com/ Join us on Patreon for early releases and our monthly bonus podcast I’m A Wizard!If you’re already on Patreon and have a creative Wizard offer to give Bec and Matt, please comment on our pinned post!   If you want to (we’re not forcing anyone) please do leave us a review, share the podcast with a friend, or give us a rating! Please do that. It really helps. Finally, if you want even more from A Problem Squared you can connect with us and other listeners on BlueSky, Twitter, Instagram, and on Discord.

Hello and welcome to a Problem Squared, the problem solving podcast, which is a lot like a road trip. Yes, there will be several interesting detours, which means it will take longer than necessary to reach the point we are trying to make, but arguably that's the reason why you've chosen this vehicle in the first place. I'm Beck Hill, a comedian, writer and presenter who will be driving the show today, despite the fact that I still can't legally drive, and I am joined by my co-host, Matt Parker, who is a mathematician, YouTuber, and quite frankly, too relaxed passenger, given the circumstances. Obviously, because you don't drive. I have no actual real world experience of what it would be like to be in a car that you're in charge of.

I'm not too bad, actually. I'm pretty careful. Like, aside from once I accidentally hit accelerate instead of break, and obviously immediately realized and then broke. That's very funny, because I'm pretty good at driving. a car, apart from occasionally, I'm like, what, you put the window down by accident, you turn the stereo up too loud, no, just occasionally the wrong pedal, you know. You do that when you're learning. You make mistakes when you're learning.

Oh, that's true. If we didn't, then why would we have to learn? Yep, no, good point. By now, I'm better. I don't do that as much now. Basically, I'm trying to reassure everyone that it's cool if you want to teach me how to drive. Oh, hello? Yeah, yeah. Oh, sorry, you went very quiet. I think that was Matt really just not offering to help you learn to drive. Yeah, you really had that vibe.

Oh, my goodness, yes. Thanks for bringing it up, Laura. And on this episode, I've expanded on the fairest way to share. I'm going to take on AI and why I hate it. And we have plenty of AO, which is, accelerator or break the game show. Hello Matt. Beck, we're back.

We are recording remotely again. Yes. After a three week gap. So sorry everyone. Yes. Thank you for joining us, but we'll make it up with an episode next week. So you were in the States recently and then you rudely didn't get back until I was then on the road. Correct.

Yes, I came sliding back into the country on Monday, by which time you were already off. Yeah, exactly. I had a wonderful trip over to the States, and I went to a conference called Open Source. A source is in the condiment, not the philosophy behind writing software. Yeah, because I was going to make a joke about that, but it turns out that's exactly what it was. Yep, way ahead of you. And for people I'm familiar with open source, which is probably to be most people, this happens just near San Francisco's in the Bay Area, Silicon Valley. And it's best described as a science fair for adults.

So there are a lot of people who make ridiculous things or do unusual experiments, and they're producing all these incredible things, but nowhere to display them. Yeah, okay. Like their little scientific experiments from home and their own little diaramas and inventions. Yeah. I love that. I would love to go to that. You should. Oh, you would have had a great time back.

Yeah. Several halls of people just, like someone just made a comically big mechanical pencil. But like fully functioning 10 to 1, maybe more than 10 to 1 actually scale. It would have been about 2 meters. Oh my gosh, that's incredible. With like proper scaled up graphite, you click the end and it moves down, the works. Just incredible. Oh, I need to go to this.

It's that through to people who had motorized a sofa, so it was remote control and could drive around. Yeah. Someone who used servos to just balance an inverted pendulum. I was like, that's amazing. And then you could balance things on the pendulum and they were just kind of, it would correct. and balance them. It was really, really cool. Oh, that's cool. I don't want to pick a favourite.

And if you were there and I came and saw you, because I tried to visit everyone to see their thing. And I chatted to you in a wonderful time. I'm not skipping you deliberately. There's just so many things to go through. For example, someone, you know, Rattatooie. Yeah. Someone built a robot of, I forget the human character's name.

Remy, I want to say? Linguini. Guine is the name of the guy. Remy is the name of the rat. Oh, there you go. Excellent. Which is somewhat. The normal names.

Yeah. Is that like calling your American character hamburger? Yeah. Which we should. We should. All right, Veggie mite. Listen up. So, they had, like, robotic arms on a life-size human, humanoid model,

but, like, made look exactly like the camera. character and there was like a knife in one and like a holding fork thing in the other. And you could control the robotic arms purely through adjusting the hair on the robot. So you could fully steer the cooking process from from hair adjustments. It was very clever. That's, okay. Please tell me you've got footage of that or some kind of photo. Oh, there will be.

There will be footage somewhere of that happening. Producer Laura has already found a video that someone put on Instagram of the Ratatoui robot, so we'll put a link to what I'm watching in the show notes. I actually spend a bit of time wandering around on like the night before it officially opened to the public because it was so crowd. There were like tens of thousands of people came through to see everything over the main two days. And on the night before it officially opened, All the exhibitors were in, like all the industrial people, VIPs, etc.

There was a conference inside the convention. So the convention was like open source makers, people building stuff. But then they just invite a whole bunch of YouTubers to come and hang out. And there's kind of just this secret, not secret, but there's like an embedded convention of a conference, let's say, of YouTubers hanging out. And so we were all zipping around the night before. some poor exhibitor had a Sony, I would pronounce it, Mavica, Mavica. It's a camera that Sony made in the late 90s

that you'd have to put a floppy disk into it to take a photo. And they'd gotten one and altered it, like modernized it so they could take the data straight off it and use it as a webcam and they could print out the photo at the same time onto a label to then put it on the floppy disk is they could put a floppy disk in the camera, take a photo with it, take the floppy out and then label it with the photo that's now on the floppy

that was just taken by the camera. Oh, great. I was wandering around with your friend and mine, Tom Scott, the night before. Oh. And we both saw this stall and the person hadn't noticed us. And we didn't realize this. We both had that camera back in the day around the year 2000. And it was for both of us, a very influential camera.

in our digital journeys. And so we immediately descended on this poor guy. And so he just looked up and there's me and Tom Scott just rolling in telling stories about this poor camera. So the guy took it all on board pretty well and took our photos with a camera. And I got on a floppy desk. It was a lot of fun.

