I think you're exactly right but maybe I'm missing something in wondering if the distinction isn't online v. offline but social media vs. other online communities that could have flourished. In my own field, higher ed, I wrote some years ago that universities should not have ceded so much online real estate to social media (Facebook) at the cost of their own campus community (seminar discussions, late night academic debates, gatherings on the quad). Even now, Facebook and Twitter (and maybe now BlueSky) are the places so many academic debates take place when they used to take place on campus. If the university role in "raising" the level of conversation about serious issues is diluted, that's partly what happened. Is social media more democratic, with fewer gatekeepers? Sure. But who profited from this and who lost status. https://www.insidehighered.com/views/2021/02/16/colleges-should-build-their-own-social-media-platforms-instead-relying-facebook
> And time-plateau doesn’t mean a revenue plateau. So maybe a large part of what we see as attention above gets redirected into agents acting on our behalf. Granted, they are likely to be less impacted by advertisements per se, but there will be new methods of economic rent-seeking.
AI agents are the obvious next step, agreed. And much like the current duopoly gets $200-$300 of value per US consumer annually (FAR more than 90%+ of people would be willing to pay for the FB or GOOG bundle of services) by offering "free" services and scale that can only come from "free" (roughly ~2B users each), we should expect similar dynamics in the AI assistants.
Some portion of people will pay for 'real' agents that are wholly aligned with their self interest, and which will compete to offer better outcomes...but the overwhelmingly vast majority are going to be using "free tier" AI agents, who will obviously be auctioning off placement and attention. Roughly speaking, FB agents are going to be pushing Ford trucks and Coca Cola, and GOOG agents are going to be pushing Toyotas and Trader Joes, or some similar package.
And on that auctioning - the duopoly not only eats everyone's attention, but eats even their *actual* customers, ie companies that buy ads, from which they have extracted nearly all the arbitrage of ads for less sophisticated players with clever auction schemas and optimization on their end.
Buying ads on FB or Goog as a non-F500 business is a furious Red Queen's Race that rarely nets positive, because you're against competitors in an ecosystem that's been aggressively tuned across multiple domains for years, so unless you want to become an expert at SEO *and* A/B testing on your own funnel *and* A/B testing niche-enough keywords to be net positive, *and* demographically modeling your own customers to align with the duopoly categories available, etc.
This also drives a Gresham's law type effect where actually good products are driven out by shoddy products that have big enough margins to accomodate the much larger ad spend needed to compete in that Red Queen's Race. And I don't expect THAT to change for the better in a "free AI agent" scenario either.
I think the very nature of AI agents are unlikely to have the same ad-auction economics as with the free internet. However, the nature of what this might end up becoming is of course highly unclear and *something* will become a chokehold is my bet, just as now.
> I think the very nature of AI agents are unlikely to have the same ad-auction economics as with the free internet.
Interesting, what's your model there?
My model is that they're going to be as big as phones in terms of engagement, and much more trusted at the individual level, so it's a more valuable insert, and to actually capture that value accurately requires auction schemas, which the duopoly is already best in the world at. So it's leaning into their strengths.
Interested to hear where yours differs, I could def be missing something in my mental schema.
An agent - both the conversational and autonomous - definitely will have much more influence but I’m not sure how much trust will survive monetization attempts.
Another consideration is that agents don’t have *all* the information in their models and so real time content access is a technical point of control by publishers/creators/etc. But those folks will transform from merchandisers that own the customer relationship to vendors that are once removed from the people.
Perhaps agents will be extensions of our attention, thus increasing the total stock of attention-tied-to-purchaing-power in the economy. Probably only for certain kinds of attention, though. "Watch the NBA finals for me" is probably not going to cut it for a fan.
I’m hoping agents allow us to delegate attention, at least the non-fulfilling kind that comes from scrolling feeds or reading a dozen sites to get a holistic picture of some topic.
Convincing story arc. Coming out of ads I realized the sheer power of the personalization of it all, and was looking for that first hop from digital to physical. I ended up buying a bunch of instacart as it seemed to me to play more "data broker that delivers stuff" than anything else. So far so good. But that's a far cry from unlocking robotics which would finally make a dent in larger industries like healthcare
Great article, and really looking forward to a follow-up on the strange ways the future might materialise.
I do think there is a limit to just how much attention people can dedicate to the digital world in its current form (as your time online chart shows). The future does point towards companies finding ways to "augment" physical experiences with the worlds people have built for themselves online in order to grow the total attention pie.
Meta and Google naturally seem quite well placed to design that augmentation, with their direct investments in AI & XR and assets in the current digital world, but it remains to be seen if building with the new paradigms in mind from the start (OpenAI, Anthropic) would instead be the key advantage.
Great article and I look forward to follow up. If I can recommend one extra piece of the puzzle to incorporate, have you looked at Henry Farrell and Abraham Newman's book Underground Empire?
I love the way you structured this as a detective story at the start.
I’m afraid I have nothing clever to add; I just enjoyed reading it, and wanted to tell you.
Well done, carry on.
