Continuing the discussion from IFTTT Pre-Launch:
Nice, thanks so much for sharing the recipe! (Though personally I think total steps is a much better metric to mind than days you hit your goal. We argue the point in âWhat To Mind: Picking a Metricâ.)
2 Likes
Total steps definitely wins overall, for me. The complementary goal of X days per week is also great. I particularly love that IFTTT will let people create those types of goal without the hassle of double entry. I especially like @tomjenâs example of counting days-above-threshold.
1 Like
I tried beeminding total steps once, but I have a high variance in daily so the useful version of the abstract goal - walk more - seems to be hitting my step goal on more days, rather than walk 10% higher total steps. I tried the habit of hitting the goal every day for a month, but that isnât sustainable, my hope is that this will get me into the right habit.
1 Like
@tomjen and @philip, no no, youâre all wrong! Beeminding number of days above a threshold doesnât help with high variance. Averaging over many days of data â as Beeminder does naturally â does.
If we really want to resolve this weâll need to understand the underlying physiology. What are the relative health benefits of super bursty vs super uniform exercise, holding total movement constant? I feel like even if super uniform is much better, beeminding number of days above a threshold is too crude a way to account for the diminishing returns to marathon sessions. (And my suspicion is that bursty is actually better anyway.)
Maybe we want something like the opposite of the new triangular beeminding â logarithmic beeminding, perhaps, where the number of points you get for each additional step gradually levels off. Let me know if youâd like to guineapig such a thing. Super easy to add more aggregation functions.
Yet another approach is to just use autoratcheting aka max safe days. For example, if your max safe days is 1 then you have to hit your daily threshold every other day. But the graph still shows total steps on the y-axis, as the good Lord intended.
(This is my Quantified Self bias. If you donât care about Quantified Self then maybe youâre fine sacrificing richness of data in order to focus on the simplest scheme to motivate yourself. Though I still feel that days above a threshold is not as aligned with the underlying heath/fitness goal as it could be. Way better than nothing, though, so if itâs the thing youâll be motivated by and stick to then go for it.)
1 Like
I love looking at the graphs, but I get like 95% of my value out of the beeminder sting, so I will guess I just have to disagree with you and end the debate since there is still a tiny hope that the thread can go back to what the topic originally was.
Besides if I want the data I can just look fitbit since they have the graphs, so no data has been destroyed.
2 Likes
This is measuring two different things:
- minding the number of steps
- minding the number of âadequateâ days
For #1, I think weâre all in violent agreement that beeminding the average is superior to many other methods. It is ridiculously frustrating when your pedometer hit 9900 steps and itâs counted for nothing by our productivity competitors.
For #2, itâs more complicated. Arguably any binary goal is tracking days of effort above a not-aways-well-defined threshold.
I like @dreevâs suggestion of setting an autoratchet threshold to force some effort every N days. But that approach also requires that the daily minimum and the weekly minimum are the same rate. (Not strictly true, but sufficiently so for the purposes of discussion)
Hereâs a great example from @aspiers about separating a daily minimum (threshold-driven!) goal from an overall goal. Heâll still need two Beeminder goals to achieve that, but at least now he can automate it using an IFTTT recipe like @tomjenâsâŚ
In short. we agree that if itâs steps (or equivalent) that weâre beeminding, then minding the raw count is the right approach. Thatâs true regardless of autoratchet or aggregation settings. But sometimes you also want to count the days on which you accomplish âenoughâ toward your goal, in the same spirit as a must-do or a user-visible improvementâŚ
1 Like
Just wanted to point out that whatever conclusion you come to regarding burstiness and physiology wonât apply to my use case of music practice. Also I do care about QS and accurate data collection.
1 Like
I have this same problem when it comes to cleaning the house. There the jury is definitely in: burst cleaning is suboptimal because if you only clean once a month your house is dirty 30/31 days of the month.
3 Likes
For #1, I think weâre all in violent agreement that beeminding the average is superior to many other methods. It is ridiculously frustrating when your pedometer hit 9900 steps and itâs counted for nothing by our productivity competitors.
That is actually a really great feature - it is fustrating enough that it gets me of my butt, whereas just tracking steps I could make up for them tomorrow (unless it is an eep day).
Anyway I decided to take your advice and create a second goal to track actual steps, with a (for now) insanely low rate (1 a week, which is literally impossible for me not to reach unless I am in a coma) - that way I can always make it more difficult and I get access to the data later.
2 Likes
Maybe I shouldâve said âridiculously frustrating when you donât noticeâŚâ, because I donât want to obsess over checking my step count, unless its an emergency day. There are plenty of other things for me to obsess over⌠
2 Likes
YeahâŚI was going to end a day with 92 flights of stairs so I ended up climbing up and down my buildingâs stairwell until I hit 100 to get the fitbit badge.
2 Likes