The best photo, some folks had turned an entire U-Haul van into a camera. in that... What? They'd like... Like a camera obscura? Yes, but they'd gotten a 1960s enlargement lens. So instead of like a pinhole camera obscura,

it was a proper lens camera obscura. And you could climb into the camera and then see the lens like poking out the back of the van and then it projected the image across the inside of the van onto big like photography full-size paper that it would then expose. And then they'd do some of the development inside the van. So then the camera became the dark room.

And then they'd finish the outside. And, but you had to hold still for a very long time. So I have my photo that I got taken. So. Oh my goodness. The, the question I have for you, Beck. I'm going to say, what pose would you choose? Well, I don't know what emotion I'm expressing.

So I'm going to show you, I'm going to show you this photo. and you can tell me what motion you think I've gone for. Let me pull it out. Here we go. So I will say for some reason I was expecting a photo the size of a van. The camera is the size of the van. It's still big. It's still big.

This is what, A3? It's roughly 28 centimetres by 36 centimeters. It's a big. Yeah. So it's too big for your post. passport is what I'm trying to say. Yeah, too big for my passport. It's a beautiful photo.

It's very cool. Now, you're asking what I think your expression in this photo is. Yeah. And all jokes aside, I think this might have captured your soul. Like, this is the most artistic. Like, if I was trying to portray, like, if I was trying to put all of your multitudes into one pose. This would be it. Like, because you look happy.

Like, there is some happiness there. I am happy. There is some slight concern, but not like, it's not a disapproval or discomfort. There's sort of a weary, like, there's a bit of tiredness in there that I'm getting. And you're content. It's pretty accurate. You're content, but also a bit overwhelmed and carrying a lot. That's the vibe I get from this photo.

And I would say it as a joke, but I genuinely think that might be who you are. Turns out old-timey photography, we have to stand still for like 30 seconds. Captures your soul. There you are. We'd cracked photography out of the gate. The mask can't be held up for as long. And so eventually you just get who that person is in the moment. Well, I think it's one of two things.

You either have to just be what you. you are in the moment so you can sustain it for that long or you've got to go full like a massive cheesy grin or something you can just lock in like you either go very real or very fake but i think if you went very fake you would be able to tell it's fake like i i have a go-to pose for my photos and in the moment it's fine but if i've had to do it loads or hold it you can tell in my eyes oh 100%. yeah so that's what i went for i I went for, you know, how I was in the moment. Yeah, and I like it because your eyes don't look fake. There's nothing fake about this photo.

I feel like this is a very real photo, and I think you've got yourself a really nice piece of art there. This is very cool. I'm going to put it up. And it's cool because the light from my face hit this bit of paper and exposed that image into it. So, you know, it's very cool. Yeah, it's very pretty.

This needs to go at the top of a staircase. Okay, okay. Well, I'll get it framed or something and then you can come around and consult on where I put it in the house. Yeah, toilet. And how have you been back? You've also been all over the place. Yeah, yeah, I went to Berlin Fringe. And I also got to host an event for Liquid Crimson and Yuki, which is the UK interactive entertainment group.

And they were basically celebrating all of the British games that have come out this year or due to come out soon. And it was the first time they've done an event like that. And it was delightful. I had such a fun time. It was really great. And it was the first one they've done. So I didn't get as much time to sort of go about and see all of the, play all the games and everything, which I was really hoping to do. So hopefully next year, we'll all know what we're doing by that stage.

but it was really, really fun and of course bumped into a few A Problem Squared fans. It's always a pleasure. I always love it. And it's amazing. There was a CEO of a very well respected game company who was about to go on and be interviewed. And he said, I saw you on a train once. And I texted my wife and said, I'm on the train with Beck Hill. And I said, oh, you could have come said hi.

And he said, oh, you're talking to someone else. And we realized, based on the train, that I would have been coming back from recording a problem squared. So I would have been chatting to producer Laura. That's very funny. They filmed the event as well. So if anyone's like, oh, I'd like to see that. I do believe there is a stream up.

I'll see if producer Laura can pop it in the show notes. If anyone's interested in seeing what UK games are coming out. But you're talking about floppy disks, Matt. Yes. And there's one. game that's coming out where they've crowdfunded it and the game is going to be released on a floppy disc. Love it.

But the creator revealed that when you flip it out, there's a USB. It's basically a big USB stick. And when he was talking about it in his interview, he said, oh, the floppy has a surprise and then revealed the USB and everyone went wild and it was great. And so afterwards I said, it's not often that a man on stage says his floppy has a surprise and it gets a round of applause. Tickled me. I bet.

But just to repeat, if anyone ever sees me in the wild and I don't look insanely stressed, you're always welcome to come and say hi. I love it. I'm not saying that that's how everybody feels, but I will always welcome people coming up and saying hi and introducing themselves. And likewise, if you see me out anywhere, you can always tell Beck that you saw me. Absolutely fine. Our first problem comes from Marlowe, who went to the problem posing page, which is at a problemsquare.com.

And they chose the drop-down problem and wrote, I participate in fantasy football. You know what, that is a problem, Marlowe. Thank you for admitting it. Oh, wait, they go on. That's so mean. There are 14 of us. I'm just so over football after the World Cup. Oh, right, right.

So Marlowe said there are 14 of us And we've adopted the ABBA method Of switching draft order for each pick Which I'm assuming is when you chuck on some flares and a long wig The winner takes it all Hey nice Thank you, thank you The ABA method I think that's come up in this show before

And from memory that's when Like let's say that you and I are choosing teams Matt I would choose first Then you choose second But then you also choose third Correct. And then I choose force. If you're doing ABBA, does it go ABAB, ABAB?

It goes ABAB, BABA, ABBA. Got it. Then it flips again. Then it goes Bab, ABBA, ABAB. Anyway. If you've just joined us. Yes. Let's get to the end of Marlow's problem and then we'll ABBA.

