I admit to being rude and to being a very bad builder myself.
But the article is bad. Just about every piece of evidence has some sort of issue. Correlation-causation, or not enough data, or just assumptions. The wolf stuff (seems to be) based on only 2 observations. The author cites big tech in CA, but then describes a single clothing company that has a high random metric that is supposedly an accurate indicator of all of the above.
And the connections between arguments are not even that good. Skimming over it, I wasn't sure what the article's point was.
As for the conclusion, the vague words on what could be done are the kind of stuff everyone is trying anyway for other reasons, and it isn't (?) working.
Pro (Newbie) tip: use a GUI for git (like lazygit or VS Code Source Control). Git is one of the few complicated CLI tools that actually works well as a GUI, so take full advantage of it.
I use navi cheatsheets for that. You can store commands with descriptions and easily execute them. This is where I put all the "will need only once a year" commands.
What I do google is ffmpeg, but not before every online editor/converter on the internet completely betrays my expectations and makes me reconsider my life choices.
I suspect the software is doing a lot of the heavy lifting, because the mechanism is (looks) quite simple.
Here are some flaws/potentially misleading features I noticed:
By design, the robot has only 2 I/O functions (3 in 3D), which are the motors on the strings. Thus it can't be any more capable than any other machine with 2 I/O.
In fact, the design is simply a function that maps limited 2d movement and grabbing onto 2 motors.
Pulling the strings affects it's tip first, so getting it into a specific position is a pain. Making an "S" shape, would require first rolling it all the way one way, and then unrolling it the other way.
In the video, they have the robot already setup in a specific position, and they don't show how hard it is to actually get it there. All the complicated moves (around rocks+drop off) are very specifically set up, and might be non-reversible.
The "contact detection" requires one of the strings to be fully contracted. And all of it relies on the manufacturing and environment to be consistent with minimal friction.
This robot is a bit like brain*uck. You can do stuff, but you need to do it in a bit of a roundabout way. Not to say that there isn't a sense of elegance to it.
e. Raycast doesn't know what the e constant (2.71828) is.
Their entire calculator is crappy. It took them forever just to get scientific notation and inverse trig functions working.
Meanwhile, it was all proprietary (so I couldn't do anything myself) and they were busy developing AI. I know they have all the fancy add-ons and scripts, but at that point it is easier to just cmd+tab to the browser for Wolfram Alpha or commandline for numbat calculator.
I just realized Albert (sort of) work on MacOS, but it is in German for some reason, so I will have figure out configs.
Why does this feel more like an advertisement for Webrecorder[0] than an actual guide?
Why add that privacy and big tech section at the end?
Thanks for the list of features and buzzwords, but I couldn't figure out what all this tooling exactly does. Also, none of the links work. And Webrecorder doesn't have a Firefox extension, so don't even bother listing any of the "privacy" stuff.
1. Make L texture look better by by shifting the arrow origin up and left. Like the turn traffic sign.
2. I am deleting my robots now, when I am trying to remove an arrow from underneath them.
3. If itch and Unity's APIs allow, make full screen, "ask before exiting", and auto-save. itch has popups/buttons on the top right, which obscures the speed controls.