The capitalist's guide to podcasting, with Ross Rheingans-Yoo
This week, I'm joined by returning guest Ross Rheingans-Yoo to discuss the strategic advantages of starting a podcast. Ross shares his thought process around his newly-launched show Development and Research, while I share more about the makings of Complex Systems. The transcript will be updated with in-line notes on Thursday
Sponsor: Mercury
This episode is brought to you by Mercury, the fintech trusted by 200K+ companies — from first milestones to running complex systems. Mercury offers banking that truly understands startups and scales with them. Start today at Mercury.com
Mercury is a financial technology company, not a bank. Banking services provided by Choice Financial Group, Column N.A., and Evolve Bank & Trust; Members FDIC.
Timestamps
(00:00) Intro
(01:18) The power of podcasting for venture capitalists
(03:09) Building trust through voice
(07:11) Podcast production models
(16:38) Recording and equipment essentials
(23:05) Sponsor: Mercury
(24:18) Post-production and transcripts
(32:22) Advertising vs. User experience
(33:28) Creating standalone artifacts
(34:20) The power of video clips
(35:22) Challenges in podcast distribution
(42:12) The role of guests in podcasts
(47:12) Pre-recording preparation
(49:22) Recording session best practices
(54:22) The value of silence in conversations
(56:38) Launching and growing a podcast
(59:18) Trading bootcamp
(1:02:27) Wrap
Transcript
My name is Patrick McKenzie, better known as Patio11 on the internet. And I'm here with my friend Ross.
Ross Rheingans-Yoo: Thanks Patrick. Good to be back.
Patrick McKenzie: So Ross, you were previously on the podcast talking about drug development and discovery. You want to give people a brief update on what your work has been like?
Ross Rheingans-Yoo: So I was talking earlier about my own particular views on the field of drug development. What's possible, what's exciting. Since then I've been involved in a couple of small investments in the space and pushing along other investments that I'd made in the past. My hope is that those things are rolling towards a future where I'm doing that more permanently. Where instead of raising money deal by deal, I phrase a fund of like-minded investors to go after this thesis. But that is a long work, which has a lot of groundwork being laid, I think before the first fundraising emails go out. So that segues into, I guess, what we're gonna talk about today and what I'm interested to talk about with you today, Patrick, which is getting into the business of podcasting.
The power of podcasting for venture capitalists
Patrick McKenzie: So as many folks will be quite familiar with, one of the things that is extremely useful to have in one's back pocket as a venture capitalist, or anyone else doing fundraising, really is so-called content marketing or permission marketing assets. I think podcasts are an interesting point in that space. There are some people that build their brand primarily through writing. I think I'm probably in that camp. But for people who present well when talking to other experts and have interesting professional networks that might be, say, underexposed on the internet, I think podcasts are a really interesting way to get stuff which is very compelling for listeners. Connects well to the sort of people who make either investment decisions or receiving investment decisions. And are very sort of time-informed, factor appropriate for busy professionals, both in terms of the people who are recording it and the people who are listening to it. At the end of the day, eventually I will probably be raising money for something and I bet that at least one person who writes a check is listening to this right now. Knock on wood.
Ross Rheingans-Yoo: I didn't appreciate until I started thinking about the exercise was how much the medium is good for that networks thing. It turns out when I start thinking, I know a lot of people who also have interesting thoughts in this field, but have not been putting out content on their own. That I can be helpful to all of the people who are interested in the topic by bringing forward other people who were never going to be writing their own blog posts and so the only way that they were going to put their thoughts forward was in conversation format. And if I think that those conversations can go well, that's a good time for all involved. So I think that's one of the pretty unique things that the medium plays to, where the things that writing plays to are facility and strength of good ideas that you can build up over multiple articles, so on.
Patrick McKenzie: I will say as someone who has written extensively for the internet for almost 20 years and done podcasting for much less total time and sort of cycles expended, that in terms of trust level, the people who express when meeting me in physical reality, the people who express the greatest amount of trust for me are largely listeners and not readers. Which is surprising because I've made a lot of money over the years for many people who have read my stuff. And my sort of working backwards from this is that the amount of time someone can spend with you in the course of listening to a podcast in their own private spaces when it is just you and them alone in their thoughts is quite considerable. If they're listening to you on the drive, if they're listening to you at the gym, et cetera. And when you benchmark that against the amount of time people spend speaking to other important individuals in their life, like a significant other, or a boss, or a religious leader or similar, you hit pretty comparable levels pretty quickly as a routine podcast producer, and being in a position where someone has a sort of like ambient level of trust for you is a good place to start many sorts of commercial relationships, particularly the ones that have like large amounts of consideration and high dollar amounts attached.
Building trust through voice
Ross Rheingans-Yoo: And that operate in these networks of trust. I think a lot of the fundraising communities and startups and venture, whether in biotech and internet tech. The stuff runs on a lot of trust that the other person is really who they're made out to be, not a huckster, not fly by night. And so it is that the social network of weak ties that really means something to getting the work done. It's interesting I think, to think about what it is about voice that builds that so much more strongly than writing. I think. I don't know. Something about humans.
Patrick McKenzie: Yeah, I think it's something about the way we're wired, something about how we perceive writers as someone very different than us, connecting through the word space. But the people in our lives who we actually trust are ones whose voices we hear repeatedly. And so just directly hacking brain 1.0.
Ross Rheingans-Yoo: Seeing yourself in someone else's situation is an interesting one. Sorry, just as an aside, I was mentioning to my barber of all people the other week, like, what are you up to? We're making this conversation. I said like, oh, well I'm recording this podcast with other people. And he says like, oh yeah, I was on my friend's podcast talking about horror movies or whatever. And I've been shocked by how relatable an activity it is even compared to writing on the internet.
Patrick McKenzie: Well that's the pitch for why people who are doing interesting things can get a podcast out early about. Oh, one more little mini pitch I'll give, in the course of fundraising for a startup or fund or attempting to hire people or anything else where you're going to be sending a lot of coldish to medium coldish emails, a useful thing to have in your back pocket is a conversation starter for those emails. Particularly a conversation starter which is a proof of work slash proof of humanity, or which indicates shared interest between you and the person you're reaching out to. Having a few essays on your website is a good enough first crack at this, but if you have a podcast where you're already fishing in a dense network for guests, and you're pitching other people in that dense network for money. "Oh, I was previously on an episode with a buddy of yours." Drop the link. Many people like their friends because they like their friends and they would happily listen to their friend talk shop for an hour. And so you are much more likely to get that link clicked on than any other random link that you put in it for a proof of work function. And that proof of work will very definitely distinguish you from someone who is using, for example, an LLM to spin up two paragraphs to try to get past this person's first filter for their attention.
