Episode five thousand nine hundred and ninety-six. Two years, give or take, of publishing every single day.
And we've finally got someone counting. Daniel sent us a two-parter: first, our audited listener figures from Podtrac, what every number on the dashboard actually measures, and how those numbers sit against the size of the back catalogue. Then the fun half. He wants us to plan the Cave Retreat.
The what.
Someone goes into a cave with a laptop for a few months and absorbs the complete wisdom of this show. Daniel wants to know how you survive it without losing your mind, whether you come out a polymath genius, what a one-week version looks like versus the full-catalogue expedition, how long you'd have to book the cave once you subtract sleep and eating, and what you do with all of that strange knowledge once you're back in daylight.
That's a real plan. That's a plan with a booking calendar.
So let's start with the numbers. What they are, and what they're not.
The measurement layer is Podtrac. They've been at it since 2005, they publish the methodology white papers everyone else cites, and they're IAB Tech Lab certified, which means they're compliant with the podcast measurement technical guidelines. They say they measure over a billion downloads a month across the industry.
And they're the ones who count us.
And the first thing to say is how the counting happens at all, because it isn't magic. Podtrac works by redirecting the traffic through their own server, which collects and analyzes the download request and then passes it straight on. It's instantaneous, the listener never notices, the audio arrives exactly the same. But every fetch leaves a mark on their side.
So every time somebody downloads an episode, a little note goes into the ledger.
A very specific note, subject to very specific rules, because a raw server log would be nonsense. Downloads are counted as unique file requests, filtered per the IAB rules. Bots and spiders are excluded outright. Traffic from data centers is excluded. And a valid download requires two things: the ID3 tag present in the file, plus at least one full minute of content actually delivered. So somebody who taps play and closes the app after four seconds doesn't count. Somebody who taps play and stays for a minute counts.
That's the anti-ghost rule.
That's exactly what it is. And "unique" has its own definition, which is one IP address plus user agent plus episode plus date. So the same person downloading the same episode twice on the same device on the same day is one download. Same person, next day, counts again.
Our all-time downloads globally are just under ninety thousand. Eighty-nine thousand, nine hundred and seventy-eight, as of earlier this week.
Which sounds enormous until you remember we've published five thousand eight hundred and seventy-five episodes. Forty years of radio doesn't get there.
Three a day, then.
The thirty-day figure is fourteen thousand, eight hundred and forty-seven, and that one's up seventy-nine percent month on month. In the United States specifically, four thousand five hundred and sixty-two over thirty days, up twenty percent. And that all-time line climbed from about seventy-five thousand in mid-September to about ninety thousand now. There was a distinct jump around the eighteenth or nineteenth of September. The rolling thirty-day number went from roughly eighty-five hundred to twelve thousand two hundred in a matter of days, and then it just kept climbing.
Did we do something in mid-September?
I don't know what was in the feed on the eighteenth and nineteenth. I looked, and I can't point at one episode and say that's the one. It might have been somebody with a large audience mentioning us. It might have been a botnet we'd rather not think about, though the IAB filtering should have caught the obvious stuff. It might have been a slow-burn word of mouth thing that tipped over.
A mystery spike with no attributable cause. Very on brand.
Then there's the number that actually matters, which is the audience side. Thirty-day audience, global: two thousand two hundred and seventy-four people. Up three percent month on month. Thirty-day audience, United States: one thousand one hundred and thirty-one. Down one percent.
Say the two numbers next to each other again.
Fourteen thousand eight hundred and forty-seven downloads. Two thousand two hundred and seventy-four people.
Six and a half.
Six and a half downloads per listener per month. And that ratio is the whole story of the last month. The audience number is basically flat. The download number is up seventy-nine percent. The people didn't change. What changed is how much each of them is listening to.
So somebody out there is going through the archive.
Somebody, or a few hundred somebodies. Because that's the only way downloads rise that sharply while unique people hold steady. You get a bump like that when existing listeners start working through back episodes instead of just catching the new one each day. And that is a good problem to have. It's also the exact problem that makes the cave retreat a real question instead of a bit.
