Last modified: 2013-06-04 (finished). Epistemic state: log.

(Wait what? words? I need to write about stuff we did? Again? Didn’t I just… oh well. Fetch me the logs, I’ll docum.. wait, it says there I mostly did Japanese grammar practice, which I can’t meaningfully talk about yet, loads of boring-to-talk-about deliberate practice, and work, which I bet I’m not even allowed to talk about because we pretend we’re relevant to the industry? Ok, so can I post a manifesto then? I have a draft and need to get some of this anger out of… no? What about metaphys.. also no?

Listen, Mister. I know you had the best of intentions with this “write about stuff you did because you like explaining stuff, which helps you actually do stuff”, but we have some better tools now, if you haven’t noticed, and while this works well for conceptual stuff, most skills are actually pretty hard to talk about in a meaningful way and learning them doesn’t really leave neat piles of notes that can be trivially converted into posts, so now you’re making me do extra work here. Can’t we go back to “deconstruct texts” and stuff? That’s at least fun.1

Yes I know the rules, I wrote them, but even Kant would agree that some rules can be broken somet.. ok yes he wouldn’t, but you know what I mean! Look, your paranoia isn’t helping me either! You’re so worried about distant effects of your actions, you’re not letting me talk about topics before you’re sure they’re fully under your control, but you’re still making me adhere to rules that were written under the assumptions that just documenting whatever happens is way awesome, so don’t you see the problem here? A lesson? What lesson could you possibly try to teach me here?

“Sources of validation”?

I see. Yeah that’s really clever of you. So if I don’t break the rule, I’ll have to find something to write about even though I don’t have anything ready yet. Yeah, I get how this extra work is great for your complicated plans. But if I just break it? What th.. whoa, that’s a big graph. That’s all conn.. even th.. whoa. Sheesh man, you’re quite the magician, you know that? Yes I know I joked about how I’m doing acausal magic when I wrote the rules. I didn’t expect you meant that bit! So if I don’t follow the rules, all.. that.. collapses? Well, that’s a pretty double bind you set up there. Listen, Watzlawick’s books weren’t instructions how to mess people up. I don’t care that it works! Don’t melon me! There are nicer ways to get what you want. I hate when you cloud your incentives with magic talk and convoluted narratives. What do you mean that’s the point?

So it’s actually a triple bind? For all three possible modes of response - loser, clueless, sociopath - you’ve set up a demand that transparently subverts it, and is in conflict with the other two modes. The loser narrative is unbelievable because I can see the strings, the clueless reward is nonsense because I see you printing that toy money, the sociopath power doesn’t work because I’m afraid what using it will do to me. If I use any option anyway, you still win, and if I transcend their limitations, you also win.

You’re quite the evil fuck, you know that?

…so what now?)

What’s the saying? If you don’t know what to do, go meta. So I’ll just document one of my recent Beeminding changes.

Starting on 2013-02-28, I redesigned several things: I’ve started to switch from open-ended time-based goals to finite task-based goals, broken my big “do useful stuff” goal into smaller sub-goals, and begun to add percentile feedback.

I explained some of the goal organization over on Reddit, relevant quote:

I used to just track “productive hours / day” so I don’t sit on my ass too much, but I started to substitute more fun projects for everything else, so I recently broke that up into various independent groups of goals: “science”, “guitar”, “code”, “polyglottery” and “job” (ugh). I still track the total too, but that’s more of a nuclear option (with the big 90$ pledge) and I haven’t settled on a good daily goal yet. If I’m too harsh on any one goal, it’s more likely I just decide it’s not worth it and fail even though I still have plenty of time to meet it.

I’ve toned down some of the time-based goals because I found “do stuff for 20 minutes” not very motivating. There’s no sense of progress and “4 hours” looks so large, I often didn’t even try to tackle it even when I had plenty of time left. Instead, I’ve begun to convert everything into a TODO list using org-mode. So for example, I’ve just turned every chapter / unit of a textbook into one todo item, which feels much more actionable and is done. (Also wrote a simple script that updates Beeminder goals accordingly.2)

So far this seems to be more productive, but I’ve not been doing it for very long. I tried a simple Bayesian comparison of a before/after dataset, mostly to figure out how R work. I just wanted to use BEST, but the normality of “useful hours/day” turned out very low, meaning the data doesn’t seem to fit a normal distribution very well. After looking at a basic histogram:

Fume Histogram

I saw why. While this looks reasonably Poisson-y, I still don’t understand the underlying assumptions and methods very well, so even though I could hack together a simple version of “I put numbers into this magic box and look at these funny graphs it gave me!”, that wouldn’t be particularly useful.

However, I just plugged my “total useful hours / day” into a simple Beeminder “weight gain” graph (so that I can track (and then raise) my fluctuating average), and you can still see two improvements (the first time I broke stuff down at the end of March, and the second slope when I transitioned to tasks):

Fume Average Graph

(Note that this is zoomed in and cuts off some >6h outliers. The points are “total useful time” for each day as logged with my time-tracking tool, the thin line is a (slightly lagging) average.)

I’m also trying to use this graph and a simple percentile number in fume to implement Seth’s percentile feedback, so instead of just seeing “I did 1.2 hours of useful stuff today”, I also see how good that is compared to the last week.

I’m not sure yet if it actually works in terms of increasing my productive time (not enough data), but at least it’s more motivating and I’ve made significant progress on several projects due to the tiny todos that feel more like “one more turn” instead of “fix ALL the things!”.

I got annoyed with Dropbox mangling my data and so thought, hey my ~/spoiler directory (i.e. all my notes, todo lists and drafts) is already a (huge) git repository with auto-commit, so I could just use some git solution to sync that between machines. Then all the stuff in Dropbox is either read-only (public shares) or written by only one machine ever (phone -> desktop) and I won’t have sync issues anymore. Shouldn’t be hard, right?

