The terminology can be confusing because in corporate speak being uncompromising can mean being opinionated. Not "we made no tradeoffs" but "we didn't split the difference."
The article means it as tradeoff rather than concession, so the argument is "to avoid design by committee you have to compromise on purpose."
This is the taste question applied to Mathematics in a similar way it's been applied to code. In an era of AI abundance, the question becomes what the goals are and what gets verified and digested (adopted by users). Goodhart’s law: the goal of producing code is not just about maximizing the number of tokens used, but the productivity gains and economic surplus.
This could serve as the template for any field in the age of AI:
"We are not trying to meet some abstract production quota. The measure of our success is whether what we do enables people to
understand and think more clearly and effectively about math (or products, or science, or hardware...)"
I find AI a joy to use to build software. I was never the type of engineer who loved the code itself. I was always the type who wanted to see how users would react, how metrics change based on the feature, etc. The code was the most boring part for me generally.
This is the template any field should have in the face of AI: screw AI, forget it. If you're using it, it's a disrespect to the field. Let's continue without it. I think that's the only sane approach.
William Thurston's argument (from the quote) was that the point of math isn't to produce more math but to better understand the nature of reality, just like the point of code is not to generate more code but to increase user value.
AI is a tool that reshapes knowledge work but the ultimate goal remains about value.
A good example of "founder bias" where big companies are read as not innovating, when in fact their goal is to squeeze as much juice from their user base and strengthen their monopolistic position and pricing power. From the outside it looks like blindness and atrophy but from the inside it's the main bread and butter.
The sighted engineer is also cave adapted, just to a different cave.
The problem is when practices are chosen, that improve things on one dimension, and harm things on another, without recognition of the long-term tradeoff being made.
Simply recognizing the tradeoff exists, is likely to result in wiser decisions with lower tradeoff costs, and greater opportunity expansion.
who is to recognize the tradeoff? communicate it to who? A giant business is built on stability. i dont think its a bad thing that they do not pivot except in circumstances where they have no other choice. The business is a brand and that brand is a certain way of doing things and if/when that way of doing things falls out of favor then the business should fail. the owners made enough to retire by then, as well as many additional people, and its a reasonable way for people to become unemployed which doesnt hurt their chances at new jobs. A new company sprouts up, or 5 new ones, and they can all be lean and quick to pivot and try to figure out the whole new way to do the same kind of business. All the experts can be hired and the supply chain keeps chugging along
this is also a wanring to young people joinging old companies in weak job positions over taking a chance on a new company trying to be better. dont just assume the old company is going to be there forever.
especially now in defense. war has changed. companies like Anduril are on the rise while classics like Boeing, Lockheed, Raytheon, etc are assessing how much they can adapt. its not extreme like theyll go out of business but the gov may order way less jets and such than in the past, and put a lot more money into drones. they will still buy jets though, and the classics may need to adapt to a lower funding pool, lower orders, and becoming a low volume legacy product hovering around certain use-cases, while new companies get all the runway to takeoff on the big investments.
this kind of thing plays out at all different scales and to different degrees everywhere in the market.
We are talking about the big cultural/organizational decisions that allocate where creative freedom lives or the lack of it.
To not recognize the decisions, weigh their tradeoffs and their implications, is to make the decisions blindly.
Creativity has to be actively protected where it is wanted or impactful. Or in large companies, it will get crushed as power to make choices flows elsewhere.
Reminds me of a blogpost about discussions within facebook about supreme importance given to engagement metrics. Then they were replaced with holistic engagement metrics.
The article frames the cost of living issue mainly in terms of demand (...because of OpenAI and Anthropic...), which is an important factor, but has very little to say about supply. San Francisco is a major metropolis that looks like a mining town: there's very few high-rise buildings compared to other major cities. It's due to many factors including strict zoning, growth caps, seismic risks (compared to say Tokyo?) and landlords that don't have much incentive to decrease the value of their skyrocketing assets.
Also some recent setbacks like our leaning Millennium Tower.
It might take a political earthquake to change the status quo given how ossified everything is unfortunately.
- OpenAI wants to be the consumer version of AI, modeled after Google and Meta, with a mostly free universal service powered by ads and e-commerce. They haven't fully shown that model can work. The big problem is the lack of zero marginal costs as each new user requires GPU spend.
- Anthropic positions itself more as enterprise AI, modeled after Microsoft ironically enough, and charges big companies for services. The economics of coding agents work but GPUs get expensive fast and open models are getting good enough for most use cases.