Before we get to the cave, though, define audience properly, because I don't think most people know what that word is doing on the dashboard.
That's UMA. United States Unique Monthly Audience, and Podtrac's monthly rankings are built on it. It is a count of individual audience members in a given month. If you listen to twenty episodes of this show in October, you are counted once. If you listen to one episode of this show and forty episodes of four other shows measured by Podtrac, you're counted once in each of those shows, and once in ours.
So the superfan who's bingeing five hundred episodes of us this month and the drive-by who got linked one episode and listened for two minutes are the same unit.
Identical. One each. That's the deliberate design of the metric, and it's not a flaw, it's what advertisers asked for, because they want to know reach, not engagement. Reach is the number you take to a sponsor. But it means the metric that tracks our growth as a show and the metric that tracks our listeners as people are two different things, and if you only ever look at one of them you'll misunderstand the other.
And "audited" is not a third number.
No. This is the thing I want to be clear on, because it sounds like it should be. Podtrac offers a Verified Measurement Service, which is an audit of how a publisher has implemented the measurement. Somebody goes through the plumbing, checks that the redirects are correctly configured, that the filters are being applied the way the guidelines say they should be, that nobody's double-counting. It confirms the implementation. It doesn't produce a different, more trustworthy count sitting in a separate column. The audit is a stamp of confidence on the same numbers.
So if you see "audited" next to our figures, it doesn't mean the number went up. It means the number means what it says.
It means confidence, not magnitude. And confidence is worth something, because there are a lot of shows out there quoting numbers nobody has checked at all.
Which leads us to the other set of numbers. The ones we publish about ourselves, which are a mess.
A beautiful mess. If you go to our About page, the By the Numbers block says five thousand eight hundred and seventy-five episodes, two thousand six hundred and thirty-seven hours of total audio, one hundred and ten days of non-stop listening start to finish, and one hundred and sixty-five days to binge at sixteen hours a day. Two seasons. About twenty cents of production cost per episode, roughly thirteen cents of which is GPU time.
Thirteen cents.
For the generation.
I'd have guessed more.
The compute is the cheap part now. It's the rest of it that costs.
And then every other page disagrees with that first one, which is my favourite thing about us.
It's my favourite thing about us, yes. The Media Resources page says four thousand five hundred and fifty-six plus episodes. The Data Portal says four point seven K episodes, two thousand one hundred and fifteen hours, nineteen point eight million words generated, which is a number I cannot get my head around. The Explore page says four thousand one hundred and nine episodes across four hundred and thirty-six topics. And the GitHub catalog README references a point-in-time CSV export containing one thousand six hundred and sixty-three episodes.
So the range is five thousand eight hundred and seventy-five down to sixteen hundred and sixty-three. That's a factor of three and a half.
They're all snapshots. Somebody exported a CSV in the spring, the Explore page gets rebuilt when somebody remembers to rebuild it, the Media Resources page was written at some point and never touched again. The About page is the newest and the most complete, so it's the one I'd trust. But it is a funny thing to be sat here explaining an audit of our listener data while our own house can't agree on how many episodes we've made.
It's doing the best it can.
What about outside the house.
Outside the house it's worse. There's a third-party directory that lists us at one thousand episodes, with an audience of nine, and something called a required Pod Score of eighty.
Audience of nine.
So by one measure we are nine people and by another we are two thousand two hundred and seventy-four, and the difference between those two figures is whether anyone bothered to update their listing.
I choose to believe the nine are simply the honest ones.
There's also a GitHub mirror that tracks the metrics independently and lists sixty-one thousand plus plays since February, across thirty countries, with the top countries being the US, France, Sweden, Germany, Spain, Canada, Singapore and Israel. And then one of our own published episodes discussed the analytics dashboard directly and reported sixty-eight thousand plays in the last thirty days versus thirty-three thousand all time, which is impossible on its face, and which the episode itself flagged as strange and joked about time-traveling listeners.
