I’m interested in a) helping out anyone who wants to make use of a similar system, and b) getting feedback on things that might work better in terms of tracking time…
Hi Joe! Robby from RescueTime here. That looks awesome! I’m curious about a couple of things:
Looks like your sampling interval is one minute. How do you account for activities where you swap back and forth a lot? Or, probably a better question, do you find that to be an issue? Seems like with a really small sample, (a few minutes) that lack of granularity might be a problem, but it would become less of one as your dataset grew larger. (Or, am I just reading that incorrectly?)
At first glance, the display doesn’t make a ton of sense to me. What do the rows represent? Previous days? And the units are minutes, right? I wonder if the rows could be labeled in a way that would make it more clear?
Do you feel like there was anything in particular that jumped out at you once you started tracking this stuff? Did the numbers match your perceptions of how you thought you spent your time?
I like it! It’s actually not all that dissimilar from early versions of RescueTime. We also started out with Applescript plus a patchwork of slapped together languages just to get something working.
I’m fairly happy with one minute sampling - certainly it’s the case that anything I’m doing where I am very rapidly switching between things is going to be productive work anyway - plus they will be in a ‘set’ so I’d be switching between say, terminal, Eclipse, and vim (Yes, I know, but…)
Yeah that would be the penalty of rolling-my-own - rows are days (I should make that clear in the description and it is in minutes.
The major thing that shocked me was how things like twitter creep up even if you are only glancing at them a couple of times a day. The other big thing for me was how big a slice of my day TV-watching was, which really didn’t match how I thought of myself at all. I’m yet to do any sustained analysis on the stuff (I’m trying to work out how best to mine the data to fill in bits of my google calendar. But it’s not massively obvious…
I’m a RescueTime user since almost the beginning, but I used to use TimeSprite for windows, which also did 1 minute sampling, and I liked that better. From talking with Joe Hruska, I understand that you guys think that having high accuracy is an important goal, but I disagree, and I don’t think that your fast sampling approach gets it.
First, you create a much bigger data problem by sampling so often (1 second perhaps?), and the apparent result is that the site is slow, especially when you look at long time ranges. Second, sampling so often drains the battery on my computer.
Lastly, in order for RescueTime to provide an accurate overview of how time is spent, you have to categorize every site and file that you open over the course of a day. For me, there’s a fairly large long tail of <1 min things that I don’t bother to categorize because it would waste too much time. The problem is that a big part of the RT userbase apparently is being tracked by their superiors with RT, and so many non-productive websites are tagged by default with high productivity scores. By not correcting this myself for all of the small chunks of time, the overall stats are unfortunately somewhat skewed.
In contrast, with 1 minute sampling, the burden to categorize everything isn’t very much, and taking over 500 samples per day is certainly enough to have a statistically very accurate idea of where the time went. Proper categorization of activities makes a much bigger difference than the number of samples once the number of samples goes above 50 or so.
I’ve been recently using time tracking tools especially at work to help me stay focus and limit wasted time. I personally used Time Doctor. I also tried Rescuetime and it doesn’t fit on my needs where I can’t organize my tasks and I feel it’s intrusive. I won’t know if it is still tracking my computer during non working hours.
