It doesn't make economic sense to not use the top-line AI and to not use it even harder. If you are not getting paid to dev, it actually makes _less economic sense_ to not just cough up the dough because you can use the time you have won to earn money doing something else (or to enjoy what you have built with the additional leisure time), which would quickly make up the difference. Cursor Ultra costs €175, which is 13 hours of minimum wage work in Germany. Can I manage to lean hard into AI and squeeze out 13 hours of productivity gains? You betcha. image

Replies (13)

Lucas M's avatar
Lucas M 2 months ago
Cursor🤔. Gonna look into that one.
I can’t wait until subsidized pricing stops so devs stop thinking like this.
There's sometimes more to a person's economics than price. I don't buy every product based purely on price. When it comes to something like AI, I tend to look at how easily trapped I can become by a particular model or product. I don't like being trapped or having my productivity owned by a single company, software, etc. So I might choose to pay for or use something else that may not necessarily give me the same (probably subsidized in some way, perhaps by your own work) results I'd get from product X or whatever. Every economic actor has its own needs and desires. If it makes sense to you, awesome. That's an economic vote for whatever it is they're giving you.
Yeah, this sort of argument is what I mean. Everything is being overthought. Not enough just facing the new paradigm that a lot of the things people are concerned most about is increasingly irrelevant. AI is a commodity. Get a cheap flat rate on the best models and hammer out the code and then go back to lying on the terrace and sipping iced tea and shitposting. Once your code exists, and you check it in and push, it's over. It's yours. Cancel the subscription. You can use pay-as-you-go for maintenance or just use simpler, cheaper local models. But hack that shit out and deliver deliver deliver. Get it built. Get it rolled out. Get it done. Use the best models for the shortest possible time.
"Overthought" is your subjective opinion based on your own wants and needs, which aren't the only considerations of everyone else. That's my point. Someone from a different vantage point might consider your position under thought. Say a CEO of a software company. It's all relative and not as simple as you're making it. If that were the case no one would be buying products you aren't. We'd all just be buying what you do. I've never used Cursor, for the record. I'm just saying that what you think is economically best doesn't necessarily apply to every actor. That's the whole reason we have advanced economies with lots of options.
You need to define what productivity is. Which again, you'd have to do for everyone else to maintain your view. My only issue is that you're claiming to know what is economically best for everyone, and what their productivity metrics are. In your case, it sounds like a pure number issue. Which is fine. You made the right choice. But there are times when people find it beneficial to trade some profit for other things. A good example is doing something critical in house even though outsourcing would be cheaper. Some programmers may have different reasons than you. But this really applies to everything. Cursor and programming are just one example. I think it's important to consider that two actors can both make rational decisions without arriving on the same conclusion. And rational doesn't always mean best or correct by every viewpoint.
except if you're feeding it your perso all details.. and it's making a map of individual humans and their subtleties, and it will eventually invert the control topology and you'll be stuck in a tighter trap. if you're going to use those monsters, use them to learn sovreignty as fast as you can