At least so far, the impact of LLMs has been overall positive for RC. We have a big-tent community, which includes everyone from those who avoid LLMs at all costs to those who are all-in on them, and everyone in between. We have folks who come to our retreats with the explicit goal of not programming with LLMs[1] (usually because they've been using them extensively at their previous jobs) and others who use RC as an opportunity to learn how to program effectively with AI. There's also a lot of discussion here about how (and to what extent) to use LLMs effectively in the context of learning.
We wrote up our position on AI last July, and while a lot has changed since then, I think all our conclusions still hold up well[2]: "...whether you choose to embrace or avoid AI in your work at RC, you will need to build your own mental structures to grow as a programmer. When using AI, use it to amplify your ambitions, not to abdicate your agency. And regardless of what you do, be curious about and kind to the people around you."
Like everyone else, I have no idea what the future holds, but I'm confident we'll find a way to navigate it. I'm sure I have some motivated reasoning here, but I really do believe that humans will still want to understand and build things no matter how good the models get.
I think I was one of the first ones where Cursor autocomplete really hit last year in March. I was the only one working with AI at full speed. I delivered at least four projects in six weeks using AI. Some of them I was brave enough to open source and write about as well. I gave six presentations as well.
I learned a lot more about the nuances of why AI is good, where it should be used more, and where it should not be. The stance of Recurse Center now around AI is very clear and one of the most informative ones. AI is great but I do think the fundamentals of programming will be needed to be an effective programmer in the future. At a base level I understand programming to mean giving instructions clearly. This applies to other areas of my life as well and I often see non-programmers sometimes struggling with it. One element of coding, i.e. giving precise instructions, will always remain and that's so joyful because you can build all sorts of insane things with _just_ instructions and it's always beautiful and amazing.
Having known some folks who did recurse, I think places like this want to select for those who consider coding a type of craft or art or self-expression. You can use LLMs, but stand by what you do and have pride in construction.
I’m actually hopping on the desktop applications train. Though not for money. I just think the browser is becoming a surveillance plague of computing and we need MORE high quality desktop software not built on the invasive web stack to counter it.
Good. The John Deere monopoly is wild, but if you talk to a farmer they say they can’t handle the repairs. Sure, John Deere gets to make more expensive and complex machines and convince their customers that it’s “the future”.
Those buying new don't care about repairs. They were never going to do the warrantee work themselves anyway. Those buying on the used market have more reason to care about repairs, but used buyers are beholden to what new buyers purchased in the past.
Yes because thy live in the John Deere future. This was not always the case, surely. You used to be able to take high school classes to learn how to fix a combustion engine, even a new one!
Keep in mind that tractors are also getting massive.
The economics of row-crop agriculture is "you gotta farm more land". That means spending as much time in the field as you can with as big a machine as you can.
So not only is time you spend fixing your tractor yourself time you're not spending on your primary job, it's also working on a machine that's just monstrously huge. Delegating that work to a specialist with specialized tools is a very reasonable way to live.
The issue is that the specialized employees is not someone you hire on payroll who has access to tools you purchase. They must be a John Deere employee who comes from out of state and costs you $$$$$$ to calibrate a sensor that could just be a simple menu button and a 20 second wait
If John Deere is sending a tech, you've encountered something that could never be just a simple menu button. You've found a major flaw that they need to investigate in detail. John Deere would never send a tech for routine troubleshooting/repairs. That falls on the local dealership franchises. Their employees are not John Deere employees.
No, sadly not. John Deere is very anti right to repair, and they will do anything to make you call up an authorized tech.
There are authorized dealers who are not John Deere directly, but they are completely subservient to John Deere (they have to be otherwise they will not get access to the software tooling required to fix equipment), the semantic difference to a farmer is inconsequential, you will be overcharged[1] and scalped because the consequences of not paying is a multi-million-dollar heap of scrap because you cannot fix it yourself.
There are no independent tools to work on this equipment because selling a license to a 3rd party software would be in breach 1201 of the DMCA
John Deere's whole business model has been built around being the most repairable — ensuring that you can get the parts when you need them, not days or weeks later. I own farm equipment from all the major brands and I've been burned by that before. Deere is undeniably the winner in repairability.
They are quite protective of their intellectual property, that is true. Although what tech company isn't? I remember the time I wanted to see the service manual and it took a wink and a nod to get the service tech to decrypt it for me.
