4 ms·

Heads up: the graph is misleading as it does not start that Y axis at 0. (Nor does it go back very far.)

by alanh 14y ago

mertd 14y ago
Well here: http://www.appdata.com/apps/facebook/225826214141508-draw-something-by-omgpop http://www.appdata.com/apps/facebook/225826214141508-draw-so... 5M users is still 33% of 15M (their peak) no matter how you plot it. It is a fad mini game without too much depth and little repeat value. Trend is not surprising. I'd say it is expected.
- alanh 14y ago
  Thanks, but how are those graphs better? They aren’t anchored at zero either and don’t show more than 30 days without some sort of premium account. > It is a fad mini game without too much depth and little repeat value. Trend is not surprising. I'd say it is expected. Most likely. I sure got tired of drawing the same, sponsored words.
scoot 14y ago
That hardly needs a "heads-up" (where else would one's head be?), as the graph is clearly labeled, so is in no way misleading. The fact that it doesn't go back very far only goes to emphasize how rapid the decline is.
- alanh 14y ago
  It’s not deceitful, but it’s still not worth much as a graph. 1. We think in terms of fractions. What fraction of users have left? It can be calculated from the numbers, but you don’t need a graph for that. If you started the axis at 0, it wouldn’t need calculated — it could be seen. 2. Not showing longer history makes it impossible to answer questions like, “how does the current rate of loss of users compare to the rate of growth a month earlier?” Your attempted point about “emphasizing how rapid the decline is” is, to be blunt, the stupidest thing I read today. How does showing less data, less clearly, help anyone? Clearly, showing more history would give an accurate picture of “how rapid” the decline is or isn’t — again, see point #2 above. Last point. As a thought exercise, consider a graph that goes up rapidly, then reaches an inflection point after which it slowly, steadily goes down — by no more than a total 1% over its run. It would look mostly flat after the inflection point, as part of the whole, complete graph. But zoom in, so the axes only contain points in that slow, mild down slope, and it looks like a dramatic slide from “a lot” to “nothing at all,” right? Not very helpful.

Heads up: the graph is misleading as it does not start that Y axis at 0. (Nor does it go back very far.)

by alanh 14y ago

Heads up: the graph is misleading as it does not start that Y axis at 0. (Nor does it go back very far.)

mertd 14y ago

Well here: http://www.appdata.com/apps/facebook/225826214141508-draw-something-by-omgpop http://www.appdata.com/apps/facebook/225826214141508-draw-so... 5M users is still 33% of 15M (their peak) no matter how you plot it. It is a fad mini game without too much depth and little repeat value. Trend is not surprising. I'd say it is expected.

alanh 14y ago
Thanks, but how are those graphs better? They aren’t anchored at zero either and don’t show more than 30 days without some sort of premium account. > It is a fad mini game without too much depth and little repeat value. Trend is not surprising. I'd say it is expected. Most likely. I sure got tired of drawing the same, sponsored words.

scoot 14y ago

That hardly needs a "heads-up" (where else would one's head be?), as the graph is clearly labeled, so is in no way misleading. The fact that it doesn't go back very far only goes to emphasize how rapid the decline is.

alanh 14y ago
It’s not deceitful, but it’s still not worth much as a graph. 1. We think in terms of fractions. What fraction of users have left? It can be calculated from the numbers, but you don’t need a graph for that. If you started the axis at 0, it wouldn’t need calculated — it could be seen. 2. Not showing longer history makes it impossible to answer questions like, “how does the current rate of loss of users compare to the rate of growth a month earlier?” Your attempted point about “emphasizing how rapid the decline is” is, to be blunt, the stupidest thing I read today. How does showing less data, less clearly, help anyone? Clearly, showing more history would give an accurate picture of “how rapid” the decline is or isn’t — again, see point #2 above. Last point. As a thought exercise, consider a graph that goes up rapidly, then reaches an inflection point after which it slowly, steadily goes down — by no more than a total 1% over its run. It would look mostly flat after the inflection point, as part of the whole, complete graph. But zoom in, so the axes only contain points in that slow, mild down slope, and it looks like a dramatic slide from “a lot” to “nothing at all,” right? Not very helpful.