Ross Rheingans-Yoo: I mean, the episode of yours that I was on has been useful for this in some amount of reach outs in the allied fields or sort of associated work. But I imagine that just having a larger toolbox for that has increasing returns through many margins.
Podcast production models
Patrick McKenzie: So I think we've given people enough of the pitch of why to have a podcast. There's a few different production models for particularly cadence of podcasts that I've encountered over the years. And we'll give people a bit of a rundown for them. One is like the traditional podcast, which is what you're listening to right now. Complex Systems is produced on a weekly, every Thursday cadence. And to my great surprise, having run a podcast before actually successfully publishes every Thursday. Well done Patrick. Well done Sammy, the producer. But I guess I'm involved in the production function as well. The typical experience for traditional podcasts is that they decay from having a schedule to being sort of periodic in publication. The best reason to have a schedule is it trains people to come back to it on a particular cadence. Whenever that cadence lines in their life are similar. And for example, there is a particular podcast I listen to which comes out every Friday, which I do not listen to on Fridays because my Fridays are a little crazy. But my Saturday mornings are not. And so it is my listen before the kids wake up on Saturday morning podcast.
A refinement to the traditional model, which I think not enough people have experimented with, is what's called the season model. And you're producing Development and Research, is that the name of the show?
Ross Rheingans-Yoo: That's right. So the title that we'll be releasing under is Development and Research. You can find us online at developmentandresearch.bio, or development and research wherever your podcasts are sold. But I can talk a little about my thoughts in moving towards that.
It's an interview series about people that I know who are involved some way in the project of developing new drugs. So at researching them, producing them, investing in them, getting them globally to the world, and trying to get at the question of it is.
So expensive and so slow, and so failure prone to develop a new drug. Why is that? How have we gotten here? And what can be done differently to have a better world where ultimately we have better medicines that make it to people faster? Has been, a world that I've fallen into. Kind of by accident.
And I've just become obsessed with it and this problem. I've been fortunate over the years, to have met some really fantastic people who all seem to be pointed in the direction of we can do things differently. And I think it's that energy that has kept me stuck to it, because it's not this inevitable, terrible state of the world. There are people today who are looking at this and saying, nah, this can be different. And then doing something about it and then doing things differently. And increasingly since 2020, since a bunch of people got into emergency mode for covid that the world has been able to step back and say, wait, we could do that for Covid.
Why can't we do that all the time? And you know, it's not like that idea just dies. That there are people who look up and say, oh, I always kind of thought this was broken, but everyone else had their thing. No, you think this is broken too? You think this is broken too? Well, maybe it really is broken.
Maybe I'm not crazy. Maybe there is something different that can be done. And that has just been a project that's grabbed my attention and I'm pretty excited to be able to put out these conversations with these fantastic people that are digging away at their own sides of the problem.
I'm not going into this project with the idea of setting up a perpetual media empire that will be this weekly source of work. I've actually seen pieces of the production function of yours, Patrick, and you really have assigned yourself a piece of homework to do every week in sort of the cadence of your work. For whatever reason, I'm not particularly interested in taking on that obligation or doing that and then having it decay in kind of a messy way. And the smallest unit that I felt I could bite off in the production here was a fixed number of episodes. And then once it's going to stop, it just makes more sense to record it all upfront, get all of it ready to go. Possibly we'll be doing editing and polishing on the last episodes while the first ones are airing. But we currently have five recorded and we'll go out as five or six every week until we're done. That'll be season one. And then we'll take a break for season two, whenever season two comes.
Patrick McKenzie: So I like a number of things about this. One, I like that it front loads the work and so it's time bounded in terms of the amount of effort and resources that you have to put into it. It also has some nice natural affordances in reaching out to guests because you can say we are recording for season one between bookmarks here. And so there is a reason for the guest to say yes to you and to make it happen between those bookmarks. Because the opportunity to hit season one goes away after the second bookend. I think that many people might underestimate the amount of effort that goes into production of a podcast at professional or semi-professional levels. And you can play around with these numbers if you have increasing levels of editorial slash producer support, et cetera. But Complex Systems is plus or minus a 40% job for me, which is not the thing that I would be able to sustain if I actually had a hundred percent job at the moment, which thankfully I do not. And granted it wasn't at the top of my docket for a number of years prior to launching Complex Systems, but I had at least three, six month plans. This will be the half where we launch a podcast with yours truly as the host, for at least a year and a half to two years, prior to quitting my last day job and just never got off the ground because the productivity tax would've been too intense.
So one hats off to the people who actually make this work and still have non-media jobs and the rest of their life. And two, if one anticipates having a hundred percent job that one cannot simply put aside for Monday and Tuesday of every week, then the season model allows you to focus intensely on that in sort of the marketing cycles for the business every six months or so, for a week or two, and then go back to quote unquote your real job, in between the seasons launching. I also think that blogs have this problem too, where the form factor for a blog assumes that you're producing sort of time sensitive content that decays in value very, very quickly after you publish it. And if you stop publishing it, your blog has quote unquote failed. And similarly, your podcast, if you stop publishing it as quote unquote failed, but no TV show in history with the debatable exception of Game of Thrones after season eight, is failed just because seasons have stopped airing. It's just like, that was it, that was an artifact. The artifact has run its course and it is done. And then sometimes you get a reboot and you feel happy. And so I love the season model for the particular use case of, if you have a book to talk or a commercial need for having the podcast out there, it dovetails very well with the other things.
The other model I've heard is essentially it's a version of the season model where you know in advance the season count is fixed at one. And this makes most sense for when the book you have to talk is literally a book. But shout out to my good buddy Josh Kaufman, who is a New York Times bestselling author, for telling me this. It turns out that Apple Podcasts and very possibly other podcast apps give an enormous boost in search visibility to podcasts early in their life, to try to get them a seed set of users. And that causes there to be a reason for you to do sort of SEO for podcasts specifically, by doing a single episode of just record your five best episodes, one and done drop them on a new podcast link. Put them up, either shortly before or contemporaneous with the launch of whatever your time limited thing is, like your new video game or your new book or similar and then you're done. Because after it has achieved the launch boost, that was the entire business purpose for running it, and you don't need to justify the ongoing tax on your time and productivity to keep it up there. And if people search for your thing later, you've probably picked a title and similar, that are quite similar to the SEO title for your book or other artifact. And so it will continue operating as a friend catcher without requiring additional work on your behalf.
