I don’t think the gates should animate up into the air. It breaks the visual logic of 2D for no benefit. It’s subconsciously confusing to see a gate I place in one cell move to occupy pixels in the cell “above” it.

I look forward to future days introducing new mechanics as well. Can I suggest a few, based on dynamics?

- Food! The horse moves on every turn towards an attractor. Have a hay bale / giant sugar cube in one corner fall off the back of a truck / helicopter :) Horses start out dumb and move directly towards the goal before backtracking. Smarter horses path find the shortest route to the goal.

- Goals! Now that the horse is moving, get the horse into a static horse box / cattle pen cell by strategically placing fences so that the path it takes towards the food involves walking onto the goal square.

- Floods! Water encroaches from the edges on a turn by turn basis. Not only do you have to contain the horse, you also have to hold back the flood.

My algorithm, by hand, was as such:

1. Start with the smallest possible valid solution (1)

2. Expand slowly, and each "step" (like, moving a wall or two around to "obvious" spaces) must be a valid solution (this brings you to 40-60 score, depending on your choices, on day 8). Continue to step 3 once you can't see anything obvious.

3. Look at possible places where you could expand, but need 1 more block. You'll find one eventually.

4. See if you can spare any walls anywhere, using diagonals for example. If so, place the solution from 3 and go to 3 (repeat). If not, go to 5.

5. Count or estimate the squares gained by doing your improvement from 3. See if you can reduce your score by less than that, pessimizing your solution, to gain 1 wall. Once you've found one, go to 3.

That got me to the optimal score within 15 mins or so.

Only nit: fix the walls. They take up one and a half spaces so are confusing, and they're sci-fi steel with flashing red lights. Turn them into one-square-only fences. You use fences to enclose horses, not raptor walls from Jurassic Park.

1. Do a screenshot of the grid (try to include walls as well)

2. Open https://enclosure-horse-solution.onrender.com/

3. Make sure the number of walls are correct in the input (bottom left)

4. Press "Solve"

PS: It might crash as it's on the free version of render. I've added a caching layer.

Here's the github so you can run it locally:

https://github.com/langarus/enclosure.horse-solution

clone it and run

make init // make web

https://enclose.horse Day 8 PERFECT! 100%

Right now, this is only used for troll levels, but I wonder if you could also use it for some actual puzzles.

Ah the famous spherical horse in vacuum

algorithm:

1. infer grid dimensions

2. color histogram analysis to designate grasses, water, cherries and horse

3. apply mixed-integer linear programming to determine optimized wall placements

4. profit!

try: https://dyigitpolat.github.io/enclose-horse-solver

source: https://github.com/dyigitpolat/enclose-horse-solver

The collected answered could probably be used to teach an AI to approach this type of problem thereby gaining some of the cognitive biases that humans have, which is exactly what you want in some cases: An AI that generates human like solutions to hard problems .

Doesn't feel outrageously difficult to put inside a webview?

Great stuff :)

https://enclose.horse Day 8 PERFECT! 100%

It would be interesting to be able to change the wall budget for each puzzle to add some variation (with a max limit).

Removing a block was a bit fiddly on FireFox (Floorp) due to the right click menu appearing when I tried to click on a tile.

Looking forward to tomorrows!

I didn't initially expect it would be a problem, but the constant squiggly movement gets very annoying.

I'd even go so far as to deny any submission with more than sqrt(size) walls.

[1] https://enclose.horse/play/44wCCO

I wonder how the wiggle animation is implemented in for the buttons and modal.