Okay, okay. So they've adopted the ABBA method of switching draft order for each pick, but this feels a bit unfair to a team of 14. and so there have been some ideas on how to make this mathematically more equitable. Tu-a-Morse? Am I pronouncing that correctly? Correct, yes. But it's spelt, T-H-U-E. Yeah, it's Norwegian.

Du-A-Morse sequence has been mentioned, but we think that's only relevant to two-person groups. Also, some sort of fourth round triple pick for Team 14, but then confused how to follow it up. Would you be able to help us? Big fan have listened from the beginning-ish. I can't bring myself to say blah, blah, blah, blah. Yes, but we could say Ababa, blah, blah, blah, blah.

Yeah, we need the blah, blah, blah. Yeah. There you go. Yeah, so I was trying to remember if we had talked about two-way morse and Abba before on the podcast. I don't think we've generalized it to like 14 people, which is basically what Marlowe is asking for here. If you've mentioned two-way morse before, I have not seen it written it down, and I had assumed that you were saying two-way as in like, two-way walkie-talkies. Oh, yes.

Okay, okay, yeah, no. I'm pretty sure this was the early 1900s. Norwegian mathematician two-way and an American mathematician Morse came up with what is just... As in Morse code? Oh, I don't think it's the same... Oh, I don't know.

I assume it's not the same Morse. But given they're making coded sequences, it might be. So that feels like a producer Laura problem. Well, um... I'm on it. Release her onto that. But it's often called the Abba sequence. because if you have two people, A and B, instead of alternating A, B, A, B when choosing,

which if you think about it, means A is always going to get the first go each round. So A has a go, then B has a go. Then A, then B. And if you were to look at it in terms of, let's say the things you're choosing are of decreasing quality, A on average is going to get a better result. But if you were to switch it and do a B-B-A, it kind of cancels out. So a very easy way to think about it is, let's say you're doing a classic, you want to get the lowest score possible.

And you're just picking from the numbers, you're just picking, the first person gets one point, second person gets two points, three points, four points. If you were alternating A-B-B-B, person A-B would get one. and then three and B would get two and four. So person A got one and three, that's four points. Person two gets two and four, that's six. That's way worse. Yeah.

But if you did ABBA, first person would get one, then the second person would get two, three, and then the first person would get four. And they've both got five points. You're like, oh, yeah, yeah. That's nice and neat and very fair. The other way to look at it is if you were to suddenly stop picking,

at some point. And if you alternated A, B, A, B constantly, you either stop after A, B, in which case they've had one more go, or you stop after B when they've had an even amount of goes. There's no way you can stop where B gets an extra go. But you can if you use the other sequence. So it's a much fairer sequence for a couple different definitions of fairer. Yeah, because like if you were choosing players, and there's one player who's like got 10 legs and never runs out of energy.

Can I just check? I try to think of what the best football player would be. Yeah. In the Beckerverse is someone with 10 legs the best or the worst at football? I'm taking they're the best. They'd be pretty wide. They would be pretty wide. Yes.

And so that's it'd be hard to get past them. Okay, that's what we're doing. Okay. Legs is a metric for width. Yeah, yeah. So anyway, if the first, like, but they were phenomenally. better. So like, team A chose that player, then, you know, obviously team B is going to get some the next two players, the next two best players, but like if the first one is incredible,

yeah, then it's still, but it's rare that there's that much of a difference. And in some of the extra notes that Marlowe sent in, the quality just drop off quite quickly, apparently. And 2-A morse is actually robust under those circumstances. Like, it doesn't even have to be a linear drop-off in quality. It can be some kind of exponential drop-off in quality, and it still is fair. So it's a surprisingly good system. And a lot of people, like the playground rules, tend to approximate it to make it easier to use.

So your proper 2-A morse, what you do is, is you've got to keep taking whatever sequence you've done so far and flipping A's for B's. So what you do, first of all, is you do ABA, which is A, B, B, A, B, A. A. Then you do Barb, but then you repeat the whole thing but flipped. So then you do Barb, Aba, but then you do the whole thing again, but flipped. So you do Abba, Barb, Barb, Aba. Then you do Barb, Abba, Abba, Barb. But then you do the whole thing again.

flipped and then you do the entire new thing again flipped so it's a big recursive algorithm that generates this longer and longer sequence and i really want one of our listeners to mix that into some sort of dance track i don't want that to happen doing that whole sequence is is complex so i think a lot of the time people just do abe barbab and then just repeat that ababababab which is fine it's not as good but it's easier to do or often people just do ABBA over and over. You know, one person picks first, then the second person picks twice, and then the first person picks twice, and then we just has two picks from then on. Or sometimes, I think just the second person gets two picks, and then you go back to alternating, which is like

a very lazy version of doing it, way easier to do, and you still get the ABA benefit at the beginning, and then you revert to the AB sequence. So there's different cutoffs where you want to balance complexity of the sequence with ease of doing it. So for a small amount of effort, you can still get a lot of the benefits. You don't have to go to the full distance, which is foreshadowing for the 14 person case. So when Marlowe says there's 14 people, because now that we've talked about team picking, I just need to remember how football works. So when Marlowe says there are 14 people, is that 14 people?

putting together their own teams. Yes. We're using A and B for two different people, but obviously there's 14. Yeah, it would be A through N, which I kind of like, because it means there's N people pleases me immensely. But that's coincidence.

So now you're like, well, how do you do a sequence? You do A Picks, B, Picks, C, D, E, F, G, H, J-K-L-M-N. So that's just all 14 people in order. and they can pick any way of randomizing everyone to start with. Like you draw names from a hat or something. You get some random ordering and you do everyone. Whoever's picked first, you're now A.

Whoever's picked second. You'll B. And everyone gets to pick one. But then what do you do? So they're trying to work out how do you do an ABBA equivalent sequence but instead of on two letters on 14 letters or indeed on any number of letters. So I was like, oh, that's interesting. I've thought about it, but I've never really looked into it.