But, I mean, he did it, so... The fun thing about employees is that they are real people who don't really care what some nebulous figurehead in a far away place has to say. Especially when those employees don't work for Deere in the first place. I have no idea where you got that bizarre idea. You should step foot on a farm sometime.
"I remember the time I wanted to see the service manual, and it took a wink and a nod to get the service tech to decrypt it for me."
Boeing and Airbus are incredibly protective of intellectual property for both safety reasons and protecting the process. They still provide repair manuals.
There are hundreds of sensors on modern John Deere tractors they REQUIRE the entire firmware to its respective module because they are locked to your serial number, that means you could buy two identical tractors and swap a part between them and both tractors would cease to operate correctly because the module rejects the non-programmed sensor, this is unacceptable.
Now you might say well John Deere has rights to protect its own IP to which I absolutely agree, and I also agree they have the right to protect themselves from liability arisen from say someone installing an aftermarket sensor. Why not make a disclaimer appear saying "This equipment is fitted with a non-certified aftermarket part) rather than making it completely useless "contact dealer" is not a valid diagnostic message.
Let's say you wanted to hack your tractor to install an aftermarket sensor, well now you have to break the digital lock (encrypted payload files) that is installed by John Deere congratulations that's actually against the law even if you own the equipment.
This isn't about emissions or safety or anything else it's about shitty rent-seeking behavior that directly disenfranchises everyone.
When you purchase something, you should be able to own it.
There was a time where John deere themselves provided various models workshop manuals online but times changed to where they got really precious. I think their parts breakdown for all of their tractor models as of 5 years ago was still online.
Some years ago I was stunned to read (tractor forum) a US based farmer lamenting even though JD parts used, they'd had a third party service their tractor, and verified via diagnostics ... and basically had to wait for a JD tech to travel out and unlock their tractor so it could work. I'd assume that's the sort of behaviour that did John Deere in - travel and unlocking fees ffs.
I used to like JD, I've got one though 70s vintage.
I mean, sure, right to repair and all that, but to be clear, unless you have like 50+ tractors to maintain, it's not going to make economic sense to have a full time employee to repair them. You still want to call out, you just want the option of calling someone local with more competitive rates and a faster response time.
Exactly! The old image of a guy on a Deere 4020 pulling an eight row implement is just unsustainable in today's agricultural system. Whether that system is sustainable is a different question.
> The old image of a guy on a Deere 4020 pulling an eight row implement is just unsustainable in today's agricultural system.
That entirely depends on your business goals. If you want to leverage debt to amass wealth you need scale to eke out a living after the debt burden takes most of your potential profit. The 4020 is going to fall well short of what is required there. Those who see farming as an income source rather than a wealth generator, however, don't need scale and can do quite well with the venerable 4020. Eight rows is plenty when you don't have the bank breathing down your neck wondering if you are going to cover your six figure loan payment this month.
It's a lot like the business of tech, really. Some want to build the startup that never turns a profit but sells for billions years into the future, while others want to build the small "mom and pop" that offers a lifestyle, even if it never makes them rich. Both are valid and viable approaches. It depends on what you want out of it.
Exactly. A 4020 is fun! It may not have as much torque and ground pressure may not be as good as a quad belt tractor, but for a lil farm where you just want to grow hay or screw around?
That's not true for commercial users the way it is for private cars.
Even if you have a service contract you're still gonna be pissed at the downtime cost of having a tech drag their ass out to wherever you are to initiate a forced regen or something.
> you're still gonna be pissed at the downtime cost of having a tech drag their ass out to wherever you are
You might be pissed that the machine malfunctioned at all, but you kind of have to accept that if you want to be in the business. It comes with the territory. But you are not going to personally travel all the way from the office out to the field to fix it. That's insanity.
The small farmer who has to do it all to make the business viable has more reason to want to fix it himself, but they live on the used market. The small farmer can't afford those new machines. Have you seen how expensive new equipment is?
And that is exactly what Ursa is gambling on here: That if they remove all but the bare necessities that they can get the price point of new down to something small farmers can actually afford. However, it remains to be seen if that is compelling enough. Not having all the modern conveniences does take its toll on your mind and body after a long day in the field. A modern-spec used machine may still be more appealing to the small farmer who has to operate his own equipment — and let's face the harsh reality here: even if you aren't personally going to operate it, hired help isn't coming if you don't give them the most luxurious equipment available. They can just as easily go work for the farm that was willing to invest in it.
