| I second this. I absolutely love the visuals, this style is pretty much what I dreamed of eventually building for myself if I ever had some spare time. But I'm also concerned with the fact that it's optimized for looks, not for utility, and I want to make a general observation about the current Quantified Self / fitness trends. What I've noticed about them is that products are optimized to look pretty, not to be useful. Those two things are related, but it matters which you optimize for. Depending on which one you care about, you make different choices about e.g. visualization tools you use. To pick on AprilZero, since it's the topic of this thread, what I see on the site are: - Charts without scales (or with very impoverished ones) - you can't read too many interesting things from them, whether about time or value of measured parameters. There's data loss. - Gauge charts, which are cousins of a pie chart, i.e. the thing you use when you care about how nice something looks, as opposed to actually making it useful for learning something from the data. - The weight chart that seems to be doing exactly the opposite of what it should; splines look pretty, but what you want to aim for is a rolling average, which acts as a low-pass filter, eliminating noise from measurements and revealing your weight trend. See how the line leads the data points? It should be lagging behind them. For an example of what I mean, see [0]. - Just data, lack of any attempt to infer a course of action based on it. Again, the visual are absolutely stunning and I applaud the work. But now that we've figured out how to make data pretty, we need to focus on how to make it useful. For that, we need to start doing the hard work - understanding what choices we do have and how the data can tell us which decisions to make. For instance, I could track my weight and things I eat, and draw pretty lines of macronutrients over time, etc. but what I really want to have is something that'll dynamically rebalance my diet to keep them at optimal levels. I want the software to tell me to eat X or Y for breakfast. And if I feel like eating something entirely different, to have it readjust on the fly. I ate too much fat today, so the suggestions for tomorrow will have less fat. An iterated version of this program could plan stuff in advance and generate me a shopping list, etc. You can probably imagine more possible improvements. But note how in the example above, it turned out that I need exactly zero charts. The input I provide is measurements, but the output I care about is right decisions. Sure, I may like (or need, for something else) to look at the measurement data presented in a nice, visual form. Until my imaginary software is good enough, I might actually need to. But you can see how the focus shifts from just having a line chart to having a representation that will be useful to make decisions. TL;DR: charts and dashboards are means to an end, not end in itself. [0] - http://imgur.com/YcXK5af - it's a screenshot of a graph I made for myself when I went on diet a few years back; the blue line connects the actual samples, the green one is a rolling average with 7-sample window. Note how the green line lags behind the blue and how it filters out noise. Incidentally, this is exactly why the common advice given to people is to weigh yourself once a week, and not daily - because it's simpler to tell them that than to teach them about rolling averages and low-pass filters. |