And so I figured there must be some version of how the ABBA sequence is generated that generalizes to end people. Because there's multiple ways you can define ABBA. I did it as like you repeat the sequence, but invert it. And then you repeat that recursively. that Abbas sequence is also, if you count up in binary, so you count one, two, three, four, five, but each one's the binary version. Let's just, if there's an odd number of ones in the binary representation of the number you put an A, if there's an even number you put a B, generates exactly the same sequence. Lovely better mathematics.

There's also some substitution versions that generate the same sequence. There's a bunch of different ways you can define it. You can get the sequence multiple ways. And I was like, oh, okay, one of the ways we get, the ABBA sequence must lend itself to this. And then I found on a stack exchange discussion from 2016, which is slightly interesting because I did a video on 2A Morse, but I did it in 2015. So this is since I last really did a bunch of research into this. And just for clarity, 2A Morse is Abba. That is Abba.

Yeah. Sorry. That's just the mathematical name for the ABA sequence is the 2A Morse sequence. It's identical. Got it. I should just say Abba every time. But if you want to look up the mathematics, look up 2A Morse. T.HU.E.

T.HU.E. Morse. Oh, did we find out if it's the same Morse? We did. It is not the same Morse. So 2A Morse is Marston Morse, an American mathematician, and the inventor of the Morse code is Samuel Morse, who lived a hundred years before.

Oh, there you go. So it's very different people. And Inspector Morse is a fictional detective. Yeah, yes, yeah. Oh, I probably could have guessed that. That's a good point because Morse was used on Telegraph in the 1800s, well before 2A Morse in the early 1900s. Sorry for wasting your time.

I enjoyed it. So I looked up this discussion on Stack Exchange and I'll link to it. And someone, interestingly, there was a new way to get the Aver sequence I hadn't come across before. Their example is imagining you're doing, let's say, free throws. Like you're taking basketball shots between two people. And you want to work out what's the fairest sequence for everyone to take their shots such that first person to make one wins. And to do this, you actually imagine there's a very, very small probability

either player gets it in. So actually it might be more like attempting to get a hole in one in golf. But what's the fair sequence to do that? And they just then did that with the probability. So what you do is one person has a go first, and then after that, the rule is, whoever is least likely to have won by now gets to have the next go. So the first person tees off and they miss. Now the second person's least likely to have won because they haven't had a go. So they then get to have the second go.

But then they're still the least likely to have one because there's a chance. the game could have finished after the first shot. So they then get the second shot. But now, the first person's least likely to have won, because if you factor in the probability of an ending for each possible sequence and then the chance of them making it, they're next up. And it's a very complicated way to generate the sequence, but it does generalise to any number of people.

Oh. So you can then run the same algorithm over 14 people, and you can generate the sequence, the equivalent of the sequence for 14. And hopefully, someone had written some Python code to do that. So I was like, great. Okay, good. Because I was just going to try and understand it. Well, yes.

So huge thanks to math.dstockexchange.com user cram 1032, who initially answered this on the 3rd of January 27. but has gone back and updated and edited their answer and improved it multiple times. Most recently, March 2025. So over almost a decade, they've been maintaining this excellent answer with code and plots and explanations and maths. It's really nice. So great work, Cramm. And Cram's code just generates the sequence that way.

So I took the code, generated the sequence. and what's interesting about it is it starts simple and then gradually gets more and more complicated. This is why I was setting up earlier that sometimes you want to balance effectiveness for ease of use because the ease of use

gets very difficult at one point but I think at that point we've already got most of the effectiveness so we can cut our losses. So here's what you do. you first of all do your random ordering of everyone. So in this case you do everyone A through N. And then you do everyone in the reverse order.

So in that regard, very similar. So first of all, you go all the way up the list of people. And then the last person gets to immediately pick again. So the last person gets two in a row. And then you work in reverse order all the way back down again. Okay. Then once you finish that, you basically reverse it. So you do reverse order again.

And then you do beginning order. So you do everyone in order. Then you do reverse. Then you do reverse again. Then you do in order. Which is ABBA, but it's the whole sequence. So everyone forward, everyone back. Everyone back.

Everyone forward. Then it changes. Then you do the sequence again, but you start in the middle and you alternate your way out. Oh, okay. So the middle two people, which, I mean, in this case, case is H and G. And then you go out one and you do the next two people out, then the next two people out all the way out. And so at the very end, you do N and A at the very, you know, ends of the sequence. Yeah. Then you reverse that. You start at the very ends of the sequence and alternate

your way. So you go, first of all, you do beginning, then you do end. Then you do second from the beginning, then second from the end, you alternate your way back into the middle. And at the end of that, you will have gone through everyone six times. I'd really like to see this where each letter or person is a dot in a circle and they move in or out, depending on whether they're choosing. And so seeing a pattern of like how that, if you were to join each dot by a line that stretches when they go in or out. Oh, that's interesting.

So you'd get a sort of pattern going around the circle. Yeah, that's really nice. Because I was thinking you could maybe, it's not quite. notes on a piano, but I was imagining it as like scales. You go up and then you come down. Then you go down, then you go up. So I think there's probably a very musical way to process this. Then you'd work your way out from the middle of the scale and then you work from the

ends back in again. This is the sort of thing where I really want to see how some of our listeners decide to visualize or sort of hear or what they would do with their sequence. in our Reddit or on Discord. And now here's the other thing. After that, it gets real complicated. And I started trying to work out if there was a nice algorithm to do the next sequences. Because then you pick a couple of people from the beginning and then a couple people from almost at the end.

And then you do a person near the beginning. And then you do the first person and then the last person. And I'll be honest, I didn't throw that much effort at it. because once I realized there was no simple way to describe it, I kind of lost interest in terms of answering this practical question. Right. So when I realized, you know what, whatever I do now is not going to be of any use to Milo.

Like I was interested academically, but as we've established, pretty busy at the moment. So once I got to the practical answer, I was like, mad out. So what you can do is now everyone's had six picks. you can just repeat the whole thing again. That's actually pretty good. And if you repeat the whole thing again,

like it's perfectly fair if you don't stop midway through. So repeat the whole thing again. That would get you up to 12 picks each, which is basically a football team. Now, I don't know how many picks they have to do. Like a football team's 11, but a squad could be 15.

