And in many cases, it is sublinear. A semi-truck costs much less per ton than a van. Fright trains are even cheaper. And container ships, even less so. In fact, I am convinced that it is sublinear for tractors too, but the benefits go to John Deere rather than to the customer: bigger tractors offer more value to the customers who have a use for them, so they can increase their prices accordingly.
Working in tech, you should know that the list price of an item is a poor proxy for how much work and how many raw materials went into it. When you don't have very many brand choices, market dynamics matter quite a bit. I think it's noteworthy that those same tractors increase in weight superlinearly with horsepower.
Let's assume price is a good enough proxy metric though:
> HP should be a decent approximation for the amount of work done.
The problem is that the amount of work done, measured in energy at the tractor, isn't a good approximation of the amount of work done, measured in useful effects on the farmland (stated differently, wider implements are less efficient). If you look at what width of tiller or other implement you can pull behind those tractors, it scales poorly with horsepower, and when you read through case studies of people who tried to get away with smaller tractors and wider implements you find that they had a lot more human toil and more passes over the farmland to make them work.
A more useful proxy metric would be how many dollars of tractor a single person needs to manage X acres of farmland (for any fixed task) in Y hours. That's a little hard to gather data on because you have qualitative shifts in who's farming for which reasons as you increase acreage, but if you look at the actual tractors' capabilities (e.g., appropriate width of implements) you can compute it as a derived metric.
> comparison to trains, semis, ...
The comparison isn't great. The whole point of a tractor is to manipulate its environment, immediately spending energy to break physical bonds in plants, accelerating and decelerating a rock from point A to a nearby point B, .... Mechanisms of transport get around that because the only "mandatory" costs are the delta-v: accelerating and decelerating the load. If you're transporting a long distance, that's a negligible fraction of the costs, and all you're doing with scale is reducing the drag along the way. Since drag is nearly proportional to frontal surface area in common speed/weight regimes, you get the observed effect that a longer vehicle (a van or a semi) is more efficient per unit of load than a shorter vehicle (with extra benefits because the shape itself has less drag per unit of frontal surface area). As you scale to something like a train, rolling resistance matters a lot more. The extra efficiencies there are basically because your steel wheels don't deform as much as rubber wheels, and more of that deformation is reclaimable (with many other benefits, e.g., you're able to waste less weight on the engine relative to the load).
The key feature making that possible though is transport over long distances, amortizing the cost of startup and shutdown, with a secondary important feature being low rolling resistance pathways (roads, bridges, rails, cargo ships with the barnacles cleaned off, ...). Farms have neither of those. Loose dirt is piss-poor rolling resistance characteristics, and the bulk of the work you're doing is severing physical bonds to move things a few inches or feet. The "work" being done is the whole point of the activity, rather than a byproduct you're able to minimize.
If you look at this https://www.farmtender.com.au/articles/john-deere-tractors-p... you will see that the price of a new tractor increases more or less linearly per HP. HP should be a decent approximation for the amount of work done.
And in many cases, it is sublinear. A semi-truck costs much less per ton than a van. Fright trains are even cheaper. And container ships, even less so. In fact, I am convinced that it is sublinear for tractors too, but the benefits go to John Deere rather than to the customer: bigger tractors offer more value to the customers who have a use for them, so they can increase their prices accordingly.