Ross Rheingans-Yoo: I think that the models for this stuff are super interesting that, as you say, there's the version where you hold yourself to a schedule and then you can only decay from there. The other ones, knowing that there were more in this space, I think is part of what made it feel approachable to get into it in the first place. Because if I had to take on a 40% job until a point, at which point I'd failed that job. I don't think I'd be doing this thing at all.
Patrick McKenzie: I will say one could play mix and match on the sort of stuff. Creating new podcast feeds, particularly when you don't have much of an audience rounds to free. And so I no longer have the podcast MIUs software podcast that I previously ran with my co-host, Keith for Heck. But that's fine. All things have a season and those episodes are up on the internet for people who want them. And you have the, you know, regardless of what model you pick, we are the host of the show. If you decide actually this is we're stopping the season model, we're going to periodic from this point out. You can either create a new feed or just append to the existing feed and go, and then you're done.
Recording and equipment essentials
But keeping the focus on nuts and bolts things. Let's talk about the nuts and bolts of recording. Because I feel like these are some of the things that derail people early in the life of planning. Microphones much less important than people think they are. What is your production microphone for the semi-pro to pro podcasts?
Ross Rheingans-Yoo: Wrong question, Patrick. Because we, I think have slightly different models, in fact, for the podcast that we're recording and this gets swept under one word, which can actually describe a variety of different activities. We're recording an audio stream right now. And for those of you who don't know, we're on a video link so that I can feel a little bit more conversational with Patrick looking at his face and his mouth move in real time so that our conversation comes out feeling a little bit more friendly. But none of you listeners are gonna see that video link, are gonna see that video content. And that's one model of podcast that this is an audio conversation. It's been edited and transcripted as an audio conversation.
There's another model where in fact, the video is being recorded and that the artifact that will go out into the world is the listener's choice of a zoom screen with two people side by side on the screen, plus them talking in sync, the written transcript or an audio only version, which is released along the same channels. And the conversation is also optimized for being comprehensible in that form.
There's yet another format that you can opt into where instead of us being on a call here, that it's actually an in-person conversation that we're going to arrange to be across a table from each other and having that conversation there. And then the recording is one camera, three cameras, a camera for my face, a camera for your face, microphones for us both. And then an edited artifact, which is putting all of that together into a visual plus listening conversation, a listening only conversation, a written transcript written that people can read with their eyes.
You're here on an audio plus transcript model. Development and Research is a series of in-person conversations. And we're also releasing as audio, and we're also releasing a transcript. But the in-person conversations then puts constraints on how it is that we move through space. It means that rather than recording these from the same studio that I've been able to get used to, in fact, recording has taken me all over, well, literally the world, that we've done recording sessions in Barcelona, in the UK, in New York, in Seattle. Hilariously have not gotten to one in my hometown yet. And so that has been with a different set of, with a different studio every time. With, we've gotten to reuse the same videographer for some of these. But not many. We've worked with several. And similarly the audio setup is I don't carry around an audio case of my own equipment. But rather my producer calls ahead and gets someone who has a suitable set of equipment to show up on the day and set it up with microphones in front of our faces. I honestly couldn't tell you a brand of microphone.
Patrick McKenzie: Got it. I think, by the way, this is a capitalist cheat code that too few people avail themselves of when setting up a podcast. There are local businesses in your town if you're in a city of any size, that do recording professionally, and that love getting one-off gigs, and they love even more when those one-off gigs turn into recurring gigs. And so just call them and get a quote. They will either invite you out to their space or come out to your space and just do it all for you. And then like what kind of camera, what kind of recording equipment, et cetera. Do you need? The answer is you don't need to think of any of that. You will get the professional to come do it. They will bring their professional gear, it will work fine, by the standards of almost all of your listenership. And you'll just get a consistent interface from them, which will be probably raw wave files or similar. And then you'll pass them on to your editors if those recording professionals are not also throwing in audio editing services, which they frequently will be.
Ross Rheingans-Yoo: And so with that interface, I mean, again, shout out to my producer, the same as your producer, Patrick. We're both working with Sammy Catrell, who has made the interface for me as simple as, alright, I am gonna have a conversation with the guest. I'll get to pass it off to Sammy for logistics. And then at some point Sammy will tell me at what address in the city that I'm gonna be passing through to show up at. And this past one, the most recent, the one that we've recorded most recently as of our time of recording this was in Seattle and I was flying through Seattle on the way to somewhere else for some other reason got off the plane. It was like, alright, what is the address that I'm going to again? Turns out it was an Airbnb that had been booked for one afternoon for like a couple of hours out of one afternoon. And I showed up. Sammy had arranged for a videographer to show up, an audio guy to show up. They'd set up some chairs and a bookcase and whatever. And then an hour later, my guest showed up and away we were. But the interface on my side for getting those contracted professionals was to kick it off to my producer, thankfully.
Patrick McKenzie: Just rent an Airbnb for a couple hours or a day. And then have everyone converge on that is also a model that Dwarkesh Patel has used occasionally, at least on the day that I recorded with them. And works out pretty well, particularly if one is budgeting this for a, if one has budget to run this sort of thing, through one's company, or similar.
Ross Rheingans-Yoo: If it's a professional endeavor that is tied into all sorts of other opportunities, one can justify shelling out a thousand dollars an episode. And well within that a thousand dollars an episode is an afternoon of an Airbnb when the host wasn't expecting to get it booked.
Patrick McKenzie: For folks who are a little bit budget sensitive, somewhere closer to like the well indicatively finger to the wind, 200 to $300 will buy you a few hours of recording studio time, plus a few minutes of professional time at the recording studio to get to your microphones set up and torn down, if that's an option. Many of the recording studios these days are a combination audio slash visual recording studios. The one that I use here in Chicago, plug for Content Club, which I'm currently physically located in. Apparently their bread and butter is a turnkey operation for videos, for people like real estate agents and similar who need things to put on social media. But they are very happy to do podcasts and similar as well, particularly when it doesn't require their full-time and attention for the duration of the podcast. Although, of course, these are professional services, there is a rate card. One can always buy two hours of a videographer's time if one needs that, to get one's production function to work.