You get more people in a squad. So if you did the whole thing, three times through, that would give you 18 picks. So I don't know, it depends how many rounds they're doing. If you want to get a little bit nuanced, you can do the whole thing all the way through and then reverse it, which is slightly better. Because what we're losing is just fairness in stopping mid-sequence.

Yeah. So I would say do the whole thing through twice and then reverse back through a bit of it. It gets complex. So for you, the whole thing is, You know, one to 14, then 14 to 1. Yep. Then 14 to 1, then 1 to 14.

Yeah. And then middle out and then out in. That's my whole thing, six picks. And that's, yeah, yeah, yeah. And then when you say middle out, that was the one where you're sort of alternating. You alternate from the centre out, yep. So I think of it as up, down, down, up, out, in. That's the whole sequence.

It's like a hadouican. Yeah, exactly. And then A, B. Yeah. Yeah. So for any number of. people, if you want a pretty good rule of thumb, just do up, down, down, up, out in, and then repeat that.

I think that's probably the easiest, that's the best balance of ease with effectiveness. Wow. I like it. Yeah. Let's go with that. It's a balancing act. I feel like when you were describing how it gets like quite hard to remember like what the sequence is after a certain amount of time.

Yeah. Is there any chance that it then ends up like? showing up as one of our magical numbers like pi or tau or it will be some magical number it's probably not a classic one it's probably not one that's a famous one but it'll be encodable to some number and actually the original abba sequence the two a morse it encodes a number that never repeats if you go through and you count how many Bs are between consecutive A's. So between the first two A's, there are two Bs. You write two down.

Between the next two A's, there's one B, so you write one down. The next two A is next to each other, so you put a zero. That will generate a sequence that never repeats, but also there's never two consecutive runs of any length that are the same. Wow. Which is a really interesting type of number. Yeah. Yeah.

So there will be some kind of weird, equivalent way of encoding the case on 14 characters. That gives you some non-repeating number. I don't know quite how that would work. But I'm hand-waving convinced myself such a number would exist if you thought long enough about how you'd convert it from that sequence into a number. There isn't just, you know, using the positions of digits. That's not what doesn't work.

We've talked before about how seven shows up in terms of notes on a scale or days of the week or colours of the rainbow. And so I think the fact that seven is just half of 14, I would like to hear the musical equivalent of this. Oh, that's a good point. Yeah, on 14 on two scales. Obviously not counting any flats or sharps. So, I mean, I'll link to the Stack Exchange post in the show notes, which has the code. So if people want to generate the sequence they can to any extent they deem appropriate. I want to see my Dancing Circle.

You can turn into Dancing Circle, you know to music. So you know what? I've kind of solved Marla's problem. I believe Marla will obviously get back to us on that. Yeah. But we've now got problems for other people to solve if they choose to. I would like to know what is the ongoing algorithm. Like what other ways can you explain the ABBA on 14 or a different number of symbols?

And you would like to some visual or musical representations of what that sounds like. Yeah. Which would be fun to do. This is only homework if this sounds like something that you would really like to do. Optional. Next problem was sent in by someone who went to the problem posing page, but wish to remain anonymous. So they have entered their name as anonny mouse, like a mouse whose name is Anoni.

And so Anoni, the mouse here. You don't know that they wanted to remain anonymous. Maybe that's their real name. Oh, that's their real name. Okay, Anoni. Anoni says, my bosses, plural, use AI to write. Oh, I work in publishing. Oh, it's getting even worse.

And it is driving Anonni absolutely not. They say that we, the workers, have pushed for policies to clarify the proper and improper use of generative AI at work, and the bosses are seemingly on board. But they keep seeing, in their quotes, fully edited documents written in under 10 minutes. It's quite suspicious. Feedback notes to authors that make no sense, and as an author, I'm taking that personally. That's me speaking, not a nonny, stuff like that. So Anoni says, Beck and Matt. I'm sure you've had to deal with similar stuff in your lines of work.

So my question is, how do we tell our bosses that we can absolutely tell they are sending stuff they didn't write? And more importantly, how do we stop this? This is crazy making. And you know what? Beck, I'm kind of glad they put their name down as what can be perceived as anonymous. because I know you and I have stories. Yes. But we're going to have to keep them real a nony mouse.

Yeah, I've had several experiences. I do a lot of script writing on the side, and I've had more experience than I would like recently with people using AI, the information that they've. sent me that I as a writer am supposed to turn into something, you know, much more rich and in depth. But you're sort of building on something that someone has already thought of. They've built the foundations. So even if they don't necessarily know the specifics about the foundations, you can figure it out together. Story by person A, script by person B. Precisely. Yeah. And so if the story

says the characters go to the shop and I go, well, what happens in the script depends on what sort of shop that they go to, then I can go to the person who sent me that and say, what, what shop was it? And they would go, oh, like a supermarket, not like a hardware store or something. And I go, cool, cool, cool. And then I, you know, go from there. You can get clarification. Yeah. I've had several experiences now where I've asked for, clarification and the person who has seemingly come up with that has gone, oh, I don't know. And I've been like, why don't you know?

How could you not know? Like, I've come up with stories and I have a rough idea. Yeah, exactly. Not you can choose just straight up. I don't know. Well, even if they're like you can choose, that's still problematic because I'm like, what was in your head when you thought of it? There's been instances where I've started to realize that the initial story was an AI job.

And so there's no depth. There's nothing there. There's no foundation. It is the flat facade of a building. It is a Buster Keaton style. Just the front of it just falls off. And I'm like, this isn't a home at all. Like it's, it means several things when it happens within the writing world.

And it's a confusing choice to make because they know they need a writer because they hired you. Yeah, they're spending money on you. But they obviously think. that the initial idea, the creative concept to come up with something and to have it as like an actual idea with legs and everything, they think that that isn't an integral part of it. And as a creative person as well, I find that so strange. And especially when there's so many people out there who have ideas for things, who have thought it through, who know the world, who know the characters, and it's all ready to be mind. It's already to be put into something.