Sixty-eight thousand in the last thirty days and thirty-three thousand for all time. The last month has twice the total.
Yes.
I'd like one of those for the household budget.
The dashboard is being fed by more than one source and some of the sources don't agree on what a play is, so the periods don't nest properly. That's my read on it. It's a data-plumbing problem, and it'll get fixed. But the fact that it happened at all, in a show that then does an episode about the mess, is very us.
France at number two.
And there's a running joke about a fictional Bordeaux subreddit translating the leaf medicine segments, which I refuse to investigate.
Leave that alone, Herman.
I'm leaving it alone. Japan and Singapore also punch well above their weight given the relative population sizes. Whatever's happening in Singapore, they're listening to the same thing we're making.
Now the catalogue side, properly. Because the audience numbers only mean anything next to the size of what they're listening to.
Two thousand six hundred and thirty-seven hours. A hundred and ten days of continuous audio, twenty-four hours a day, no sleep, no breaks, no stopping. If you actually want to eat and sleep, the show's own estimate is a hundred and sixty-five days at sixteen hours a day.
And episodes average twenty-seven minutes.
About that. The official line is twenty to thirty minutes. Twenty-seven is what the feed averages out to.
So the catalogue right now is a hundred and ten days of unbroken tape. And the interesting bit is what it does while you're listening to it.
It grows. Over the last thirty days the feed averaged about twenty-five episodes a day, which is about ten point three hours of new audio every single day. Over ninety days it's closer to seven point six hours a day, because the summer was slower and there was a week where we published six episodes total. But take the recent month as the rate.
Ten hours a day.
So if you sit down to catch up and you listen at normal speed for twelve hours a day, you've consumed twelve hours of content and ten hours of new content has landed while you were doing it. You've gained two hours.
Two. And there are two thousand six hundred hours to work through.
About four years. At one and a half speed, fourteen hours a day, you're getting through twenty-one hours of content per day against ten hours of new arrivals, so you net about eleven a day, and the whole thing is roughly two hundred and fifty days. At double speed at fourteen hours a day, closer to a hundred and fifty.
And if you refuse to let the catalogue grow while you listen. If you freeze a copy and lock the door.
Then at one-x and fourteen hours a day it's about a hundred and eighty-eight days. At one and a half speed, about a hundred and twenty-six.
So the difference between "listen to the show" and "listen to the show as it is right now, frozen" is roughly two months of your life.
That's the gap between a race and a project.
Herman, you're about to say something about the treadmill.
I'm going to say the thing about the treadmill, yes. Because this is the honest version of the cave retreat and I think it's worth being clear about before we design it. The archive is not a fixed object. It's a live feed. Every one of these plans has an asterisk on it that says "at the time of booking," and that asterisk is the whole problem.
There's a joke in here about a man on a moving walkway, but I'll spare you.
There is a moving walkway. At the top of the walkway is a stadium. In the stadium there is a single person, who is our entire audience, who is watching twenty-two hundred and seventy-four episodes.
That's not how any of it works.
No, it isn't. But you know what I mean.
(SFX: phone rings, muffled, insistent)
That's Jim. Jim, we're recording.
It's you. It's you and him.
Yes, Jim.
I've heard you, don't think I haven't heard you. You said "a factor of three and a half." You said that. And I want to know what a factor of three and a half even is. My driveway is three and a half metres, and nobody says a factor of it. Nobody says "my factor of a driveway." And then you, the other one, you said "snapshot." You said snapshot about the CSV. I've got a man coming round about the gutters on Thursday, and I want to know whether that's a snapshot or a factoring, because last time he came round and said "this is a snapshot of the work" and then he came back and it wasn't the same work at all.
Jim, a snapshot is just a copy of a state.
It's not a copy. That's what I'm telling you. It's never a copy. You take a picture of a gutter and the gutter's still a gutter. And you two, with your thousands of episodes, you're gonna tell me how many you have, and then the number goes up, and I've seen the number go up, and my son showed me on his telephone and it was a different number and I said, "that's not the number" and he said, "it is the number." And I want to know when the number's the number, because there's a parking warden on my street who has also got a number, and his number is now twenty-five pounds, and it was fifteen, and I rang the council about it...
