How many bots to make a tracker?
I have had discussions with some of my friends about the use of trackers. To me the big no-no is the massive use of bots. I really despise bots in any game, so to me the use of that is out of question.
But how "massive use" is it really? To get a good close-to-realtime coverage I suspect you probably need more bots than the typical number of players in all but the most active big city areas. Anyone knows?
Answers
We just got our first tracker about a week ago, a rocket map one that has some statistics exposed.
In my city there are around 90k mons per 24 hours; so around 3750 per hour, so around 60 per minute. But I'm not sure those are relevant numbers for estimating bots number.
I think there are around 4000 spawn points that needs to be covered each minute since the report seams to be at about 1 minute after spawn.
The bots also covers gyms, but not like gymhunter with detailed info, only team, lvl and prestige. Gyms seam to be scanned once per minute (I monitored last scan date on one of them), we have 68 scanned gyms.
Would be fair to say there is a bot per spawn point and a bot per gym, continuously monitoring?
A, and there are also pokestops monitored for rules. One more bot for each of almost 400 pokestops?
This is totally a guess...a relatively educated guess, but I could be totally wrong.
I would imagine that the number of bots needed would be determined by the size of the area covered, rather than the number of spawns...but there is probably also a few different ways to do it.
I would hazard to guess that each bot covers an area similar to what you can see when you stand still - so about 20m in every direction. It would pick up any mons and stops within that area.
There would need to be overlap as the area covered would most likely be circular.
If the bots move around too much then they would be subject to soft bans, and even permanent bans - so each individual bot could not cover a lot of area, nor could it move very much.
I can imagine that to cover a square km, you would need around 40-50 bots to pick up everything.
I could be way off on my prediction, so don't take it as absolute - but this is how I imagine that they work.
might be so since also that way they will be able to get the info during first minute of spawn.
And your way will work also from the start, when spawn points are not known.
From personal experience in my city I think this will only half the number of bots compared to a bot per spawn point, so will still be thousands.
Area around 100 square kms, so 4-5000 based on your approximation per square km :).
I think at least 10x more than number of players :(.
Even at half of my prediction, that is still 2000-3000 bots just for that one map.
This explains why Niantic are so persistent in trying to remove maps and scanners - the load on servers and resources must be huge.
It does make me wonder if it would be more cost effective for Niantic to produce their own service...and charge by the month for access - even if it only covered a 500m radius around the players current position.
It would be another revenue maker, and would give players a more legit way to get what they are clearly screaming out for. It may vastly reduce the number of players who use maps.
It could be a win/win for them.
I would like that but not with the exact spot given. Let me tell you why:
I found the only Unown spawned in my city in the first day of the tracker, spawned 200 meters from my office in a place I usually hunt. I excused myself for the start of my upcoming meeting and caught it. Happy? A bit more than caching a Rhydon before using the tracker. No surprise, I went targeted to the exact spot. So did several other from the city.
And since we have the tracker I understood GGTW's mentions that he plays by foot.
To leverage the tracker you need some kind of motorization. I used that in first 2 days and in previous weekend but somehow feels fake :(. But also addictive, I don't see myself giving up on tracker unless it gets closed or asked for payment per month bigger than what I give to Niantic in 2 weeks (it is free for now).
That is the second reason I do not use trackers, I know for sure they would diminish my feeling of accomplishment.
In my opinion the ingame tracker need some serious improvement to stop trackers.
Skip the pokestop dependence, and cover longer distances for mons you catched less of and prioritize those (maybe up to 500m). Clicking on the mon in "sightings" would give you the direction to it. That would both give a nice hunt feel and increase cooperation by triangulation.
the question is maybe not necessarily how many
scripts exactly you have to run to cover
a given area in about realtime, but how many times.
well, bots are scripts not people.
they don't walk around anywhere looking for pokemon or something.
someone took a close look at the datapakets send between the app and the gameserver,
then wrote a script to
establish such a connection,
read out the relevant parts
and store them in a database
in order display them on a map.
i guess they just connect to the server using a false gps-dataset
long enough to get the initial dataset and disconnect again.
a couple of milliseconds beginning to end.
then again with a slightly different gps-dataset and so on, and so on.
at least this way a large area can be covered in a very short time.
they use up bandwidth -and- rise the dataload the server has to deal with,
for every connecting instance of the script gets a player-id, and all that goes with it.
not as straight forward as a genuine DOS attack, but depending on for example how long the gameserver stores
those informations maybe just as harmfull.
but who knows, its all just guesswork.