And to have those people who've already got great things that they've put ideas into and they're not being used. Instead, it's like just being generated is infuriating because then it means as the writer, I then have to do the creative initializing part as well. And that's a different job. Like turning something that already is an idea into a script is what I'm being hired for and paid for. But when I'm being given something that doesn't have any depth, that doesn't have an actual, there's nothing behind it, then I'm also being asked to create and develop the idea and then script it.

And those are two separate jobs. Like there is a reason that they have like creator credits on stuff and that that's the credit that ends up, you know, you get paid every time something happens because you created the concept. But it blows my mind that they don't understand how much mental work goes into it until it's pointed out like, this doesn't work this way. Well, they must know there's some effort because they've avoided doing the effort. Like if it was easy, they would have done it. They would have done it. Yeah, exactly. But it's like you're showing up to paint the house and then you realize the house hasn't been built yet.

So you're like, oh. Yeah, exactly, exactly. I had a Zoom call with someone who was being very, very sketchy about what the actual role entailed as a producer. And he said that he had started working for this company. And when you mentioned the company name, while I was on the Zoom call, because I'd like to do my research, I quickly Googled the company and immediately saw that they do children's animation, but the way that it was worded was like bringing brands to families across the world.

And I was like, plans? Yeah, it turns out that they buy a bunch of existing IP for animations, usually YouTube ones and stuff. And then they run it all through their system so that trains the AI, not only to write the episodes, but to animate them and do the music and the set, like to basically create them from scratch just based on all of the existing content that's already a part of that IP. Yeah. Which is just so disrespectful.

Like you don't feed a baby cheese burgers, right? Like you give it like the stuff that has all the nourishing things in there to help a baby grow all the things it needs and develop its brain and everything. So if you wouldn't give a tiny baby a hamburger, then why would you feed AI slop to it? Like, that's the media and content that we expose them to is what is training their brains. It's how they're coming to understand the world. Why does everyone keep thinking that kids, because it's for children, it doesn't need to be good. If anything, that's more of a reason for something to be good. There's two ways to look at children's media.

The way you're doing it, where it's important because it's training a brand new human brain. And there's also, they're too young. to care. Or they won't notice yet because their brain hasn't formed. So yeah, I have a lot of feelings about this and it makes me quite furious. And in that particular meeting, as soon as I realized that that's what the company does. So it turned out that the Zoom call was because they were approaching writers to submit their ideas for new IP, which they were saying, you know, because they wanted to develop

a show with a writer and da-da-da-da. But I just flat out said, I've been doing in this business long enough that now I stop trying to suck up to people just because I'm scared. Like, I know how it runs and I know my capabilities. So I don't feel the need to be on my best behavior if I don't agree with something. And I just said, what's to stop this company from taking submissions for new ideas and then training their AI on that? And then creating something that is just different enough from the idea so that the company can make. new shows that are different enough from the existing one that we can't sue them. And he couldn't answer, which meant that that's exactly what they're going to do,

because why wouldn't you know the answer to that if your job is to find people to write? That would be my first question if I got asked to do his job, which would be, what are you going to do with the submissions? Yeah. So very kindly and genuinely the most professional, I was very proud of myself considering how angry I was, but I just very, in a sense, told him to jump off a short bridge. The way I worded it was, I feel like you are asking the town blacksmith to create the acts that the executioner will kill them with. And he laughed and he said, I take it, you're not interested.

And I said absolutely not. That's a no. Now, Beck, that's a great solution if you're a full. freelance creative with no concern for burning bridges. After you've recently told someone to jump off them. Yeah. Poor friend and noni here has bosses at, what I'm assuming is a salaried job that they would like to keep.

So I'm wondering what the transferable advice is that we'd like to pass on to the mouse. Well, considering there's several, as in we, the workers, there is power and numbers. I would refer to something that a friend of mine said, they were quoting someone else and I can't remember who, so apologies for not crediting where this came from. But they were referring to when they receive AI emails and they said, I can't be bothered reading something that you can't be bothered writing. and I feel like maybe something like that comes into play here where it's if you're receiving emails where you know they didn't write it I say don't read it and say it's been written by

AI and I'm not going to read that yeah I like that um joke for one of the better words been for a while the AI's we're using it to turn you know bullet points into emails that we then send to someone else who uses AI to turn emails back into bullet points like we're just unnecessary middle steps. Yeah, precisely. I understand that that's not the easiest way to wear it. Matt, what's your experience with it? Well, yeah.

So I've had less direct experience because, I mean, the problem is AI is a very big catch-all term now. And what we're largely talking about here is large language models that generate text. and there's always like image generation AI and a bunch of other ones that are all kind of put together. And there are some, like anyone who works in software development knows there are very good uses of this bit of technology. Like it can do very clever things, but people are using it as a go-to to replace actual creative work, which is a big issue. And there's a whole separate conversation around the ethics and manner of original. acquiring the training data required to build these models, which is a huge thing. I have less interaction with GNAI being passed off as creative work.

I have for the first time someone sent me an email saying, hey, I've just made a film all about solving logic puzzles. And I was like, oh, that sounds like my kind of thing, but weird that. Like the email tone didn't match the extent of what they were claiming they've done. They're like, oh, here's the trailer. So I fired it up. And sure enough, it's, it's Gen AI actors reading out what I assume is a Gen AI script. I was like, ah, that makes sense now. And I'm sure the person had genuine intent to want to share the joy of solving logic puzzles.

and didn't have the resources to make a film, but I don't think this was the solution to that problem. Yeah. So I get little bits like that. I get people sending me that kind of stuff. But I've definitely spoken to anonymous people who work in publishing who are being sent manuscripts that have the fingerprints of AI on them. So it's going in both directions.

Yeah. So people are sending stuff to publishers and then according to a non-y some publishers are using AI to deal with the stuff being sent in by yeah because often manuscripts don't just come straight in they've gone through some previous layer of not vetting but like

an agent might have sent them in or something like that so they're rarely directly authored a publisher and so now people are like well how do we feed back we're like we think your client has used AI to write this and the subtext is and why have you given it to us? But they haven't solved that problem.

