I read something (actually it might have been posted in this thread, as I think about it) from developer who posted a conversation they had with reddit, where reddit mentioned the guy's app was inefficient with their api calls. The dev's response... actually, instead of trying to remember, I'll just post it:
Quote:
Another common claim by Reddit is that Apollo is inherently inefficient, using on average 345 requests per day per user, while some other apps use 100. I'd like to use some numbers to illustrate why I think this is very unfairly framing it.
Up until a week ago, the stated Reddit API rate limits that apps were asked to operate within was 60 requests per minute per user. That works out to a total of 86,400 per day. Reddit stated that Apollo uses 345 requests per user per day on average, which is also in line with my findings. Thats 0.4% of the limit Reddit was previously imposing, which I would say is quite efficient.
|
So the guy's app uses (on average) 345 requests per day per user, compared to other apps that only use 100 per day per user... and the his response was essentially "who cares, we're allowed to use up to 86,400."
I mean, that mentality isn't exactly help anyone. If I was in Reddit's shoes, I would set realistically acceptable efficiency targets and then give the dev community 3 (6, maybe?) months to get their calls within those targets. After that, I would install fair/competitive rates for apps that complied (maybe even reward the most efficient apps with lower charges). Then I would subject any apps that don't come come under those targets to ridiculously higher rates on all their calls (not just the overage) until they complied or shut down.
Edit: Also, the dev I quoted above used this analogy:
Quote:
|
As an analogy (can you tell I love analogies?), to scale the numbers, if I was to borrow my friend’s car and he said “Please don’t drive it more than 864 miles” and I returned the car with 3.4 miles driven, I think he’d be pretty happy with my low use. The fact that a different friend one week only used 1 mile is really cool, but I don't think either person is "inefficient".
|
But what they need to think about is that api calls are fairly predictable (on a per call basis, using reasonably similar parameters). Like, everyone is making the same calls. So, using his analogy, it would be like the car he's talking about can only go to the same store from the same house, no matter who drives it. So are the two people using this car the same number of times each? If so, his 3.4 miles is not nearly as efficient as the other person's 1 mile.
The one thing missing from this guy's post is a way to differentiate the two useage totals. Since they're only providing per day numbers, it's hard to really understand the full picture without #of unique visits or even calls per view or... I don't know... anything tangible, really.