> Because often enough my guests can’t just answer pretty complicated fractal questions in a satisfying way on the spot
Given this, perhaps you should try a dialogue with a guest over multiple emails, condensed into an essay. This could work in cases where you feel there is benefit from going deep and the guest has the time to dedicate to writing arguments and responding to yours?
Dwarkesh, the work you are doing is awesome. As was pointed out in another comment, please don't underestimate what others are learning from the podcasts. You have many very intelligent listeners, I'm sure, but few have the deep understanding of any particular topic that you and your guests have. Your hours of preparation allow for conversations that are deep, detailed and highly-informative for the rest of us. So while you may not be getting that extra kernel of knowledge you are looking for, the rest of us are getting a popcorn bag of full of insights and understanding.
And extra congrats for the string of unbelievable guests and conversations you've had of late. A+ stuff.
>> The main angst I’ve kept receding back to over and over is, “Okay what did I actually learn from this interview? And if I didn’t get that much concrete insight and understanding out of it, despite a week+ of research and hours of conversation, what hope is there for the audience? And if no one learned anything, what the fuck are we doing here?”<< Your audience doesn't have the benefit of your prior research; so your questions themselves are often a learning, even if the answers are not fully satisfying. In addition, if you do a thousand conversations and only ten survive as meaningful for future generations fifty years from now, that in itself would be worth the endeavour.
RE: helping people actually learn valuable things from podcast interviews:
The podcast that has probably given me the most valuable insights is Econtalk, and if I had to guess what Russ does that causes that I'd say that he's willing to go really slow and linger on one seemingly simple point for a long time and explore it from every angle.
I think in the moment it probably feels more entertaining to listeners to move on from topics faster, but doesn't leave much of a lasting impression because absorbing new ideas is hard. You already linger on points more than most other podcasters but I'd be curious what'd happen if you leaned into that even more.
Another thing I wonder is: how would your podcasts be different if you pretended the convo wasn't being recorded and your only goal was your own selfish learning during the time you talked to the guest? I think you're already way higher on the dimension of 'having the convo you want to have' than other podcasters but it seems like it'd be interesting to try to max that out.
The spirit of enlightenment and thanksgiving truly resonates.
In 2025, I can't think of anyone else who has had a bigger impact on AI discourse, particularly in a forward-looking and productive way.
A multi-channel approach makes sense going forward since it's ever more difficult to maintain foresight and to communicate. Some amount of personal sharing helps as well. It's good to share the ingredients of "taste."
The thought I had after listening to the podcast for a while had been that this has the potential to be the defining voice of the coming era (had shared the same in my email). Glad to see your vision for the podcast.
In 2026 - I would love to see the pod focus more on the real world impact (both guests / content, and the audience). Someone (really, a number of people, who hopefully would be aligned) is going to have to serve as the bridge between labs / big tech and the real world. It is definitely not going to be the politicians. It is not going to be tech leaders. It is not going to be the regular media. It is not going to even be most of the tech / new media (Conflicts, trust gaps, knowledge/understanding gaps). Fine needle to thread - but you are well positioned
Zeynep Tufekci have an amazing and incredibly bleak talk about AI and the future at neurips today. I think she'd make a great guest. I'd love to hear you two disagree about basically everything.
This is cool as an opportunity to experiment with format. Perhaps a mix of pre-written questions provided in advance and spontaneous questions, interstitials where you break to elaborate on something like pablo torre's pod, exploring some concepts with the guest before you go live to give them a framework, and essays that frame a pod and orient the guest are some possibilities.
> Because often enough my guests can’t just answer pretty complicated fractal questions in a satisfying way on the spot
Given this, perhaps you should try a dialogue with a guest over multiple emails, condensed into an essay. This could work in cases where you feel there is benefit from going deep and the guest has the time to dedicate to writing arguments and responding to yours?
Any more book reviews coming our way? I quite liked the one on Robert Caro's LBJ series, and Notes on The Prize was a good read as well.