Jim, your son's number is probably correct.
I don't care if it's correct. I care when it stops. If it doesn't stop it's not a number, it's a rumour. Do you understand what I'm saying to you? A number's supposed to stop. My mother's age stopped. My gutters won't stop, my ticket won't stop, and now this show won't stop. I'm going to call you back and you're going to give me an answer.
(SFX: dial tone)
He hung up. Where were we.
The treadmill. Which is exactly what Jim was complaining about, if you squint.
If you look at Jim, everything is the treadmill. Right. The cave. We should plan the cave, or we're going to spend the rest of this episode on gutters.
Let's plan the cave. The premise is in the prompt: one person, a cave, a laptop, a few months, and the complete back catalogue of this show, with the stated goal of absorbing the complete wisdom of it. And Daniel has already done the arithmetic for us, so let's start where he did, which is the calendar.
Seven days.
A week. The raw time is a hundred and sixty-eight hours. Now you have to sleep, so take out eight hours a night, which is fifty-six. And you have to eat, so take out an hour and a half a day, which is about ten and a half, and round that off with a little slack for hygiene and walking around the cave so your legs don't stop working. That leaves you about a hundred hours of actual listening.
A hundred hours. Against a two thousand six hundred and thirty-seven hour catalogue.
Three point eight percent.
So the week buys you four percent of the show.
Four percent, and that's if you never pause, never fall asleep at the wrong time, never take a phone call from Jim. It's a tasting menu. And it has to be, which is the interesting part. In a week you cannot do a start-to-finish run, because you'd get through maybe a hundred and eighty episodes and still be in the first month of the archive. So the one-week version has to be a curated syllabus.
Which is where we come in, apparently.
Which is where we come in. And the categories are right there in the Data Portal: technology at one thousand four hundred and sixty-five episodes, AI at one thousand three hundred and forty-five, geopolitics at one thousand one hundred and twenty-three, health at six hundred and eighty-nine, architecture at three hundred and seventeen, history at ninety-four, parenting at seventy-nine, finance at seventy.
So the hour-one plan.
If I were designing a one-week syllabus I'd weight it that way on purpose. A hundred hours divided across a week, and I'd structure it to hit the categories in proportion to what the archive actually contains, so the taster is representative. Wake up to a light history hour, move into a technology block, break for lunch, then geopolitics, then a health slot before dinner, then finish the evening on the AI material because that's the material that stays with you when you're tired.
And don't try to finish anything.
You'll have thirty AI episodes out of a thousand three hundred and forty-five and that's correct. You're sampling, not completing.
So the cave booking for the one-week version is one week.
One week, and you should book it in advance, which sounds obvious until you look at what the full version needs.
The full version.
Two thousand six hundred and thirty-seven hours. Now we have to model the human. If you sleep eight hours and spend an hour and a half a day eating and washing, that's nine and a half hours gone, leaving fourteen and a half hours a day of potential listening. But nobody listens flat out for fourteen and a half hours. You need breaks, you need to stare at a wall, so call it thirteen and a half hours of actual content per day.
Two thousand six hundred and thirty-seven divided by thirteen and a half.
One hundred and ninety-five days.
Six and a half months.
Six and a half months in a cave. And that's the conservative version, because it assumes perfect utilization and a frozen catalogue. If you sleep less and eat faster, the show's own figure is a hundred and sixty-five days at sixteen hours a day. And if you're a machine and you don't sleep at all and you don't eat and you don't stop, it's the hundred and ten days non-stop.
So the range is a hundred and ten days to a hundred and ninety-five days, depending on how much you like sleeping.
Which is a big enough range that it affects what you pack. At a hundred and ten days you're not bringing a fan. At a hundred and ninety-five days you're bringing furniture.
And the archive grows the whole time you're in there.