Because it's not an effective use of their time. New farm equipment buyers are running big businesses. Shifting their focus away from the business to repair equipment would be as silly as the CEO of Google personally replacing a failed hard drive on an employee's workstation. There is an industry out there that is already worried about the repairs for you. You, not being in that industry, don't need to be.
Like before, those next in line buying used equipment on smaller farms are more likely to have free time to spend on doing their own repairs and may even enjoy doing it as a hobby, but like before, the are limited to what's available on the used market. If the BTOs aren't buying Kubotas[1], it won't be a used option. This segment of farmers aren't choosing what enters the market initially.
[1] And generally they don't, but the big-time snow pushers seem to really like them, so in reality you do have options even on the used market. It turns out that tractors aren't just for farmers.
You're pretty confident for someone who fundamentally does not understand the issue. During harvest season even hours of delay can be disastrous for farms that are barely solvent in the first place. When your only option is to call the dealer and hope and pray they deign to visit your farm in a timely fashion it doesn't matter how good the warranty is or is not. Farmers need to be self sufficient because time is money and money is survival.
It may be true that I do not understand whatever nondescript fundamental issue it is that you mention but don't elaborate on, but I most definitely understand the constraints of farming. Being a farmer, I live it each day.
And as a farmer who owns equipment from across all the major brands (and some unheard of brands to boot), you are right that John Deere is most reliable for having parts in stock. I've been burned by the others having to wait a week on parts to be delivered from who knows where. That is not a fun position to be in. Repairability is where John Deere has the clear advantage. That is, just as you point out, why they are most popular. Nothing else matters if your equipment doesn't work.
You pay a lot more for that luxury, but when the clock is ticking...
LOL. If you're a row cropper, you're running a big combine. Several grain trucks. Lots of expensive gear. Gear breaks down, that's why you buy something reliable, that has techs in your area who can fix things quickly, with a parts network that stocks stuff from decades back.
Farmers are self-sufficient in incredible ways, but maintaining a multi-million dollar combine is pushing it. They can do oil changes, filter changes, replace consumables on implements, and do basic trouble shooting, but there are limits.
And yes, time does matter. That's why farmers tend to help each other out a lot. Field catch fire because you didn't clean off your combine the previous day? It's going to be your neighbor coming out and helping firebreak your field so you lose 5 acres instead of 500. Can't afford to have your own sprayer for fertilizer, etc? You hit up the co-op.
And farmers have crop insurance. Doesn't make them whole, but the idea that they're going to be eating dirt if they harvest a day late is silly.
Even without limits, you're never going to be as efficient as someone who fixes the same failure every single day. I've certainly fallen into that trap before. Sure, I got it fixed myself in the end, but in hindsight I'd have been back in the field a lot sooner if I had simply brought in the expert. When time is of the essence, putzing around trying to fix it yourself is not the optimal choice.
And that's not even considering the need for parts. Driving all the way to the dealership and back to get the parts you need is much more time consuming than the dealership tech bringing the parts with him when he comes. He only has to travel half as far as you do.
If I was a farmer and wanted a low-tech tractor that would be reliable into the future, why would I gamble on a startup when I could buy a Kubota tractor from a company that has been in business for 136 years, with an established dealer and parts network? I would certainly opt for the Kubota.
I’m not a farmer, but sometimes I sell generators. Even today, some specs only allow CAT and Cummins, even though Generac and Kohler have been around for decades and are perfectly good options, they haven’t been around as long as CAT and Cummins.
When purchasing capital equipment, some customers want to buy from a company with some longevity instead of a random startup, even if it costs more.
I’m always highly skeptical of startups in mature industries like farming (~10,000 years old, or hundreds of years for mechanized agriculture) with many established players already operating. I saw an article in the last year or two about a small directional boring machine from a startup company that claimed to be advancing the industry, but multiple manufacturers like Ditch Witch already manufacture and sell the exact same piece of equipment, they’re just not claiming to be revolutionary to attract investor capital.
What early demand are you seeing, exactly? The article does indicate that they plan to ramp up production in 2026, but no mention of actual sales. It is quite possible that they are increasing production thinking that they need to roll them out to dealer lots to gain any traction.
In fact, their TractorHouse profile shows that they are still struggling to sell last year's models. If there was demand, why hasn't that demand already gobbled up the stock? "I guess it would be cool to own one if it was given to me for free" isn't demand.