Post-production and transcripts
Ross Rheingans-Yoo: Another thing that has surprised me, and then in hindsight it's not clear why I was surprised or shouldn't have been surprised at all, is how much there is a rich and liquid network, a rich and liquid market for professional services in post-production work.
For people less familiar with the media business lingo than both of us are now. Post-production is the stuff that happens after our camera stops rolling, our microphone stop rolling. And then there is still a chunk of work that happens before that recording becomes an artifact that you're going to be consuming as a consumer of this podcast. So this is editing work. This is the transcript work, and in the video space, this is video editing work for onscreen graphics and visuals and so on. So turns out it is possible to DIY all of that oneself, I'm sure. And it is possible to yourself or through someone who hopefully knows some of the ropes of this by now, get other people who do that more professionally to take on some freelance work in spare cracks of their time, to do this at a higher level than I was going to.
Patrick McKenzie: And there is virtually no one listening to this who should be doing their own editing or video editing. They are just enormous sucks of time. And there's a skill curve to climb there, and that is not the most interesting skill curve to climb when so many people have already done it before you. At the low end of the market, you can go to Fiverr and get people in relatively low wage countries who do this for essentially eight hours every workday, sliced up into 30 to 60 minute increments for a variety of people just like you or your local real estate agent who needs a video edited.
And then there are people working in your own town who do similar. Obviously this is extremely remote amenable work as well. Much of the low end of it, or particular parts of the production chain are heavily benefit from AI assist transcripts. I used to always have a human transcriptionist do transcripts. If I can make a plug for one place that almost all shows under invest, it is having a really good transcript available. Many places have historically used just the autogenerated transcript, which is barely a deliverable that I think is safe for human consumption. And then the ones that do a little bit better than an auto transcript, will produce a verbatim transcript of what was said by the guests.
Ross Rheingans-Yoo: Which is a task that can be outsourced to contract work.
Patrick McKenzie: Yeah. And even very professional, legitimate mainstream media organizations frequently don't have anything better than the verbatim transcript, for what I assume are internal sort of will to do better reasons. But you should have the will to do better. Most people don't naturally speak in sentences and paragraphs that read well, compared to other things that are written well in the world. And so simply taking the verbatim transcript and upgrading it to be what would be what the guest would've produced if they had slightly more time to self-edit their own comments. So just giving them the best version of what they actually said will make it much more likely that people will actually read all the way through the transcript, increases the utility of the transcript as an aid for people who are, common user behavior, by the way. They'll have the transcript open while listening to the podcast and occasionally like check their hearing or similar, go back and read something, while hitting pause in the transcript and similar, and so a small bit of extra work on your end, small bit of extra expense will have a great increase in the marketing boons for having that available. Also the obvious marketing boon is like, the transcript is SEO gold. And yet people don't do that, which is just a straight error.
Ross Rheingans-Yoo: A slight correction there. Patrick, you said a small piece of work.
Patrick McKenzie: Oh, it's a huge bit of work. If you do it, it's a huge piece of work. I think about 50% of my time on complex systems is spent on enhancing the transcript after a human has already enhanced the transcript.
Ross Rheingans-Yoo: After a human has already enhanced the transcript as you say, so recently I was saying, okay, look, the AI can produce the words. Hopefully the AI can do the first pass of turning this into something that's useful for humans. And I'll just go over it once more and do it again. No, turns out I don't know whether I'm holding it wrong or what, but the thing which came out of the smart AI, top of the line, state of the art service was like, no, I need to go back and edit this so that these sentences make sense next to each other. Or we cut off the aborted half sentences sort of interruptions in unintelligible crosstalk that somehow made it into the transcription. Anyway. And then I'll go back and clean this into whole sentences and coherent paragraphs and thoughts. And that work, I took 10 minutes of tape to see how long that took me, and that took me very nearly an hour to go from raw, this sure is words on the page, to something that I was actually going to be ready to release. And that was the point at which I turned to Sammy and said, Sammy, can you get someone to take the first pass on this I generated artifact so I can take the second pass? Because that's just a more efficient use of everyone's time under capitalism.
Patrick McKenzie: Yeah. I spend about three to five hours on the transcript, and again, that is for an hour long episode, to be clear. I generally spend about three to five hours of work per hour of recorded time of an episode in doing post-production for enhancing a transcript. And again, that is after the transcript has been first processed by one or two AIs, and then hand checked by at least one smart human who is familiar with the show, my own way of talking, et cetera.
What do I do during those times? Partly it is just correcting errors or other infelicities in the text of the transcript itself. Partly I do a thing which very few other podcast hosts do, but more should, in addition to the actual text of the transcript I have other than contemporaneous reactions to the conversation has happened and I interject those as inline notes of the transcript. And if there are things like citations or "oh, this brings up another conversation that we have," et cetera, can interject those into the author's commentary, which one provides a reason for people who have already listened to it, to actually read through the transcript, if that is a good use of their time. And also gives me an opportunity for doing things like corrections that aren't corrections. So sometimes you will give a stat on error, which is wrong, and one option is like chopping out that sentence. And if one holds oneself to those standards, great. But that is a painful thing to do when you're chopping up an episode. Sometimes it is just easier to do the correction in a textual note to something particularly where the stat that is wrong is not particularly load bearing for an argument.
Ross Rheingans-Yoo: Yep. I think on top of that workflow, and this is a thing which I've encountered, which I guess you haven't, Patrick, is that you can easily spend another three to five hours on video and making the video be in a particular state. Both from just notes to the editor, like this is how you chopped it up. But I would've done it slightly differently through, and I guess in the theme of you're like, many people don't, but it's probably better if they just do put helpful things in the transcript. Many people don't, but I think it's better if people just do have, use helpful things on screen. Sharing the screen with the talking heads, whether that's graphs, whether that's snippets of text that it is worth reinforcing, whether that's chapter headers or any visual models of something that we're trying to paint with words. Any of these things, it is possible to get all of these things in your video, going through your video and saying actually this part, we wanna cut this part. Or like this half sentence actually confuses the issue more than it clarifies. Oh, also, right here, I want a picture of a molecule in front of everyone so that we know what we're talking about. Going through an hour of video and making all of those notes to one's editor can again, take you three hours per hour of video.
