Conway and Doyle can divide by three, but I can't

[worez]

https://arxiv.org/abs/2309.11634

[tromp]

Discussed 7 weeks ago at https://news.ycombinator.com/item?id=37618301

[ganzuul]

Perhaps doing the thing with a compass and straightedge would feel therapeutic at this point.

[onenukecourse]

Im sure this is about some higher end math than I dont understand, but in day-to-day mental math just

1. multiply by 3

2. divide by ten

3. If more accuracy is needed, add 10% to the answer.

Of course, you can always add up all the digits and see if it's divisible by 3. If not, add/subtract to make it divisible by three, divide it and then correct for the borrowing.

[jprete → onenukecourse]

It’s a set theory problem, calling it division by three in the paper is a bit of a joke.

[montroser → onenukecourse]

4. If even _more_ accuracy is needed, add another 1%

5. If even _more_ accuracy is needed, add another 0.1%

5. If even _more_ accuracy is needed, add another 0.01%

and keep going ad nauseam!

[philipswood]

I keep seeing {𝐴𝑖 }𝑖∈N

as "alien".

[bitwize]

Dividing by 3 is hard, man.

While an engineer at Xerox, my father modified an electromagnetic calculator to turn it into a form of stored program computer (similar in form and function to the Olivetti Programma 100). While experimenting with the calculator he found a way to crash it: divide 10 by 3.

He got a patent for his electromechanical computer invention, for which Xerox paid the standard nominal fee of $1, but it went nowhere.