What this chart doesn’t show is the speed of the shift. The 1985-2015 gap took 30 years. The 2015-2025 gap happened in 10. Concentration accelerates as winner-takes-most dynamics meet software’s near-zero marginal cost. The next 5 years chart is going to be harder to look at.
In the "media trust" section, how/where are Substack blogs (like this one) categorized? They're certainly not legacy media, but are they actually social media? To me, "social media" means sites like Reddit, X, Facebook, etc.
I would consider Substack blogs to be a separate category. Specialty/niche media, perhaps? I think it's relevant because the news sources I trust are virtually all hand-picked Substack commentaries and newsletters as well as a few independent ones not living on the Substack platform.
I don't consider legacy media or social media to be trustworthy sources, but I will put a lot of weight into what my curated "news and opinion" email feed shares with me.
“Software ate the world” always undersold what was actually happening. Software didn’t replace industries — it consolidated them into a few platform companies that own the customer relationship across categories. The chart isn’t really about software; it’s about distribution moats compounding. AI is about to repeat this one layer up — not eat the platforms, but consolidate the model layer the platforms now depend on.
Why not capitalize those countries at some multiple based on their growth rates? One could also view it as the rise in importance of intangible assets.
The Railroad GPT thread lands. From scaling engineering teams, I'd add: middle management isn't just routing information. It's context translation, coaching, and absorbing risk.
AI seems to collapse the routing layer cleanly but that judgment work doesn't vanish, it relocates. I think figuring out where it lands is the real org design question of the next decade.
What this chart doesn’t show is the speed of the shift. The 1985-2015 gap took 30 years. The 2015-2025 gap happened in 10. Concentration accelerates as winner-takes-most dynamics meet software’s near-zero marginal cost. The next 5 years chart is going to be harder to look at.
In the "media trust" section, how/where are Substack blogs (like this one) categorized? They're certainly not legacy media, but are they actually social media? To me, "social media" means sites like Reddit, X, Facebook, etc.
I would consider Substack blogs to be a separate category. Specialty/niche media, perhaps? I think it's relevant because the news sources I trust are virtually all hand-picked Substack commentaries and newsletters as well as a few independent ones not living on the Substack platform.
I don't consider legacy media or social media to be trustworthy sources, but I will put a lot of weight into what my curated "news and opinion" email feed shares with me.
Ate the world how?
Are you arguing that the valuations of tech companies are as accurate as GDP figures? No fucking way.
Also, comparing valuations to GDP is plain wrong
“Software ate the world” always undersold what was actually happening. Software didn’t replace industries — it consolidated them into a few platform companies that own the customer relationship across categories. The chart isn’t really about software; it’s about distribution moats compounding. AI is about to repeat this one layer up — not eat the platforms, but consolidate the model layer the platforms now depend on.
Why not capitalize those countries at some multiple based on their growth rates? One could also view it as the rise in importance of intangible assets.
The Railroad GPT thread lands. From scaling engineering teams, I'd add: middle management isn't just routing information. It's context translation, coaching, and absorbing risk.
AI seems to collapse the routing layer cleanly but that judgment work doesn't vanish, it relocates. I think figuring out where it lands is the real org design question of the next decade.