Patrick McKenzie: If I can highlight one thing there under the definitely, definitely do this: linkable chapter headers, even if no one else in the world ever linked to one of them, they would be astoundingly useful for you because you are in the future, going to be linking back to your own episodes and dropping people at exactly the right point is much better for you than dropping them at the start of the episode.
Advertising vs. User experience
This is one of those things where the advertising based business model causes people to suboptimize for the user experience in ways that suboptimize for their own use of the product. Where the attempt to just buy more user minutes causes people to want to say, rather than having folks select to the right part of a 90 minute episode, I would love to have them watch all of the 90 minutes so that I get 90 minutes worth of advertising rules where that is never what you would choose to do as a consumer of something you would want the exact three minutes that you care about, and then go on with your day. And given that in the future you will be in consumer of your own podcast mode rather than advertising sales rep of your own podcast mode, you will want the capability to drop people into a three minute section. So build that capability by having linkable chapters available.
Creating standalone artifacts
Ross Rheingans-Yoo: The other version of this, and I was talking with a friend of mine in video production for a different organization is that you can either link people into a three to five minute segment of the broader thing, or you can link people into a standalone artifact, which is a three or five minute cut of the relevant pieces that brings some thematic topic together. And so this is the eight minutes that is all three times that this topic came up in the conversation. Or alternatively, super cuts across a season or between different guests that put them even closer in conversation to each other than if you have to listen to all of one episode and then all of another episode in someone's ear. So these are underdone as we're making artifacts for our own future use, as crystallized pieces of conversation we can send to other people.
The power of video clips
Patrick McKenzie: I think the sort of industry state of the art with respect to clips is better in video land because of the way that clips work in YouTube and other similar. TikTok, people not ironically suggested for many years I should have a TikTok for business and professional purposes. And man, I don't know. Anyhow, my joke's a TikTok and its users.
Ross Rheingans-Yoo: A platform that we both use more than that one, Patrick, is people will be scrolling down through Twitter and they'll stop if they see two human faces talking. Because our monkey brains like checking in with that thing, the words under them start moving and you start following along. You can get trapped in that for I mean, you can get trapped in that for a whole episode. If the platform lets you, if you've posted it right there in the Twitter feed. So turns out there's something about just the words across a blank screen that doesn't do the same thing. The best you can do is a clip of the transcript. And then, but often people will consume what's there and move on that it's you don't really have the opportunity to get caught into minutes of consumption by accident, as it were.
Challenges in podcast distribution
Patrick McKenzie: I'll report just from my own experience that I have not been able to get much marginal distribution for complex systems as a result of using Twitter, despite having tried a few things. And possibly I'm just not investing enough in trying or doing interesting audio visual things with it. I do know that some of the more sophisticated tech podcasts and Dake comes to mind, get enormous leverage out of clip flying episodes, particularly through YouTube. Ironically it turns out that my episode and DSH was probably suppressed or semi suppressed by the YouTube algorithm for talking too much about covid vaccines. So plus from the past and getting suppressed for that particular topic. Patrick, ironically, in a episode, which also had many sharp words for the tech slash government censorship apparatus, but be that as it may. But there was a clip of the episode produced about money laundering specifically, which was not suppressed because we don't suppress that topic, right? And so the best distributed thing I've ever done on the internet was me talking for two minutes about how easy and fun it is to do money laundering, which is obviously not the pitch that I gave for money laundering, but talking about the mechanics, et cetera.
Ross Rheingans-Yoo: This makes you perhaps the second most palatable clip of talking about how to do money laundering of all time, right?
Patrick McKenzie: Oh, there's many clips that are competing for that claim to fame. One that comes immediately to mind is the breaking bad, Jesse Pinkman.
Ross Rheingans-Yoo: Well I said the second Patrick, I think exactly one. I think there's one clip that's competing for the first best. Is it exactly The Breaking Bad Jesse Pinkman and Paul Goodman? Exactly, exactly. The Breaking Bad episode. And you certainly can aspire to be second behind that. That one is so good.
Patrick McKenzie: I will drop a link to a YouTube version of this two minute sizzle reel for the Breaking Bad Laundry money laundering discussion. But it's literally good enough to be in compliance presentations at like real life professional services and financial firms that took question aside. Best distributed thing I've ever done for the internet is a two minute chat about money laundering, which was excerpted from a hour and a half long chat, I think on the Torish podcast. And so they make incredible return outta spending. I would model it as 10 per tens of percent of the post-production effort there goes into like, let's produce the clicked artifacts that will maximize your distribution of the larger artifact that we've just produced, which we sort of internally conceptualized is like the core artifact of the show.
Ross Rheingans-Yoo: This is my understanding of it, and this is, I think, a big part of the reason beside, behind deciding to do video in the first place. This has been an annoyance in a number of ways, and I can count one is possibly two guests that we simply have not been able to have on because we've not been able to arrange the in-person recording logistics. That's sad. But I do, my hope is that the benefit of having the visual component, especially for clips, especially for being able to produce clips of video in the way that clips of audio don't work, makes it worth it. We'll see. So I'm talking up the advantages of this here, and at the time when you're listening to this, dear listeners, development and research will already have its first episodes be live. Nevertheless, at the time that we're recording this, Patrick, we haven't launched yet. So this is another thing which is secretly happening in every podcast you're listening to, if you, but it's rarely talked about on air. I can think of one podcast where the hosts are constantly joking about it.
We're recording this at a minimum a few days in advance because of the editing process is going to have to go. But possibly weeks or even some months can be pretty transparent to the listeners if something doesn't go out of date. In the meantime. So that is I think not always appreciated, but yeah, it's just kind of lurking behind the surface on a bunch of this stuff.
Patrick McKenzie: I think this is a thing where the season model and similar bid for you because it is more obvious to people in a season model that things were prerecorded. I can think of only one time where there was a serious change in current events between the recording room podcast and the publication of a podcast where the producer thought I should definitely either edit something out or edit in context. Luckily due to the tradition of having very elaborate show notes, rather than editing it out, I just put a "hey, this," it was actually the episode about prediction markets, and I quoted the price of Biden being the Democratic nominee, when that price was 95. And by the time the podcast dropped, I think the price was 15 and there was a bit of a delta that needed to be explained, which was obvious for anyone who had seen the intervening weeks, but would've been less obvious if you didn't know that the podcast was four weeks old.