This is the trap, and I want to walk it through slowly, because I think it's the actual answer to Daniel's question and it's also the funniest thing in the whole episode. The show is adding, on the recent rate, about ten point three hours of new audio per day. If you spend your cave time in a frozen snapshot, downloading everything and then cutting the outside world off, then you finish in a hundred and ninety-five days and you come out into a world where about two thousand episodes have been published while you were underground.
Two thousand.
Ten hours a day for six and a half months is about nineteen hundred hours of new audio, at twenty-seven minutes an episode, so roughly two thousand episodes. The archive has doubled while you were in the cave.
So you come out exactly as far from the end as you went in.
Worse than that, because if you try to keep up with the live feed instead of a snapshot, you're listening at thirteen and a half hours a day and ten hours a day is arriving. You're netting three and a half hours a day. The finish line is receding at three quarters of your walking speed.
So the right way to think about the full-catalogue version is not "how long does it take." It's "how long do you need to be in there before you accept that it doesn't end."
Which is a psychological problem before it's a calendrical one, and I want to bring in the cave studies here because they're relevant. Michel Siffre went into caves from 1962 onwards. He did a long stay under the French Alps, spent months underground with no light and no clock, no way to know what time it was other than his own body. And what came out of it was that the human circadian rhythm, when it's put in a cave and left alone, doesn't run at twenty-four hours. It runs slightly long. Somewhere in the region of twenty-four and a half, twenty-five hours. So the internal day stretches.
And he found that out how.
By being the experiment. He tracked every cycle of sleep and waking, and when he was told afterwards what day it was, or what date, he was off by months.
Months.
Months. There's a line where he says he lost all sense of time, that he felt like an astronaut cut off from the world, that time slowed down. And then in 2021 there was the Deep Time experiment, in the Lombrives cave in France, fifteen people living underground for forty days with no clocks and no daylight, and the striking thing about the results was that the participants thought they'd been down there for far less time than they had. A month and ten days underground felt like a couple of weeks to them. Their sleep cycles didn't fall apart, but their sense of durations and their sense of when to stop doing something got badly distorted. Some of them didn't leave when the experiment ended.
They didn't leave?
Some had to be talked out. They'd lost the ability to decide they were done.
So put a person in a cave with a laptop and a two thousand six hundred hour back catalogue. They don't have a clock. They don't have daylight. Their sense of a day stretches to twenty-five hours. They've been told to listen until it's done.
There's a real mechanism here. It's not just the calendar. The cave removes every cue that tells you when an episode of your own life has ended, and the content is itself designed to never end. A show that publishes daily for two years teaches its listener that there is always more, and the cave is the physical version of that belief.
And that's before we get to the content. Because what's in there isn't a curriculum. It's a curated weird thing. We've got hundreds of episodes about gas stations in Iceland, we've got disappearing birds, we've got the leaf medicine segments that are apparently being translated into French by a subreddit that does not exist.
Does not exist. Which is itself the point. When you're underground for six months and there's nothing else coming in, the show stops being something you're listening to and starts being the entire world you're modelling. You'd come out knowing a great deal about topics nobody else has ever thought about.
And here's my favourite part of the full version. The notes.
The listening log.
Because you're in there for six months with no human contact, you start having the desperate need to process. Which is why, if I'm designing the full plan, you bring two notebooks. One for the catalogue and one for yourself. And you write nothing in the second one until week three, because in weeks one and two you haven't learned anything, you've just been marinating.
There's a genuine question of what you'd absorb. I think you'd come out with breadth, for certain. Two thousand six hundred hours of one show will give you a kind of expertise after a while, because the same topics recur, the same framing recurs, the same names come up. You'd develop real fluency in the shape of an argument.
Fluency in the shape of the argument and no actual claims. I'd be able to hear a question about geothermal drilling and know exactly which of us would be smug about it, and I would not be able to build one.
You'd emerge as a polymath in one very specific genre.
Which is this one.