♥️
I think you're exactly right but maybe I'm missing something in wondering if the distinction isn't online v. offline but social media vs. other online communities that could have flourished. In my own field, higher ed, I wrote some years ago that universities should not have ceded so much online real estate to social media (Facebook) at the cost of their own campus community (seminar discussions, late night academic debates, gatherings on the quad). Even now, Facebook and Twitter (and maybe now BlueSky) are the places so many academic debates take place when they used to take place on campus. If the university role in "raising" the level of conversation about serious issues is diluted, that's partly what happened. Is social media more democratic, with fewer gatekeepers? Sure. But who profited from this and who lost status. https://www.insidehighered.com/views/2021/02/16/colleges-should-build-their-own-social-media-platforms-instead-relying-facebook
I sold my attention and the ultimately ended up with this blog post from Rohit.
You’re welcome?
Attention => Facebook => VC $ => Substack => cool posts
We get lemons, we make lemonade, I talk about lemonade, I meet fellow lemonade vendors, life's good.
> And time-plateau doesn’t mean a revenue plateau. So maybe a large part of what we see as attention above gets redirected into agents acting on our behalf. Granted, they are likely to be less impacted by advertisements per se, but there will be new methods of economic rent-seeking.
AI agents are the obvious next step, agreed. And much like the current duopoly gets $200-$300 of value per US consumer annually (FAR more than 90%+ of people would be willing to pay for the FB or GOOG bundle of services) by offering "free" services and scale that can only come from "free" (roughly ~2B users each), we should expect similar dynamics in the AI assistants.
Some portion of people will pay for 'real' agents that are wholly aligned with their self interest, and which will compete to offer better outcomes...but the overwhelmingly vast majority are going to be using "free tier" AI agents, who will obviously be auctioning off placement and attention. Roughly speaking, FB agents are going to be pushing Ford trucks and Coca Cola, and GOOG agents are going to be pushing Toyotas and Trader Joes, or some similar package.
And on that auctioning - the duopoly not only eats everyone's attention, but eats even their *actual* customers, ie companies that buy ads, from which they have extracted nearly all the arbitrage of ads for less sophisticated players with clever auction schemas and optimization on their end.
Buying ads on FB or Goog as a non-F500 business is a furious Red Queen's Race that rarely nets positive, because you're against competitors in an ecosystem that's been aggressively tuned across multiple domains for years, so unless you want to become an expert at SEO *and* A/B testing on your own funnel *and* A/B testing niche-enough keywords to be net positive, *and* demographically modeling your own customers to align with the duopoly categories available, etc.
This also drives a Gresham's law type effect where actually good products are driven out by shoddy products that have big enough margins to accomodate the much larger ad spend needed to compete in that Red Queen's Race. And I don't expect THAT to change for the better in a "free AI agent" scenario either.
I think the very nature of AI agents are unlikely to have the same ad-auction economics as with the free internet. However, the nature of what this might end up becoming is of course highly unclear and *something* will become a chokehold is my bet, just as now.
> I think the very nature of AI agents are unlikely to have the same ad-auction economics as with the free internet.
Interesting, what's your model there?
My model is that they're going to be as big as phones in terms of engagement, and much more trusted at the individual level, so it's a more valuable insert, and to actually capture that value accurately requires auction schemas, which the duopoly is already best in the world at. So it's leaning into their strengths.
Interested to hear where yours differs, I could def be missing something in my mental schema.
An agent - both the conversational and autonomous - definitely will have much more influence but I’m not sure how much trust will survive monetization attempts.
Another consideration is that agents don’t have *all* the information in their models and so real time content access is a technical point of control by publishers/creators/etc. But those folks will transform from merchandisers that own the customer relationship to vendors that are once removed from the people.
Hence https://paywalls.net
https://open.substack.com/pub/paywalls/p/building-a-better-web
Perhaps agents will be extensions of our attention, thus increasing the total stock of attention-tied-to-purchaing-power in the economy. Probably only for certain kinds of attention, though. "Watch the NBA finals for me" is probably not going to cut it for a fan.
I’m hoping agents allow us to delegate attention, at least the non-fulfilling kind that comes from scrolling feeds or reading a dozen sites to get a holistic picture of some topic.
GAMMA > FAAMG
btw
Convincing story arc. Coming out of ads I realized the sheer power of the personalization of it all, and was looking for that first hop from digital to physical. I ended up buying a bunch of instacart as it seemed to me to play more "data broker that delivers stuff" than anything else. So far so good. But that's a far cry from unlocking robotics which would finally make a dent in larger industries like healthcare
The trend stabilizing at 6-7 hrs is interesting. What do you reckon is competing for attention now?
Yes I do
It is cliche but AR glasses could be one. just ambient computing 24x7
Great article!
Great article, and really looking forward to a follow-up on the strange ways the future might materialise.
I do think there is a limit to just how much attention people can dedicate to the digital world in its current form (as your time online chart shows). The future does point towards companies finding ways to "augment" physical experiences with the worlds people have built for themselves online in order to grow the total attention pie.
Meta and Google naturally seem quite well placed to design that augmentation, with their direct investments in AI & XR and assets in the current digital world, but it remains to be seen if building with the new paradigms in mind from the start (OpenAI, Anthropic) would instead be the key advantage.
Great article and I look forward to follow up. If I can recommend one extra piece of the puzzle to incorporate, have you looked at Henry Farrell and Abraham Newman's book Underground Empire?
Farrell's Substack: https://substack.com/@henryfarrell
The book: https://www.amazon.com/Underground-Empire-America-Weaponized-Economy-ebook/dp/B0BST43C5D
My own eyeballs were overwhelmingly on Twitter (2010-2022), and it weirdly (happily?) did not monetize!