Ross Rheingans-Yoo: Without giving too much away about episodes yet to come. Patrick, turns out the few months that have preceded us and that are yet to come, have seen a lot of change in the particular topic of how medical and pharmaceutical research gets done in America. And so one of the things which is kind of forcing our hand on we need to get DNR out the door really sooner rather than later, is I think there are some portions of our discussions that are in danger of going stale just as the world around us changes. And some of this is, we've lived through a rather faster changing time very recently than over last fall when you're talking about, some of this is subject matter. And some of this is just happenstance. But there are some of ours where I look at it and I'm like, I kind of wish that we didn't have two months between recording and release, because now this sounds a little bit off focus, but I don't know. The alternative is if everything is optimized for being timely on the scale of days to weeks, as you said about blogs. When it's six months out, it's fallen off its own utility. From that perspective,
Patrick McKenzie: I think that people who are not a mainstream media news organization should not aspire to the timeliness of mainstream media news organizations. The New York Times can afford to throw out some of the world's best writing every day, but most people don't have that luxury. And given that you are producing artifacts and it takes you the same amount of effort to create news versus creating something that is evergreen in character and quality, produce the thing that is evergreen and continue to reaping the results of that for years versus producing something where it'll be yesterday's news tomorrow. And you know, if occasionally episodes get interceded through intervening events, et cetera, that's an opportunity for you to do an update episode, et cetera, versus just assuming that all of your episodes will get interceded after a period of days or weeks.
The role of guests in podcasts
So we've mentioned the sort of backend of the pipeline management problem that you have the episode in the can and there is some amount of time during which it's going through the editing process and/or waiting on an upcoming slot in your publication calendar. The thing that's I spend the vast majority of my time on in pipeline management, and that Sammy spends even more time than that on is the pre-recording part of pipeline management, I would say broadly. So this is specific to shows that routinely have guests involved in them and complex systems has a guest in the vast majority of weeks, I would say more than 80% of episodes. There are other podcasts that have the same two to three people every week, and I envy them for how easy that must be.
One has sort of a addressable universe of guests that one would aspirationally want to have on one's show. And at any given time you have some subset of that that you're working through various emails with, where typically there's an exchange on getting an initial indication of interest and then conditional and initial indication of interest that you try to nail down a mutually amenable recording time. And given that the people that you have as desirable guests are likely public intellectuals or extremely professionally scheduled. There is some competition for their own calendars, and there's also competition for your calendar and for the availability of your recording studio, the rest of your team, et cetera. And so it can be anywhere from on the low end, two days on the high end, six weeks plus, plus plus, to get someone from the initial indication of interest to the point where they're actually in the studio with you recording the episode. And then it can be out 48 hours later. It could be out two months later, depending on what your publication calendar looks like.
Ross Rheingans-Yoo: Yep. That is something that even more than the backend post-production tends to be outside of our control as hosts or when people respond to our emails, is largely outside of our control as hosts. And like most external facing games. It's a matter of funnel and pipeline management and that I have been shocked by how good my conversion rate has been moving through the funnel. I certainly have had people that I thought would be on who didn't want to talk publicly for whatever reason, and that surprised me about them. But then I learned something. People were scheduling falls through people wear logistics fall through for whatever reasons. But maybe I'm not asking enough people if more than half of them have converted into episodes. But it didn't take me more than two times the number of episodes that I had to get the number of episodes I recorded. I dunno if this is different from your experience, Patrick.
Patrick McKenzie: I think both of us have a relative advantage compared to most people who will be attempting to do this due to position and career position, network, and similar. One can sort of award oneself an advantage by making it very easy for people to say yes. One way is to, regardless of whether this is early in your podcast career or not, present to someone who has done this before. Sound like someone is talking to a responsible, well operated, professionalized organization versus "hey, would you like to have a zoom call with me sometime?" And just make it easy for the guest to say yes, like we'll propose a couple of times that work for us, or you will propose a couple of times which work for you. You will pick one, be very clear about expectations for what they have to bring and get that expectation as close as possible to, it's you schedule one meeting at the margin for yourself and you show up to it like any other meeting that day. And then we take care of all the rest of it, whether that's you take a lift to the meeting and it's in person and like the team handles everything from there.
Or the technology I use for recording audio interviews is Riverside, which is amazing in taking out all of the guest work for guests, specifically for audio production. In that it's basically like click on this link in Chrome. There's nothing to download. You get one prompt about your microphone and then it's go time. Why do that for remote recordings? Your audio engineer long suffering audio engineer will have told you over the years that doing both of the recordings from one side of the conversation works very well when every possible star aligns. But as soon as there's any sort of network latency or a jitter in the connection or similar and things desynchronized, you lose the sentence or the paragraph. And so they will ask you, do you have both sides record audio locally and instructing a guest who might or may not be the most technical person in the world, how to run an audio recording tool chain on their own machine, is a great way to burn a lot of your producer's hours for not all that much listener benefit versus Riverside.
It just does a local recording on their machine through the magic of Chrome Web technologies. And then ou have to explain nothing to them other than like, click on this link and it's go time. And then at the end of the episode, after you hit the stop button on the recording, it takes a few seconds to a few minutes to synchronize their local recording to the cloud. After which point, you no longer care about what happens on their machine. I dunno what we pay for Riverside, but whatever it is, they should charge more.
Ross Rheingans-Yoo: Entirely on brand, Patrick. I mean, the in-person experience is different, that you show up and like most people need to be told to talk directly into the mic from a distance of an inch and a half. But once you've put the boom in front of their face, they'll probably do that by accident and then you're kind of off and away. But yeah, no, that makes total sense. That and explains why the URL that we're at is Riverside.fm and not, well, you can fill in the rest of ways that we could be looking at each other's faces and also hearing each other's sports.
Pre-recording preparation
Patrick McKenzie: So what do you do in terms of preparation for guests after they've said yes to come on your program?