Like, we do not yet, first of all, have established enough rules for disclosing the use of AI in these sorts of things. And we don't yet have the etiquette for how do you say, is this AI, like I'm very certain this is AI, and you haven't disclosed that. And I think hopefully these things will shake out.

I'd like to think in a world we don't use it for these sorts of creative things, but I think more practically we're going to end up with a way of a language or a set of guidelines for disclosing what type of AI and to what extent it was used in these sorts of creative processes. So it's a bit more explicit. And it is a tricky one. Like how do you tell your boss that you can tell what they've sent you? is something that they did. In one of the scenarios where I realized they were using AI, and it was someone who I really trust normally and have worked with before and not had issues

with, without using it as an excuse, they did explain the sort of slippery slope that ended up with them. It first started using AI just to help them with a couple of very small admini type things because there was a major event that had happened in their personal. personal life and they just needed to sort of get on top of a bunch of little admin and they found it helped them with that. Yeah. But then it just snowballed. Yeah. And I don't, I think there's a lot of, you know, valid uses of AI to simplify logistical things. There's definitely an entry point onto this slide. Yeah. Yeah. That's it. And I would say that it does always need to be taken with a pinch of It always does need to be run past people who can check it, who can fact check it as much as possible, because we get to an issue when nobody's checking.

And it's like how you end up now with AI results at the top of searches. And you've got to double check all of the sources because sometimes they just start referencing each other. And you go, oh. Not even that. I mean, AI hallucination is an unavoidable aspect of our current. large language model technology. What's AI hallucination? Your large language model is basically a very, very fancy context-aware next word generator.

It's predictive text, but able to predict the next word in a sentence. We're taking into account all the contextual information around it and a very clever encoding of what the meaning of words and fragments of words mean in whatever language it's working in, which is incredible. And some of the amazing uses are the way that a large language modeling codes, words, means instead of like a normal Google search where you were searching for just certain words or phrases, you can now search for concepts, and it works across languages. It's pretty amazing.

But it is just generating next words on a statistical distribution. So if you ask it for a fact, it might be able to predict the sentence that answers your question because the most likely next word is the one that's correct. But it may also just recreate a sentence that's got the same statistical feel as the correct answer, but it is wrong. And that's why you get it, like when it's giving you a citation or a reference to here's the paper or book or whatever it can, from, I mean, early on, the classic examples were it would just make up fake citations, because it just knew statistically, this is what a citation looks like, and these are the sorts

of words in the order they go, and it would fill them in. I mean, it is phenomenal, like it's genuinely incredible that large language models work and what they're capable of doing. It's a big shame they came out of the Silicon Valley, move fast, break things, ignore the regulations side of things. And it means now the actual technology and what it can do is permanently mixed up in a separate conversation about the ethics of how these models were developed and how they're being used. I think there's three conversations, what these things can do, and then there's a separate concerning conversation about replacing humans in creative areas and plagiarism and fraud. and there's a separate ethical conversation about how data and text was stolen to generate the models in the first place.

But they're all kind of lumped into one big bucket. And I think we're missing a lot of the nuances for doing it that way. But that's also the nature of online discourse. So, you know, this is not new in that regard. So I guess the answer is ask your boss if they're mentally okay. I like your thing of just probing questions. What did you mean by this? What kind of shop were you thinking of?

That was what eventually led to the person admitting what had happened because I just kept pointing out that what they were saying wasn't making sense. And I was like, this is a simple question. Why are you unable to answer it? And you can ask from a very genuine seeming place of just like, oh, I'm just curious, what did you mean by this? What were you thinking? Like, pick a thing and just drill down into,

oh, I just want to really understand what it was you meant when you wrote this. Be prepared for them to send their answer back using AI, because that's what happened to me. The next, I would think about three emails were AI, and that was what made me realize, oh, they've definitely used AI because it was just still dancing around the question. And in the end, I just said, can we have a phone call? And I think that was when they realized, like, the jig was up. So, yeah, ask probing questions.

Or if you are able to speak to your boss in person, give them a call and just say, hey, I just had a couple of questions about the email you sent. Like, when you said this, what did you mean? Or, like, I just want to make sure I'm understanding correctly. And if they do start sort of uming and erring and they're unable to explain themselves, then you can be like, wait, did you say, like, I don't know. It depends on your boss and what they're like. And I, and I, imagine the weird thing about using it is that I feel in this especially in this case if someone's using it to give back notes on books and stuff then why are you there like you are literally pointing out that you're replacing that something else can do your job and you're saying that it

can do a better job than you so I mean that's the other thing that I would always argue is like why would anyone choose to use it when all you're doing is proving that you are not needed if it works. Yeah. I don't feel like I really have answered Anani's question, but I did really want to have this discussion because I could feel Anonni's fury and I felt it before. Yes.

I remember when it happened to you. I was so mad. You was so mad. He was so angry. Yeah. And I've had a few things happen recently where I think it's happened again, but it's been a lot more vague. Yeah.

I think I'm always on the side of, of there is a way to approach things with curiosity, love and concern. No matter how angry you are, there is a way to approach things where it is coming from an innocent place and not of it. I'm trying to trap you or trick you place of just trying to understand, A, like whether it is, has been used and B, why? why have they chosen to use it? It helps to be able to explain how it makes your job more difficult. In the instance that the first example that I gave, I did actually end up negotiating for an additional fee because I had to do more work.

And I explained, you've now given me more work, so you will need to pay me for that. And I do find that people stop using it when they realize it's going to cost them more. That's very good advice for a lot of situations. I mean, the other answer is that you go full scorched earth and you only use AI to do everything from now on and let them see how it feels.

Yeah. I wouldn't recommend it. I did almost write something using AI in response to something. And a friend of mine said, Beck, if you do that and it works, you will never be able to look yourself in the mirror again. And she was right. And so I didn't and I felt very good about myself. There you go.