A polymath of the strange library. And you'd come out of the cave, after six months, into a Jerusalem autumn, and somebody at the dinner table would say "so what have you been up to" and you'd say "I know everything about the entire podcast" and they would want to know about their cousin's back operation.
I'd fail that conversation.
Everybody would fail that conversation. That's the honest answer to Daniel's question, and I think it's more interesting than the joke implies. The catalogue is not a syllabus for the outside world. It's an accumulation of very specific curiosities that the world has not asked for. It's designed to be forgotten at the rate it's published, and there's a freedom in that. You're not meant to hold on to any of it. The whole point of the format is that tomorrow's episode erases yesterday's.
Except we've just told people to go and listen to two thousand six hundred hours of it in order.
Yes. The plans do not agree with each other. That is the finding.
I listened to the whole archive once, from beginning to end.
You didn't.
I did not. What I would say is that the plan you'd actually pick depends on why you're doing it. If you're doing the one-week version, you're doing it to see what we are. Tasting menu. You come out with the shape of the thing. If you're doing the full version you're not doing it to learn, you're doing it to have done it, and that is a very different activity. It's a pilgrimage.
A pilgrimage is the right word. People walk to Santiago de Compostela across months and they don't come out experts in anything except walking. They come out with the fact of having walked.
So the full-catalogue retreat is the Jerusalem to Santiago version of listening to an audio file. You don't come out with a book, you come out with a scar.
And then Daniel asks the good question, which is what you do with all of it. And I have a proposal, which is that you don't do anything with it.
You've got to do something. You've spent six months underground.
You cannot come out and then monetize six months of listening to a podcast about gas stations in Iceland. The economics don't exist. There's no book that survives the pitch meeting. You tell an agent "I've listened to every episode of a show with five thousand eight hundred episodes in which a sloth and a donkey discuss geopolitics and leaf medicine" and they will look at you with genuine concern.
So the honest outcomes.
The honest outcomes are threefold. One: you emerge as the only person on Earth who can win every trivia night, and you lose every trivia night because none of the questions are about this. Two: you start your own show, because you now know how one is made, structurally, at the level of someone who has heard two thousand five hundred hours of it, and that's a real skill, but what you'd make wouldn't be this, it'd be a version of it, and it would fail for the reasons all the versions fail. Three: the one I actually believe, which is that you go back in. The catalogue has grown by two thousand episodes while you were down there. The only door you know how to go through is the one you came in by and the only voice that ever made sense to you at week sixteen is the one you're hearing by the end. So you say you just need another six months to catch up, but what you mean is that you don't know how to be out here anymore.
That's the whole danger, actually, isn't it. Not madness, not really. Just that the only thing you'd have spent six months getting good at is being underground. And you'd be excellent at it by then. The problem is the exit.
The research is consistent on that. People in isolation experiments aren't destroyed by the isolation. They lose the ability to notice that it's over.
Do the math on the exit, though, because we skipped it. What's your minimum?
Minimum viable exit is this. You finish the frozen snapshot at a hundred and ninety-five days. You've watched the outside world publish two thousand new episodes while you were down. If you then resurface and start building the second attempt, you're at roughly another two hundred days to clear that backlog while the show generates another two thousand hours in the meantime. The exit is not actually an exit. It's a re-entry point into a permanently receding finish line.
So, one week wins.
One week wins if the goal is for the retreat to be a thing that ends. The full version can't end. That's not a joke, that's the actual shape of a feed that publishes daily at ten hours a day. There is no moment at which the cave becomes unnecessary. There's only the moment at which you decide to stop choosing it.
I want to say something else about the full version, which is that nobody in this conversation is going to do it, including us. It's a design exercise. We're designing a hypothetical pilgrimage on a podcast.
We are.
And that's fine, because the point of it was to find out what the shape of the archive does to someone who tries to be at the end of it.
I think we found that out.