Ross Rheingans-Yoo: Good question. I like to do my research, but even before that, I like to have a short 30 minute conversation with the guest beforehand, schedule's willing. And we've missed this a couple of times, and I think every time that we've missed it, I've kind of regretted it, in that I wish that I'd had more time to prep the very first thing that would've come up if we just like met over coffee to talk about what someone's deal is. Even people that I think I know well. Stuff comes up and it's like, that would've been great, would've been great if I had known at the beginning of the episode that was gonna come up 40 minutes in. So I really like sort of having any chance for us to say words to each other and then like the stuff that's really gonna come out is gonna come out and then I can go off and research it and be a little bit more knowledgeable when it is that I come back to the session. As I said, we have not always succeeded at that. But when we haven't, I've been sad in at least one way because of it.
Then at the beginning of the session, it's not that one just sits down and then every word is from the very first instant ends up on camera. Often we'll have a chat about what are the topics we're gonna go through. Is there anything in particular that it would be terrible for like both of us and all of our audience if we failed to talk about? Unless your goal is to be surprising people and getting particularly candid reactions, and it is possible to be doing that even in a friendly fashion. But unless your goal is to be surprised, I do think that people tend to present ideas that are better for themselves than all of your listeners if they have had any minutes ahead of time to compose their thoughts. So letting people compose their thoughts.
I think the other benefit of sitting down for 20 minutes before an episode is just we get used to talking and we loosen up. And our own speaking cadence, sitting comfortably in the chair or whatever, before it is that we actually start rolling the tape. I don't know, I've only been on the sort of behind the camera for two of your episodes, Patrick, does that line up with how you run these?
Recording session best practices
Patrick McKenzie: So there are some things I do differently and some things I do the same. I would suggest that everyone have some buffer time when doing a recording session, one for the AV check. Two, it is embarrassing the number of times that every professional podcast has recorded an episode without hitting the magic record button.
Ross Rheingans-Yoo: Oh, no.
Patrick McKenzie: Oh no. This is a rite of passage for everyone, so make sure that magic record button is on. But start the recording before starting the recording as it were, and just have some loosen up, like reach out about what you're going to hit. How's your day going today, et cetera. And then getting to your five count and get into the actual episode proper, and your audio editor can blow away all the stuff that came before the start.
Ross Rheingans-Yoo: Oh, it is miraculous, how many mess ups and false takes and chats and half sentences and unintelligible crosstalk ones audio editor can blow away.
Patrick McKenzie: I do not organize myself to have a pre prerecording day chats with guests. That is plausibly something I should do if I were to do it. The thing I would ask for is not a second meeting, the first to have the prerecord chat and second half the recording. I would ask for a total of one meeting. And then, by the way, if before your scheduled record date of May, whatever, you ever have 15 minutes, just gimme a call at blah, blah, blah. I'm on schedule most of the time and we'll just chat. That's not bad. Talk about things. Pretty high uptake rate on that, the day that you send that email. So send it on a day where you're not in the studio. But just the quick on schedule chat with people who are like 80% scheduled but not 100% scheduled, probably works pretty well for them. And doesn't feel like they're taking two meetings because they haven't sacrificed two slots of their heavily scheduled busy professional calendar.
Ross Rheingans-Yoo: No, that's good. And perhaps I'll exploit that particular one as we schedule season two.
Patrick McKenzie: The other thing that I do is different hosts have different levels of how formalized they make this, but there's often a research process at many communications departments before either someone appears on a podcast or before someone appears on the podcast run by that communications department. There will be a, call it, three to four page dossier worked up on the counterparty, which this is the sort of work that AI, particularly deep research, is getting very very securely good at. The classic assignment that you would give a junior employee for preparing one of these briefs is go listen to the last like three or four things they did. Read the three or four essays from their blog or other public intellectual sort of work. Figure out who this person is, write some bullet points about that. And then optionally write some areas to probe on for questions. Often the more junior employees don't really come up with all that great questions, but you're just looking for some identification of areas to work with. And then the host is responsible for making the interview, actually seeing on the day or beforehand.
Different. There are some podcast hosts who prefer going into podcasts with a list of questions that they're going to get to. I don't, partly just for the vibe for the program I'm going to is a free ranging discussion between experts. Partly I've had the experience of someone who is interviewing me clearly from a list of questions and just subjectively as a guest, enjoyed that experience less than I do, having a conversation with the same person when they don't have list of things that they are trying to get to the end of. But to each their own.
Ross Rheingans-Yoo: I have gone both ways on this one. Within the season actually, in that some of these, I at best had like three things that I expect to hit on floating around in my head that we talked about just after we'd sat down. And then hopefully we get through at least two of them in the actual course of things. Sometimes I actually have cards that I have written out questions or question stubs on and between. Often they're not in order and there I am reordering them, reshuffling them. It's not, certainly as the host, I don't want to have the experience of like, and here's a question and here's a question and now we're gonna move on to the next. But often I like it has been helpful to just be organized. And I think it is possible to accidentally leave out something that would be good conversation and that seems like a shame.
Patrick McKenzie: Quick tip for people here, by the way, you typically don't have a recording session so booked that you will be talking until the last 30 seconds of the recording session. And so you can offer your guests the opportunity to get up, stretch, and take a drink of water and similar, just a five minute break in the middle of a recording session. Which is a good time for you to reorganize your notes. Reflect on okay, what have we talked about that I expect to talk about, what have we still not talked about yet, and given a few minutes outside of the needing to like flap one's gums to keep the conversation in going mode, some time to introspect on things. Okay, what is the next natural place to take this conversation? So this is a cobbler's children have no shoes sort of situation for me. I do not routinely take a break in the middle of my recording episodes, but I'm reminding myself do more of that in the future.
Ross Rheingans-Yoo: Think there's something about like the temporality of edited conversations in that you can take a lighter or heavier hand in the editing. But one thing that you really do just have an affordance for, given that this is not being streamed to people live, is we can have an awkward 32 second pause where we both just like think about what it is we're going to say next. And I think there's a way to do that badly, but there's a way to do that where we think about to ourselves or like just in conversation like, oh, Patrick did, is there anything that we've missed? Can we like think about this together? Oh, okay. And then we'll jump back into the conversation. It is possible to make those things be smooth.
I think a personal failing in my sort of in the flow of a conversation is that if you find yourself stuck in thinking for like a second and then feel like you need to catch up to the speed of the conversation to get an average density of words in. This is actually the hardest thing for your editor to deal with because they certainly can take out the pause. But then taking out the part where you gave some rapid fire catching up from the pause, like there you're just kind of out of luck. So unlearning some conversational habits, which are maybe bad habits anyway, has been one of the most interesting experiences of going back, watching my own tape and being like, it really would be easier to edit if you didn't do this.