I'll consider doing more of these, thanks!
Dwarkesh, the work you are doing is awesome. As was pointed out in another comment, please don't underestimate what others are learning from the podcasts. You have many very intelligent listeners, I'm sure, but few have the deep understanding of any particular topic that you and your guests have. Your hours of preparation allow for conversations that are deep, detailed and highly-informative for the rest of us. So while you may not be getting that extra kernel of knowledge you are looking for, the rest of us are getting a popcorn bag of full of insights and understanding.
And extra congrats for the string of unbelievable guests and conversations you've had of late. A+ stuff.
This is great stuff, Dwarkesh. You're a shining example of the pay off one can get by just working hard at something other people are half assing.
>> The main angst I’ve kept receding back to over and over is, “Okay what did I actually learn from this interview? And if I didn’t get that much concrete insight and understanding out of it, despite a week+ of research and hours of conversation, what hope is there for the audience? And if no one learned anything, what the fuck are we doing here?”<< Your audience doesn't have the benefit of your prior research; so your questions themselves are often a learning, even if the answers are not fully satisfying. In addition, if you do a thousand conversations and only ten survive as meaningful for future generations fifty years from now, that in itself would be worth the endeavour.
RE: helping people actually learn valuable things from podcast interviews:
The podcast that has probably given me the most valuable insights is Econtalk, and if I had to guess what Russ does that causes that I'd say that he's willing to go really slow and linger on one seemingly simple point for a long time and explore it from every angle.
I think in the moment it probably feels more entertaining to listeners to move on from topics faster, but doesn't leave much of a lasting impression because absorbing new ideas is hard. You already linger on points more than most other podcasters but I'd be curious what'd happen if you leaned into that even more.
Another thing I wonder is: how would your podcasts be different if you pretended the convo wasn't being recorded and your only goal was your own selfish learning during the time you talked to the guest? I think you're already way higher on the dimension of 'having the convo you want to have' than other podcasters but it seems like it'd be interesting to try to max that out.
The spirit of enlightenment and thanksgiving truly resonates.
In 2025, I can't think of anyone else who has had a bigger impact on AI discourse, particularly in a forward-looking and productive way.
A multi-channel approach makes sense going forward since it's ever more difficult to maintain foresight and to communicate. Some amount of personal sharing helps as well. It's good to share the ingredients of "taste."
Ignore the haters and keep up the good work!
Love these self reflections Dwarkesh! Thanks for posting them 🔥
The thought I had after listening to the podcast for a while had been that this has the potential to be the defining voice of the coming era (had shared the same in my email). Glad to see your vision for the podcast.
In 2026 - I would love to see the pod focus more on the real world impact (both guests / content, and the audience). Someone (really, a number of people, who hopefully would be aligned) is going to have to serve as the bridge between labs / big tech and the real world. It is definitely not going to be the politicians. It is not going to be tech leaders. It is not going to be the regular media. It is not going to even be most of the tech / new media (Conflicts, trust gaps, knowledge/understanding gaps). Fine needle to thread - but you are well positioned
The most impressive part of this whole post is that you can regularly get enough friends together to play Blood on the Clocktower!
Excited for the new essay era - and interested to see you experiment with more long-form narrative conversations on the podcast as well.
You've made me fall in love with AI. It feels like concrete philosophy that is impacting us TODAY. We're basically Plato.
Can't wait for you to find the next Sarah Paine
Zeynep Tufekci have an amazing and incredibly bleak talk about AI and the future at neurips today. I think she'd make a great guest. I'd love to hear you two disagree about basically everything.
The Twitter drama was hilarious and I had a great time
Well done. Been an inspirational journey to watch from afar
This is cool as an opportunity to experiment with format. Perhaps a mix of pre-written questions provided in advance and spontaneous questions, interstitials where you break to elaborate on something like pablo torre's pod, exploring some concepts with the guest before you go live to give them a framework, and essays that frame a pod and orient the guest are some possibilities.
And perhaps having multiple guests paired selectively could work too