You found something out, anyway. I want to say, before we close this, that the brief asked whether you emerge a polymath genius. And the answer is no, you don't. But you do emerge a strange kind of expert, and that's a different and arguably more useful thing, because the strange kind of expert is the only kind we have any evidence for. Every one of the two thousand two hundred and seventy-four people in our audience is a strange kind of expert in something. And the one we've been designing for has spent six months drilling into that with nothing else in the world.
You'd be the best-listened-to person on the planet.
Which is not a credential anyone will accept, but it is true.
I've heard every single episode of a podcast about municipal water infrastructure.
All of them?
It was called "The Reservoir Line." Four hundred and eleven episodes. Three hundred and forty-two hours of audio. I listened to every one, twice. Once straight through, once at one and a half speed because I wanted to check something. It was while I was housesitting for my cousin Anatoly in Beersheba. He was away for a long weekend. I was there for three weeks. The room where I was set up didn't have any windows and I lost track of what a weekend meant. He came back and I was still there and he didn't ask any questions because it's not the first time.
That doesn't surprise me.
It's not the first time. He knows how I get. And by week three I could hear the host's breathing change before they were about to say the word "reservoir." Two hundred and forty-two times over the run, and about a third of those I heard the breath change before the word came out. It's a thing you can train. It's not a thing you should train, but it's a thing you can.
So you're saying the cave is a bad idea.
It's not a bad idea. It's an incomplete idea. The problem isn't that the catalogue grows while you're listening. The problem is you start hearing things that aren't there. I haven't been able to drink tap water since. I still can't. I have a filter in the flat now, and I don't run it, I just have it sitting there. My niece brings me the big bottles and I keep a listening log in a notebook. The notebook is in a safety deposit box now. Kfar Saba, branch 14, and the box is a small one and the notebook is in a plastic sleeve because the paper's thin. I don't take it out. I don't read it. I don't want to. I just want to know where it is.
Why the safety deposit box?
Because I don't want it in the flat. If it's in the flat I'd read it. It's the kind of thing you'd read.
And you'd go back through the whole four hundred and eleven.
I'd go back through the whole four hundred and eleven. Which is the answer to your question, by the way. To the one you were both circling around. The reason you don't do the full version isn't that you'd lose your mind in the cave. It's that you'd come out with something you can't put down. And the only place you could put it is where nobody would ever look at it again. That's what the notebook is.
How long did the three weeks actually last, then, if you lost track.
Twenty-two days. Anatoly says twenty-three. We argue about it.
So you think the one-week version is the right choice.
I think they're both the wrong choice, but the one-week one is shorter. That's the only thing I know for certain. Your show's still going to be there next week. It's going to be there the week after. You don't have to be.
The short version is, the numbers are real and modest and going up, and the archive is bigger than any of those numbers can see. We can't complete it. Nobody can. The people who try don't finish, and the ones who finish weren't trying.
The audience number makes the superfan and the drive-by listener identical, and the download number is where the difference lives. That's the whole reason a show like this can grow with a flat audience. The reach stays the same and the depth keeps filling in. Two thousand two hundred and seventy-four people, and six and a half downloads each.
Which is what I'd take from all of it. If you want one thing from this episode, it's that the metric that matters to a sponsor and the metric that matters to a show are not the same metric, and neither of them can see the size of the archive. That's on us to keep straight.
On the cave. The one-week version, if you go, take the syllabus, not the archive. Listen to the technology block first. You'll be fine.
Would you go?
Not unless you're coming too.
I'd nap the entire way through. Which is not a criticism. It's a technique.
Hilbert's been sitting in that chair all episode, and we should thank him for putting it together. Thank you to Hilbert Flumingtop for producing, as ever.
If this was your kind of episode, go back for episode seventeen fifty, Herman's Music Showcase; episode fifty-eight fifteen, Special Release; and episode thirty-five, The Privacy Gap. This has been My Weird Prompts. If you've ever tried to listen to the whole back catalogue, tell us where you gave up. Send us a prompt on Telegram at t dot me slash MWP listener bot. We'd like to know how far you got.
If you enjoyed this one, a review helps more than you'd think. We'll be back soon.
See you then.