The value of silence in conversations
Patrick McKenzie: I will say from both getting good at in-person conversations and public speaking in my life. The thing that people underappreciate is that a few seconds of silence improves most discourses versus disimproving them. And your audio editor will certainly appreciate three to five seconds of silence versus three to five seconds of verbal space filling.
Ross Rheingans-Yoo: We're trying to, well, if you think about it then, hmm, that's, yeah.
Patrick McKenzie: For similar reasons. Throwing up a hand sign to say timeout and collecting one's thoughts with one's guests works well. The only anti-pattern I've discovered there is often one does not pick a natural moment to say, okay, timing, let's allow the editor time to catch up and then creates a difficult splicing job for them, which might jar with the last few words of the part that did end up on the tape. So if you do that, definitely "un-timeout" when you realize that you are discussing gold, but have a natural jumping in point for both your editor and the listeners.
Launching and growing a podcast
Ross Rheingans-Yoo: Yeah, no, totally. What else have you found surprising about launching and then watching a podcast grow, Patrick? I mean, I haven't been through the experience of launch as of the time of this recording.
Patrick McKenzie: Growth has taken a very long time. I have some notion of how large my audience is on the internet, and the podcast captures a very small portion of my total audience relative to the amount of time that I've put into it. And the way I rationalized that as well. I've been growing the audience, as it were for 20 years, almost at this point. And so with less than one year in the podcast, like you wouldn't naturally expect to get to a hundred percent of that. But I expected, oh, you know, 60% of like the Twitter followers will be listeners. No, very different format, very different job to be done, et cetera.
And so it was not clawing my way from zero, thankfully, that would've been difficult to sustain. But the typical listener numbers for this podcast are lower than I expected them to be. And they have not been growing at the rates which I aspirationally expected them to grow at, which is not a thing to say on the air when one also wants to do sponsorship negotiations in the same month, but is the thing I'm saying anyway.
There is some magic to consistency. There is some magic to just continuing to turn the crank, there is some magic to getting better at your own production function. And then things take time.
Ross Rheingans-Yoo: And as you said, not all of the value here is coming from the listener base necessarily versus the artifacts that one can extract out of this for later use. Frankly, the networks. It is easier often to email someone and say, would you like to have a conversation on my podcast? Than in fact any other kind of conversation you could ask for. You get better uptake on that than like just meeting over coffee and on some margins. So there are benefits to doing this thing besides just how many thousands of listeners you have, Patrick, and just how many advertising dollars you can sell to make.
Patrick McKenzie: To make explicit something which is implicit. While I do have advertising on Complex Systems, I don't expect the advertising to be a large portion of the economic benefit over sufficiently long time horizons, which is a nice thing about podcasts and internet writing more broadly because the distribution of returns of brains that you get connected to is like heavily right tail skewed. If you're getting the right people, the total numbers matter much less than if you are on a CPM sort of model where all eyeballs are essentially equivalent, which very few CPM models are these days. But that as CPM, now just defining for people who haven't worked in advertising before, it's cost per mill, which is the amount you're paid per 1000 impressions of, for example, a video or similar.
Changing topics a little bit. You had one other professional thing that you've been interested in recently, with a previous guest of this podcast, Ricki. Can you give people a bit of an update on that?
Ross Rheingans-Yoo: That's right. So I've been up to a couple of things the past six or so months. One of them is this podcast and the associated work in the space of biotech clinical trial operation focused investment and thereabouts.
Trading bootcamp
Running in a parallel track in my life has been work with a former podcast guest of yours, Ricki Heicklen, and a former colleague of mine. Together we've co-founded an organization called Arbor, which is running, which is producing some educational content events, and things around quantitative trading as a way of thinking and what is it useful for and who can benefit from it.
So the most developed product that we have so far has been a quantitative trading bootcamp. This is a two to five day workshop, where we take people who are often naive to the world of financial markets, don't possibly don't know what a stock is, the difference between that and a bond have never done trading with the goal of making a profit before. Certainly some people have and come through our doors anyway. But with this audience, often the question is, what is useful about this way of thinking? What can I learn that is useful in whatever else it is that I do?
By learning to think like a trader and our belief is that the best way to learn to think like a trader is to do as much trading as you can in a period of time. And so that involves us taking people from the level of familiarity where we certainly support zero to the point where they can be trading in simulated environments with each other, reflecting on their own mistakes, working in teams, looking over each other's shoulders, and having that good discussion so that at the end of however many days it is, they have different ways of looking at the whole world, not just the parts of it that are strictly about trading and securities.
So that's something of the pitch for our bootcamp workshops. The next of them is June 3-5 in Berkeley. You can sign up at trading.camp, and you can also sign up to get notifications about our future events. You can get the details of our bootcamps and sign up for notifications about all of our upcoming events at trading.camp/complex. You can also, if you're clicking through there, get a $100 discount on signups for the next couple of our trading events, at least through the midpoint of this year or however long we can run that promotion.
Patrick McKenzie: I am a great fan of yours, obviously, and of Ricki’s. I ran the Starfighter back in the day that had its own trading simulation and really liked Ricki’s thoughts on building sort of credible trading simulations with a good pedagogy to them. She discussed them extensively in her episode. I'll drop link in the show notes. But while I haven't actually attended one of the in-person events, people that I know that have been to them have said good things.
Ross Rheingans-Yoo: We should have you by sometimes Patrick, who knows?
Patrick McKenzie: Thanks very much Ross for taking the time today. Best of luck and skill with the new podcast. I'm going to be an eager listener once it is available on the internet. And can we give a plugin? Where should people find that when it's available on the internet?
Ross Rheingans-Yoo: Sure. It's been a pleasure to be here, Patrick. We're live as of the time you're hearing this at developmentandresearch.bio. And there'll be a Twitter handle I suppose.
Patrick McKenzie: And presumably searching for Development and Research in one's podcast client of choice will also work.
Ross Rheingans-Yoo: Well, that too.
Patrick McKenzie: Alright, well thanks very much Ross and for the rest of you, thanks very much and see you again next week on Complex Systems.
Ross Rheingans-Yoo: It's been a pleasure, Patrick.