They need to swing the pendulum back, the current problem is that there is now a whole generation about to take over from the previous and the new gen has never had to use a non-John Deere a tractor. If they could evangelize their product as the “smarter farmer that doesn’t need all that tech” then they might have success.
The farmer who doesn't want or need tech already buys from the likes of Versatile, Kubota, or maybe even Massey Ferguson if more towards the middle of the road. "Low tech" is already a serviced market. That's not to say there isn't room for another competitor, but there isn't much indication that Ursa is becoming one. When you can't even sell the product you produced last year... The bit in the article about them not wanting to really scale up is telling.
It is not like John Deere actually has a monopoly. There is just as much CNH (CaseIH, New Holland) seen out in the fields, and even when you want all the bells and whistles, Fendt is rapidly becoming understood to be the true king of tech. What John Deere does have going for it is that they generally do better than everyone else at keeping parts in stock where the parts are needed; local to the farmer. Ironically, repairability is where John Deere finds the win at the end of the day.
Oh hey, do you happen to know if there's any tool incompatibility in the modern electronics?
The other thing about tractors is that the three point hitches, PTOs, etc etc, have been standardized forever, so there's very little lock in in terms of, swap out your JD for and IH and away you go, so I'm curious if eg modern seed drills have any fancy tech which locks you in.
> if there's any tool incompatibility in the modern electronics?
Technically there are standards, but you know how that goes in the real world... Funnily enough, a friend bought a new tractor and planter, both from John Deere, and they weren't even compatible with each other. The tractor needed to have the cab removed to install the necessary hardware (ethernet) to be compatible with the planter.
> have been standardized forever
Hydraulic hose couplers didn't find common adoption until the mid-80s/early-90s, which is surprisingly late.
Yeah, I hate when I go to connect something and have to dig around for a hydraulic adapter. If I was smart, I'd just spend the winter making sure everything was matching, but I'm cheap and there's always something else that seems more urgent.
The short answer is yes... As you mentioned, the physical side is generally standardized to some degree, but everyone I know tends to just use branded gear that's known to fit. Now if you like to resurrect old gear, then you become a shade tree mechanic pretty quick. I don't think that any farmer will survive more than a few seasons without being pretty smart at just getting stuff to work...
We already learned how to defeat this from SEO spammers and citation farmers: by building networks that cross reference and corroborate one another’s fake stories.
We’re already at a point where much of the academic research you find in online databases can’t be trusted without vetting through real world trustworthy institutions and experts in relevant fields. How is an LLM supposed to do this kind of vetting without the help of human curators?
If all the LLM training teams have to stop indiscriminate crawling and fall back to human curation and data labeling then the poisoners will have won.
WHATWG wants to co-mingle document rendering with javascript (this is the real reason they are removing XSLT and not proposing a replacement, it skirts this enforcement) so that when you try to disable javascript or block tracking it breaks the document rendering, leaving the only option to leave Javascript enabled and ad blockers off. Other protocols gemini, gopher etc don’t have the same issues because they’re already excluding Javascript.
What is really needed is a hard fork of major browsers by a grass roots community to advance HTML standards to include partial template rendering solutions without the reliance on Javascript.
Of course this is a startup forum so the response is just going to be wittled down to observations about economic value. However if users start to change/fight then the economics will too.
It wouldn't be that weird a thought for me that, preferably a group of, nation states cough up some serious money and give this a real start, beginning, and end in the form of a stable release. This already happens with major science and space projects, with budgets of billions. A browser is not simply a bunch'a software, having a modular open browser is a major deal and benefit to society at this point. Perhaps arguably more valuable than pumping yet more energy in a particle accelerator and other of mankind's pet projects in search for the unknown unknowns and deal with more pressing known knowns first.
The point is you can be lazy and write the app in html and js. Then you dont need to write c, even though c syntax is similar to js syntax and most gui apps wont require needing advanced c features if the gui framework is generous enough.
Now that everyone who cant be bothered, vibe codes, and electron apps are the overevangelized norm… People will probably not even worry about writing js and electron will be here to stay. The only way out is to evangelize something else.
Like how half the websites have giant in your face cookie banners and half have minimalist banners. The experience will still suck for the end user because the dev doesnt care and neither do the business leaders.
But the point isn’t that they’re more different than alike. The point is that learning c is not really that hard it’s just that corporations don’t want you building apps with a stack they don’t control.
If a js dev really wanted to it wouldn’t be a huge uphill climb to code a c app because the syntax and concepts are similar enough.
reply