So it's a race between ads and e-commerce offsetting AI spend and open source eating almost everyone's lunch.
Even open source models need hardware and energy to inference. Therefore, anyone offering a free ChatGPT competitor will be using the same unit economics.
My bet is that OpenAI will make free ChatGPT work through ads.
It's a very open question whether the economics for ads work though. Ads aren't just an infinite money pit you can just reach into, they have a price ceiling you have to stay under before they stop making sense for the advertiser. And if some of the rumours about OpenAI's ad CPMs are true, the inventory is going to be massively expensive.
It makes sense for it to be expensive, mind. The unit cost of serving an LLM response is so much higher than the unit cost of serving a bunch of Instagram posts, or traditional search results or whatever.
But price too highly, most advertisers won't be able to justify them. Why spend $60 for what you can buy from Meta for $6? It's a brave media buyer who runs with that as a long term strategy. But if you don't price highly, you're just offsetting some of the losses. The whole reason Google, Meta et all's ad networks exploded what because they cost less to get reach than traditional media did before, which opened them up to a bazillion small businesses who otherwise didn't have the capital to get off the ground through traditional media. ChatGPT's will cost more than what's available now. Massively more. There's not a lot of history of that working out!
They'll get some buyers for a little while, the $60 vs $6 equation balances out if the ads are 10x more effective, and companies will throw a bit of money into campaigns to get a feel for how well they perform.
(Google is in a different position, they make basically near infinite margin on other ad types and can lump budgets in together to get still-attractive blended CPMs. It's a hit, but it's worth taking to protect their wider network, just like they did for years when Youtube wasn't remotely profitable.)
Interesting they agreed to sell after their rebrand to Fin a month ago.
There's increasing competition in the customer support AI agent space: Sierra valued at $15.8 billion, Decagon at $4.5 billion. It looks like Salesforce CEO Marc Benioff is trying to compete directly with Sierra, which was started by his ex-Co-CEO Bret Taylor. Also about preventing independent AI support agents from becoming a control point outside the CRM.
I've been using Claude Code regularly since the 4.5 release, and 4.7 was a significant regression: very unreliable, arguing about changes, deciding that fixes weren't needed, etc.
I'm hoping they recreate the magic of 4.5 but it's as much about the quality of harness, the memory and efficiency of the tools than simply the models at this point.
Dropbox failed to find a second act: they struggled to find PMF with their acquisitions and new products: Dropbox Passwords, Dropbox Paper, Carousel etc.
As Steve Jobs warned Drew Houston, Dropbox was "a feature, not a product"
It's a great product. They had the brand, the capital, and the user base to become what Slack, Zoom, or Notion became. Instead, they spent a decade fighting a losing battle over storage pricing with Google and Microsoft. Their lack of a second act is due to a failure of product vision and enterprise execution.
And they had Paper, which was an excellent product (I was at Dropbox a decade ago; we all used Paper constantly and it was great) very close to what Notion later became. They never got it over the hump to wider PMF — like you say, a failure of product and of enterprise execution.
(Given that it was so close to Notion, I think Paper is one area where the product vision was on to something good; but they didn't succeed at product execution, connecting customer feedback to iterating correctly on product improvements.)
Never worked at Dropbox, but I absolutely loved Paper.
The problem at Dropbox seems to have been that there was no cohesiveness to all the products. Paper, Passwords, Sign, all seem to have never been truly integrated into a single experience. Each one felt like it was trying to have its own identity.
Yeah, when signing into Paper it always felt pretty silly how the auth flow was all like "are you sure you want to share your Dropbox account info with this Paper thing?" as if it was some third-party service.
Ironically, just within the last year Paper has gotten much more integrated into Dropbox as a single UX. And… it's significantly worse: slower, clumsier, harder to navigate. (I don't think there's any inherent reason those had to be correlated; it's just that Paper has clearly been destaffed a lot in recent years, so naturally any new changes will tend to be less polished.)
Another issue: Paper was tied to your Dropbox account.
From Dropbox's perspective, this sounds great. Accounts become more useful and valuable. The addressable market of a Dropbox account grows! Plus, everyone has a Dropbox account already, right?
Unfortunately, it turns out that business customers generally don't deploy Dropbox wall-to-wall. It's expensive. Not all employees need file sync.
A Dropbox account ends up being an obstacle to adoption.
And a distraction: a common account creates an irresistible urge to spend a lot of time finding ways to tie this new product into the old one.
I'm not sure I ultimately buy that Dropbox is expensive - for one thing, Notion now charges basically the same as Dropbox, and you don't even get file sync in the deal.
