Remember Fractional Beeminding? We’ve been putting it to great use for such goals as

- Our meta goal to keep churning out autodata integrations
- Project Euler, and
- Workerbee-Visible Improvements.

A key clarification before reviewing how it works: fractional beeminding doesn’t mean beeminding something with non-integer datapoints. Quite the opposite: it means beeminding an integer-valued metric where getting the next +1 is a big chunk of work, like on the order of days.

If you want to beemind reading books, usually you don’t want fractional beeminding. You just want a finer-grained metric than “books read”. Pages is the obvious choice.

If you find yourself reading part of a book and then jumping to a new shiny one and rarely finishing any — and if you want to commit to *not* doing that — then you *might* want fractional beeminding.
So here’s the review, using book-reading as the example:

- Again, you’re beeminding whole number of books read.
- Each day, or whenever you have a beemergency, enter whatever fraction of the book you estimate you’ve read. It need not be the actual fraction based on page count.
- Once you enter a fraction for a given book, you
*must not*enter progress on any other book until you’ve 100% read the book you started.

As we discuss in the original post, the seemingly loosey-goosey point 2 is backstopped by point 3. At worst, fractional beeminding of, say, reading a book a month devolves into normal beeminding, where once every 30 days you suddenly have to get a whole book read.

That was the problem I used to have with my Project Euler goal. [1] I want to commit to solving one problem per month. And unlike with reading a book, there’s not actually any natural way to measure incremental progress. You either have the right answer or you don’t. Before switching to fractional beeminding, my Project Euler beemergencies would sneak up on me. Suddenly I’d have to solve a whole Project Euler problem in days or hours and I am sadly too dumb for that. With fractional beeminding I can keep adding datapoints like “+3%, tried naive brute force in Mathematica” and “+3%, got that to work for up to n=12” or even several days of just “+3%, another stab at debugging the A-star search version” or “+3%, described status to Bethany”.

Many of those are dubious chunks of progress, but it’s getting me to at least think about the problem a little each day. By the time the final chunk is due and I have to actually get the solution, there’s a good chance I’ll be able to, having banged my head against it bit by bit over the last month.

## Officially transitioning the blog to being fractionally beeminded

I struggle with writer’s block and procrastination. It’s 2:30am as I write this. [2] Blog posts are a very natural use case for fractional beeminding. If you have a schedule you want to stick to (we’ve settled on fortnightly here) then you can’t beemind wordcount. You don’t know how long each post will be ahead of time. But you can make guesses for what percent complete a post is. Maybe you’ll be wildly optimistic and when the final chunk is due you’ll still end up pulling an all-nighter. But it’s strictly better than the status quo, where I sometimes am still staring at a blank screen with an entire post due that same day.

To further commit to making meaningful incremental progress, we’re hereby committing to — unless or until this gets too boring or tedious — sharing our quanta of progress in the forum or Discord. You can also keep an eye on the datapoints of the blog meta goal. Wish us luck!

## Footnotes

[1] We have a nice Project Euler autodata integration but you’ll need to use a manual goal if you want to do fractional beeminding.

[2] Why don’t I set the deadline for the blog goal earlier, you ask? That’s a different kind stress, trying to juggle the blog deadline with all my other beemergencies due at 5pm. And I do kind of like writing late at night sometimes.