This summer I helped for a few hours to build Beaver Dam Analogues (BDAs) at Philmont Scout Ranch in New Mexico. I am really looking forward seeing the positive ecological impact when my future grand children trek Philmont. Building BDAs is good fun. You should try it.
Interesting, tell us more! Do the beavers use/upkeep them, or are they purely to create pools where there are no beavers to do so? If the former, is it needed because there aren't enough beavers or not enough wood/vegetation for them to build?
Do the beavers "adopt" the structures? Or are they too dissimilar to their constructions? I imagine the main goal is actually less about the beavers than it is restoring habitat that beavers otherwise create. Do you ever build them in places where there never were beavers?
I have this funny picture of beavers refusing to enjoy the dams because they didn't build them. No "pride of ownership".
Beavers were hunted so intensely that they completely disappeared decades ago in the area. Without the beavers building dams and thus slowing down the flow of creeks, more and more erosion took place and area that used to be wetlands dried out. With the gradual drying the willow tree disappeared, which is one of the major food sources for beavers. So while beavers are starting to repopulate, they don't move in where there is no food available.
So Philmont is building BDAs in order to slow down the creeks, providing suitable habitat for willows - and once this food source is once again established, the beavers should return and take over maintenance of the BDAs.
Honestly might give that a go, yeah. Brand new, low stakes, throwaway projects are one of the few things these tools are actually genuinely pretty useful for.
Bad data on graphs, demos that would have been impressive a year ago, vibe coding the easiest requests (financial dashboard), running out of talking points while cursor is looping on a bug, marginal benchmark improvements. At least the models are kind of cheaper to run.
Feedback:
- At end of game show statistics correct/incorrect
- At the end of game show which particular color weaknesses were identified
- Show progress meter (1 of 20)
- I have a red/green weakness and I expected to run into issues. One of the greens was hard to differentiate but I still got it right. I expected it to be harder. Perhaps look up exact pallets for different color perception issues and use those.
LLMs are very good at knowledge extraction and following instructions, especially if you provide examples. What you see in the prompt you linked is an example of in-context learning, in particular a method called Few-Shot prompting [1]. You provide the model some specific examples of input and desired output, and it will follow the example as best it can. Which, with the latest frontier models, is pretty darn well.