I can definitely believe that that was an objection customers had, though. I just suspect that what it really meant was another way of expressing that there wasn't PMF - if the PMF were there, they'd have willingly paid, just like they do now for Notion or Slab or what have you.
And yeah, there was definitely some energy going into trying to tie the products together more (putting Paper docs in your Dropbox folders) - and when that finally shipped it sadly made the Paper experience worse, not better.
I don't see the need to become bloated like slack and a one size fits all application. They do a great job with the product they have. Is there anything wrong with just being what you are? Why does the lack of a second act need to be a bad thing if your first product is great and still extremely valuable?
I would also if anything put Zoom in with Dropbox, they have a product that is by far the most enjoyable to use in that space, but any other offshoot is not worth it.
Teams being free doesn't make teams good. Just like Google Drive and OneDrive being packaged in for cheap doesn't make them good. It is clear why they are lower value.
> They had the brand, the capital, and the user base to become what Slack, Zoom, or Notion became. Instead, they spent a decade fighting a losing battle over storage pricing with Google and Microsoft
Is the alternative not likely that they would have spent a decade fighting a losing battle over office software with Google and Microsoft? Paper was a great product but the big guys have vertical integration so companies prefer their end-to-end solutions (GSuite etc) and I don't see how Dropbox could have easily overcome that.
> Is the alternative not likely that they would have spent a decade fighting a losing battle over office software with Google and Microsoft? Paper was a great product but the big guys have vertical integration so companies prefer their end-to-end solutions (GSuite etc) and I don't see how Dropbox could have easily overcome that.
Slack, Zoom, and Notion all argue against that. Yes, they have to compete against Google and Microsoft's integrated solutions, but they're good enough that they have held their own. Of course they would be bigger if Google and Microsoft didn't have such products.
Every generation of builders believed their tools defined their value. Then the tools got easier, faster, automated, and the definition had to change.
But programming didn’t disappear. Writing didn’t disappear. Designing didn’t disappear.
AI flips the equation: when creation becomes cheap, value shifts from how much you can produce to what changes because you showed up. The ability to have a positive impact has actually expanded.
I don’t think this whole thing had anything to do with AI or not. It has to do with ‘teaching kids in a gym’ or ‘sitting in front of a screen in an office’.
“Every generation of builders believed their tools defined their value.”
People anchor identity to the hardest part of their work.
• Assembly → craftsmanship
• Hand coding → engineering skill
• Complex stacks → seniority
• Writing longform → intellectual authority
The difficulty of the tool becomes proof of worth.
“If few people can do this, then my contribution matters.”
So value feels intrinsic to the technical act itself.
⸻
“Then the tools got easier, faster, automated, and the definition had to change.”
Historically, this always happens.
Compilers replaced assembly expertise.
Frameworks replaced boilerplate knowledge.
Cloud replaced infrastructure mastery.
AI replaces a lot of implementation effort.
Each time, people initially interpret it as:
the skill is dying
But what actually dies is the old measurement of importance.
Value moves up a level:
from execution → judgment → direction → taste → responsibility
The activities persist, what changes is why they matter.
You still code, but code is no longer scarce.
You still write, but writing is no longer the bottleneck.
You still design, but layout isn’t the achievement.
The work shifts from producing artifacts to choosing which artifacts deserve to exist.
⸻
“AI flips the equation: when creation becomes cheap, value shifts from how much you can produce to what changes because you showed up.”
This is the core claim.
Old model:
effort → output → value
New model:
judgment → outcome → value
Previously you proved worth by volume, speed, or complexity.
Now production is abundant, so the scarce thing is:
causal impact
Not:
Did you make something?
But:
Did reality change because you were involved?
You’re moving from manufacturing to intervention.
⸻
“The ability to have a positive impact has actually expanded.”
So the post ends optimistic.
AI doesn’t reduce agency, it removes the cost barrier to acting.
Before:
You needed a team, funding, or org permission to affect the world.
Now:
A single person can teach, fix, organize, build tools, or help communities directly.
Meaning shifts from scale (reach) to consequence (effect).
⸻
In one sentence
The post argues that AI doesn’t eliminate human contribution, it removes technical scarcity, forcing value to relocate from producing things to changing outcomes.
> Mr. Petersson and Mr. Backlund said they thought Luna would eventually get smarter and more business-savvy
> having an A.I. boss can be like working for a teddy bear with amnesia.
It makes you question their assumptions when setting this up. Makes sense as PR but it ends up shaping how many readers see AI.