|
Author |
Thread Statistics | Show CCP posts - 13 post(s) |
|
CCP Bayesian
886
|
Posted - 2013.06.25 13:58:00 -
[1] - Quote
OMG FIRST! EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
887
|
Posted - 2013.06.25 14:57:00 -
[2] - Quote
War Kitten wrote:What constituted a hacking attempt prior to Odyssey?
What constitutes one now?
Data is pretty but needs more definition.
The 65% failsuccess rate seems to me that you'd be counting two tries at the same can as 2 attempts, whereas before it might have only been one attempt? I can't imagine you counted every cycle of a hacking mod as an attempt in the pre-odyssey data.
EDIT: going back now, i see the top graphs were for completions, not attempts. It makes perfect sense that people are trying a new feature more often. Let's see what the graph looks like long term. :)
EDIT 2: 65% success rate, my bad. Knew what I meant, but labeled it wrong.
There was no such thing as a hacking attempt counter pre-Odyssey so we're only considering what has happened since release in that one graph. Similarly the ratio graphs only are tracks of change since the release of Odyssey.
EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
888
|
Posted - 2013.06.25 15:55:00 -
[3] - Quote
oklem wrote:I'm sure the frigates added in forsaken hubs accounts for some of the completion drop.
I asked around and CCP Fozzie just told me about this as well. It seems reasonable as an answer. EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
889
|
Posted - 2013.06.25 16:15:00 -
[4] - Quote
Kai Pirinha, I'm mostly considering the statistics in terms of what people did in the sites and trying to find comparisons in other places that shed some light on that. The big problem is that I can only work from the information that was available before Odyssey to do that. In the future we'll endeavour to add statistics in a patch prior to releasing an expansion so we can see change over that boundary for elements we want to measure. Also our system needs some additions to aggregate data which it doesn't have in our automated viewer/graph generator so I was forced to just pick some interesting sites rather than looking at the aggregate data for all Data/Relic Sites versus Combat Sites.
I don't think it's sensible to draw too firm conclusions from this data at all. As mentioned in the blog we're working to add more fine grained information in. The big ticket numbers are also easier to talk about in the generalities that the blog sticks too.
We consider all sorts of data sources other than raw user numbers though. Our Research and Statistics guys have been conducting surveys for many expansions now, as just one example. EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
897
|
Posted - 2013.06.26 08:37:00 -
[5] - Quote
Thanks for all the comments!
For those of you asking for statistics on other features that changed in Odyssey, I hope some will be along soon as other teams take a look at their specific areas of interest in more depth. EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
897
|
Posted - 2013.06.26 09:17:00 -
[6] - Quote
Nexus Day wrote:You show one year worth of exploration being flat lined, then a spike when it is "improved". How does this spike compare to past "improvements"? In other words how much of the spike is due to something being new?
Thanks for the excellent questions.
It's a fairly common trend across videogames for the new new to be extremely popular for a short period of time before declining to a stable population then declining further until death. How long that tail is depends on how good the game is. This landscape is changing a bit with the advent of alpha-funding and other similar models though.
On EVE a lot of changes that have predictable market effects will also get lots of speculator uptake as well. You can see this in the Hacking Attempts graph as the first week is full of activity which begins to tail off at a quite quick rate which rapidly begins to decrease with minor bumps for weekends. This is an expected trend as the players who still don't enjoy the feature and speculators leave to new pastures. I'm basically taking a bet on where the stable level will be, the 100,000 line is the one the graph is dancing above and below. The downward trend during the week also stopped for the first time today and we saw a slight uptick in attempts yesterday (which might just be due to the publicity this blog generated or it might just be noise).
Quote:Since the "improvement" the trend for relic and data sites is downward. You could easily say this isn't indicative of future results due to the small sample size. But then why would you show results with a small sample size to begin with?
Downward after, in some cases, a more than ten-fold increase in use. As above this is what we'd expect to see with any new feature addition or major change.
Quote:Hacking attempts is not indicative of people using the feature. As a matter of fact none of your graphs are, they just represent runs or attempts. The whole point of dumbing down exploration was to attract more people to it. Do you have any data to support that, or are the old explorers just taking advantage of you making it easier?
They definitely don't measure directly the number of people using the feature but they are definitely indicative of the number of people using it otherwise we wouldn't see an uptick at the weekends when we know more people are online. One thing that suggests that this is not just the old players exploiting an "easier system" is the scale of uptake.
This data collection and aggregation on a larger scale is quite new and as you reasonably point out this captures only at the character/user level. It says nothing about the subscriber level where people have multiple accounts. However characters are owned by users and users are subscribers so it should be possible to aggregate up to the subscriber level. This sort of data mining will be done by our Research and Statistics guys. EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
897
|
Posted - 2013.06.26 13:56:00 -
[7] - Quote
Prop Wash, you tell me not to mistake correlation for causation and then go on to tell me how the changes we made caused the change in the number of people using one of the sites!
That aside we definitely don't see rising numbers as a compliment but it's one metric amongst many that we do look at to judge the success of a feature against our goals for it. EVE Software Engineer Team Prototyping Rocks |
|
|
CCP Bayesian
897
|
Posted - 2013.06.26 14:19:00 -
[8] - Quote
The changes to scanning aren't something I worked on directly or commented on apart from as an obvious confounding factor in the devblog, so it's not really a topic I can actually address. EVE Software Engineer Team Prototyping Rocks |
|
|
|
|