What I considered:

  1. Use git-annex. Long-term that’s fine, but currently the assistant ignores the .gitignore (wtf?!) and I’m beginning to develop a deep-seated hatred against all Haskell programmers.3

  2. Sparkleshare is, quote, “an Open Source collaboration and sharing tool that is designed to keep things simple and to stay out of your way”. It’s also written in C#. Good joke. Is funny. I laugh.

  3. dvcs-autosync is similar, but “takes a more minimalistic approach”. Which includes no documentation, limited directory awareness and, best of all, requires a central repository. It has “distributed” in its fucking name. Who’d be so dement.. oh, guy uses Arch Linux. I’m now convinced every Archer is clinically insane.4

So there isn’t any good tool yet. Well, I could turn one of those (probably dvcs-autosync) into something usable, but ain’t nobody got time for that. So it’s back to my cronjob’d commit-and-pull-everything solution until git-annex’ assistant stops sucking. (Nonetheless, I still put “use dvcs-autosync to auto-commit based on inotify and insta-push via git hook” on my todo, in case I do got time for that.)


You know what’s really annoying? Hair feedback time.

So I wanna do this totally awesome undercut thingie5, but it will take me months, if not a full year until some parts have grown enough that I can pull it off, and until then I can’t do shit.6 I just practice a bit by doing regular maintenance, but how am I supposed to level up my self-sustaining fabulousness if I can make maybe 1-2 big decisions per year and mistakes might require rollbacks so large, you gonna have to wait months just to repeat them?

On the plus side, while I was going through, I felt - for the first time in my life - a sense of fear of being outdated, that by the time this “just long enough to poke me in the eye” mess has grown into something that can unambiguously blow in the wind and has enough surface area for a dozen Hello Kitty hair bows, that by then fashion will have moved on and I’ll look so wannabe 90’s, I might as well buy a fedora.

Then I remember no one cares and I go back to Pinterest.

A mini-review of the newest Crusader Kings II DLC, The Old Gods. It adds a new start in 867, the year Rurik decided that Russian brides are just the best and he’s gonna grab himself some in Novgorod7, and more importantly, before the spread of Christianity into the East. I’ve always wanted to play before the Ostsiedlung, so I just jumped right into Lusatia and decided that minority status isn’t good enough for us Sorbs, we’re gonna carve ourselves a kingdom in the midst of Vikings, Poles, invading nomads, and worst of all, Bavarians.

The new raid mechanics are a beautiful addition. At first, I was just assembling little hunting parties from time to time because I thought that it was my duty as a Pagan to pillage, but by picking vulnerable Christian provinces too busy with bigger assholes than me, I had quickly transformed my fiscal policy into “whatever the Christians can’t defend”. Greater Poland acting up because it thinks it should own Silesia? Better “borrow” some money from the cities to the south. Vikings landed again? Let’s make it Nürnberg’s problem and buy some mercenaries with their gold. Once I realized I can finance the majority of my armies with ransom money, I couldn’t help but start blasting Death Metal and cheer the burning of churches. The Christians might worship a pathetic god, but at least their coffers come back from the dead year after year, ready for another harvest.

The fascinating thing is how this high lasted for some time until some even bigger Pagan decided I would make a nice addition to their ever-growing list of vassals, and I realized I ain’t got no Divine Right of Kings, I just got a big axe, and after the last succession crisis of 6 sons (did I mention those concubines?), I was exhausted and without friends, and maybe all those saints I gladly turned into martyrs might’ve been on to something and I should’ve taken up the Axe of the Apostles instead. But even though the March of Progress might be against me, and even though stability and order win in the long run and I got neither, at least for a few decades I was feared by all and when death inevitably comes, I won’t have any regrets.

My sons, though, are fucked.


  1. But seriously, I don’t always like to shoe-horn fragments, cool links and “omg they didn’t just write that” parts into logs. I know people use their FB or G+ for that, but I don’t like either site (because at least one of them is probably run by actual soul-eating vampires).

    I wonder if I should start a tumblr. While I don’t have angry opinions about gender yet, I am confused about my personal identity, sexuality, hair style and species. I would like to regularly reblog sassy pugs and summarize complex arguments with a reaction.gif, but I’m not sure if I’m ready to become a 15-year-old girl full-time.

    There needs to be a place that is kinda in-between FB and tumblr. A fumblr, if you will.

  2. Beeminder has native Trello integration, which is probably a good alternative if you’re not already familiar with org-mode. Still, Trello doesn’t have nested items and can’t be used offline, so that’s somewhat annoying.

  3. Seriously, Haskell has a worse dependency hell than the ARM code I maintain which still uses gcc 2.95. Oh, I need to update a minor library from x.y.3 to x.y.4? Every single Haskell program ever on my machine is now broken. I hate academics.

  4. Sometimes I feel like ranting about Arch, but then I remember pacman doesn’t even have versioned packages and I don’t think any more needs to be said about it. At least the suckless guys make great trolls and occasionally write cool projects. Arch just seems to rot the mind with cargo cult minimalism.

  5. Look, I have strong opinions about the metaphysics and epistemology of 19th century Dutch theologians. It’s pretty much mandatory that I either look like Skrillex or an 80’s bum. Which isn’t mutually exclusive, fortunately.

  6. I expect to get drunk sometime this week, do it anyway, freak out, complain on Twitter, then love my decisiveness and use the experience as yet another guilt token later, admonishing myself why I can’t be as awesome as I used to be. That’s how I usually get stuff done I have issues about.

  7. As mrout on Reddit said, “Hey Japan! This is how you do a rape game without being weird about it.”.

blog comments powered by Disqus
dlog » daily log » bel-marduk