On Jul 5, 2022, at 10:50 PM, Bret Victor wrote:

The dot code can either be:

- aligned with the magnets (and presumably, crossovers), in which case we get a uniform grid of dots, but have to read odd rows backwards.

- or, centered within the block, in which case we have staggered dots.  And also have to read them backwards.

I guess the first option is the only one that makes sense, assuming there's enough space to squeeze in that last dot.

<EE30C6FC-C1FB-459A-9F21-59908D1C8C66-438-0001C0D4AF377C75.png>

On Jul 5, 2022, at 10:43 PM, Bret Victor wrote:

I think this magnet pattern works.  This is a single species of block that can be flipped around to mate with itself with the proper stagger, and can be shifted left and right by two turns.  This is a top view, showing the pattern:

<7B6F1B40-412B-46F7-BD27-2920C1699AE7-438-0001BFA5AFE5246F.png>

<63EA2870-2BA0-4777-8B4B-7ED8B3378**********BFA986712866.png>

The actual magnets should probably be placed on the side along the second helix from the bottom, so blocks of any height will snap together.  This is 12 magnets for a 6-turn block (one magnet per turn per side) which is a lot, but I think that's the minimum to get the snaps that we want.  

<32C8D4B5-267A-4C3E-8757-29DEF7906**********C02EBE6E3D78.png>

I think this works??