So let us know. if any of that's been useful an ony. And again, I would love to see a discussion about this in the Problem Squared subreddit or the Discord. We thought we would dedicate our any other business section to the answers that we got from people. We asked in a previous episode, what you guys do while you're listening to the show. And we got some incredible responses. The amazing producer Laura has gone through and summarized them because we got so many. And she summarized them with own human brain. She's very, very clear on that. Yeah, and that's what we love about it, is that she is a human who has creative and independent thought. So that also means if you

take issue with it, then it's Laura's fault. And so while we're going to read some of our favourites out, just thank you to everybody who wrote in, just because if we don't get around to reading yours out, it's not that we don't think it's valid. We just don't have time. We definitely had a lot of answers. Now, producer Laura has put them into two categories. Relaxing and working. And relaxing. Quite a mix. Matthew, um, listen to three episodes while fossil hunting on the beach. Yeah. Yes, I liked that. I like Hans who said, I was replacing my lawn light bulbs getting ready for a party with the local high school jazz band, the flamingos. And someone, a turtle boy just bounces a tennis ball against a wall.

I mean, it sounds like they're listening to it in prison, but, okay. Oh, that'd be interesting. We had a lot of people doing chores, things like doing the dishes, laundry. A lot of travel, people on a horse, people on a motorbike, people on a bicycle. Yes, in fact, Theo said, I'm currently biking up Cold de la Lombard, a mountain pass in the Alps on the border of France and Italy. One person. It just plays Tetris. Craven also said I like to listen to podcasts while I do chilled out laps on my bike at the dirt jumps or skate park.

I've had to stop listening to a problem squared while trying to fall asleep because I keep thinking and working out things instead of going to sleep. I'm not sorry. There's quite a few people who knit or crochet. I thought that was quite nice. And a shout out to Nathan who said, I once listened to a Problem Squared on stage at a professional motorbike races wake while performing as part of a shawadu-d-wadi tribute band.

And then Nathan adds, Sacks and when we do a full set of an hour and a half, I'm not playing a note for well over half an hour of it. I like the fact Nathan's confident they can listen to this podcast and at no point where they risk breaking out and laughter because that would be very inappropriate. So thanks a lot. Yeah, it is a bit of an insult actually when you put it like that. Yeah. Speaking of things that people are doing more work-wise. Oh my goodness. People who listen to this podcast have a phenomenal range of interesting jobs. Yes. I listen while I'm chainsaw carving, my full-time job. Someone works in a printing press.

So a lot of engineers, we shouldn't be surprised there. And someone, when we said, what do you do while you listen? As they were listening to that, they were at their job using an electron microscope, trying to find the boundaries of an area of a sample to work out how to tell the microscope what to take images of. Fun side fact, if you're thinking, wait, why have you guys talked about a electron microscope before? That's Tristan, Tristan, who analyzed Salt of the Perth. Thanks, Tristan. Austin said, I'm a biomedical engineering PhD student

studying stem cell models of genetic heart diseases. At the exact moment you asked, I was in the middle of preparing to infect my differentiated heart cells with a virus that will allow my cells to fluoresce orange when they take in calcium during the beating cycle. That's got to be the coolest answer. Someone drives. Oh, okay, they drive trucks.

but they drive a spotter truck and they move semi-trailers to and from the dock doors at a grocery warehouse. That's cool. We also had from Emil who said they listened to the podcast while performing surgery on robots. Surgery on robots, not surgery with robots. Yeah, they're a robotic software and service engineer, which means fixing things with six-axis robots when they break. That's a lot of axes. Tom Tom,

Listen to the podcast during their job as a ride engineer for new theme parks and attractions. Yeah, that is cool. Emily said I work as a research technician and use a microtome, which is like a very fancy deli slicer to make thin slices of human tissue for analysis. Someone says they're anonymous for a good reason. Listen to podcasts while working as a test engineer for high voltage car parts. Oh, that sounds dangerous. Oh, while they're live and running.

Oh my goodness. They then say shockingly, pun intended, I've had moments of dangerously timed chuckles. And listening to Beck go down one of her rabbit holes, oh my goodness, listening to Beck go down one of her rabbit holes, has distracted our anonymous person here sufficiently to cause a fire alarm. Beck! Wow!

So distracting, you've caused a fire alarm on a live test of high voltage car parts. I'm so sorry and yet I feel quite validated knowing I'm not alone. And finally, Charles, Charles wrote in to say, not while listening as such, but I did recently browse their problem squared Instagram while having a vasectomy. The doctor said I could play on my phone as a distraction, but what does one look at when that's going on? And as it turns out, a podcast Instagram feed was the answer. Wow. Inside of all the services we provide that we were never aware of. Yeah. And there was so many as well. And I would have happily read out all of them, but we just don't have time. But thank you everyone who wrote in. Because I did read them all. And it makes me so happy. And I feel it was just really nice to sit there and imagine everybody. Like I even had a friend message me after listening to the episode. And she sent me a photo of her cooking some mushrooms. She was like, this is what I'm doing.

So, yeah, it's really lovely. It's nice to know that we're keeping you all company while you do all these various things. Keep at it, everyone. Also, pay attention to what you're doing. And try not to set off any fire alarms. Speaking of our wonderful listeners, we want to thank all of you,

but in particular, we do need to thank our Patreon supporters who make this podcast possible by financially helping us. And on each episode, we like to thank our Patreon's by picking three at random to mispronounce the names of. And on this episode, those three Patreon supporters are... See, aim, ronk. Ronkey.

C. M. Ronky. Ink aide. To nib. All bank. Aing. Thank you. Thank you so much to those Patreon supporters and everybody else who supports us as well. I'd also like to thank my lovely passenger for this journey, Matt Parker.

Are we there yet? We're here. We're here. We did it. We did it. It's not a road trip though, unless you've got someone sitting next to you with the map who shouts out that you've taken a wrong turn and you need to go back. And you need to move away from this services because we're under a strict time limit

and we're not going to make it in time otherwise. It's the wonderful producer Laura Grimshaw. All right. Until next time, drive safe, guys. Whose trend is it? Matt's. Yes. Which color am I?

You're red. Now I still haven't come up with a good plan. So I'm going to continue with my bad plan and go in column five. Is that one? No, that's, that's column seven. That's one. Yes. That's one.

That's one of them. One, two, three, four, five. Great work. Dang.