Outfit alerts

Created by: Anonymous

This one's a bit complicated. On the simplest level, it caches parameters about other outfits - outfit activity, members (including leaders, people in leader positions etc) and expected daily activity - that's to say the maximum amount of players that are expected to be online at random given any moment. For simple purposes, the database could simply be provided these values, however in a more complex setting, it could be taught to learn and estimate these parameters based on random sampling from the API. Accuracy versus amount of data.

Then comes the second part, and that is using that data. If at any point certain conditions of an outfit are met, we can assume that players just mass-logged in for an event. Seeing as though most players use external communication prior to such, it's safe to assume that they logged in with similar goals in mind. Every 5-10 minutes the bot will run a sweep of all outfit member listings given a list of specific outfits, and determine an event in the following fashion:

-A certain number of players log in within a short time of one another based on the cached parameters. -The number of players online exceeds the estimated maximum -People of a leadership rank in the outfit are online and earning SL ribbons, or have recently logged in with a certain amount of other players.

These factors would trigger an alert, where the bot would routinely suggest the possibility of ops taking place, stating the requirements met. The more met, the more likely an event is.

Third point has us at the even more complicated stuff. Where we have already stored parameters in a constantly learning database that aims to estimate values for an outfit with regards to activity in a normal range, and have then programmed the bot to detect abnormalities in those values when compared to actual live values and report them routinely. That's a summary. Now we're trying to figure out what exactly it is that the outfit in question is doing.

If the conditions for abnormality are met, the bot will check the continents that are unlocked and their populations, then, using a method to be decided later, will decide which continent is most likely to be the active site of the operations, granted it might not be accurate, but given Hossin and Esamir during primetime, for example. This could be as simple as just checking populations of each continent and the chances of people playing (again, Hossin versus Esamir), but could get as complicated as analyzing spikes in population (a bunch of people on Hossin during primetime would suggest an organized outfit is there actively doing stuff - if it matches the faction and rough number of people, it could be a pointer. Also, checking how likely players are to go to a continent at any point given some simple statistical analysis is also possible, but, again, more code and numbers and shit, so maybe all that's not necessary until the basic features are down). We can then use the knowledge of which warpgate the faction of the outfit in question owns, and so guess as to possible sites/lanes of contention. This would narrow down the search for said outfit drastically, but the accuracy might be a bit off unless we get the methods right.

Fourth point

Not really necessary, but it would be fun to see just how accurately we can predict stuff about players based on API data. In this case, ribbons earned recently are saved in the API, and checking which they are could hint as to their methods. Certain outfits are known for using certain tactics, and so key players could be tagged with a certain ribbon, however each player tagged with squad leadership as to determine who the squad/platoon leaders are/could be. The bot would then summarize the ribbon findings to guess squad/platoon leaders (and honestly if more than 4 people online are earning squad LS ribbons or someone is earning P leadership that might be a hint as to whether or not ops are going on). The tactics could be discerned, as said, through weapons and so the ribbons could be just written out, and let us draw our own concvlusions based on knowledge of the outfit.

So that's all I have. I'm not sure as far as coding goes how much time or work, moreover how much code or workload from the API this would take, how much of it is plausible and if this would even prove conceptually effective. I have pretty solid knowledge in stats, but i know nothing about coding - what I know stops where info about the game stops.

That being said, I want the info, if finally found, to be presented inconclusively and broadly as to allow us to draw our own conclusions.

To summarize:

Abnormality is detected in outfit membership logins, is reported. Whereabouts of outfit are approximated roughly. Leadership of ops is estimated. Tactics of outfit are guessed.

This would allow us to quickly log in and "shit in the cheerios" of any outfit when they are having ops, within a margin of error. The more data the bot routinely collects, the better it becomes at profiling and accurately estimating events.

That's my idea. At its core, i would be fine with just knowing how many players are online for certain outfits, but these complex features would make for a nice "abuse" of the API to gather apparently live info, with some latency, albeit, but still.

And, hopefully, with this information, we could record and observe when the ops happen for certain outfits, and work out probable times for when they have their ops scheduled (outfits are sort of routine-oriented like that - i would know, I run one!)