OpenStreetMap

Pokemon GO Mappers - What They Do and Why They Do It

Posted by WoodWoseWulf on 21 June 2019 in English (English)

It’s been three years since the launch of Niantic’s Pokemon GO app, a lifetime in the online world. After all this time, there remains a certain frustration and tension between Open Street Map (OSM) volunteer mappers and a subsection of Pokemon GO trainers who seek to edit the map in order to enhance or even cheat their game experience. At least in part, this is caused by a series of misunderstandings about the OSM project, Pokemon GO and the relationship between the two.

Several guides exist for budding Pokemon GO mappers, such as https://wiki.openstreetmap.org/wiki/Tips_for_new_(Pokemon_Go)_mappers. However, there are scarce resources aimed in the other direction, that is to say, guides to help mappers understand the motivations driving trainers to make changes to the map, and ways to understand what kinds of changes said trainers are likely to want to make.

Glossary

Some of the terms I will use here may be unfamiliar to mappers, so I will define them here.

EX Raid/EX Eligible:

EX Eligibility is a special characteristic of certain Gyms (see POI definition below) that allows that Gym to occasionally output invitations for a local event in which trainers can obtain a rare Pokemon. (See “EX Eligibility” sub-section below for more detail)

Biome:

A term used by the Pokemon GO community to describe locations where certain kinds of Pokemon might appear. For example, proximity to a water biome might result in more fish like Pokemon and forests might attract more bug-like and plant-like Pokemon. Niantic occasionally refers to biomes as “habitats”. (See “Biomes/Habitats” sub-section below for more detail)

Blocked Area:

An area where Pokemon are forced to stop appearing.

Nest:

A preset area on the map that has an increased occurrence of a particular kind of Pokemon. These change species of Pokemon on a two week cycle and also occasionally when in-game events start and end. (See “Nests” sub-section below for more detail)

POI:

Points of Interest - Niantic keeps a user submitted locations and displays these on the Pokemon GO game map as “Pokestops” and “Gyms” - these are used to generate map pins that trainers can travel to and interact with in order to gain virtual items, earn in-game currency and also play a jazzed up version of “King of the Hill”. POIs in Pokemon GO are not sourced from OSM.

Spawnpoint:

A point on the Pokemon GO game map where a Pokemon appears each hour. Spawnpoints are aligned to a level 20 S2 cell grid.

Trainer:

In this context, someone who plays Pokemon GO.

Infrequent Updates:

Niantic has seldom updated their OSM derived database since the launch of Pokemon GO. Additionally, Niantic has repeatedly shown that they are capable of using different versions of OSM data from different times simultaneously when updating distinct game features.

OSM+Pokemon GO Timeline

March 2016 - Pokemon GO Beta Testing Begins.

April 2016 - OSM-like tags are found by data mining the Android application package for the Pokemon GO Beta client.

June 2016 - Nest areas are present in the late stages of the Pokemon GO beta.

July 2016 - Pokemon GO begins to launch around the world using Google Maps to generate its visual game map, but hidden OSM data is observed when it comes to Biomes, Nests and Blocked Areas.

December 2016 - Niantic distributes spawn points along certain pedestrian highway tags such as “highway”=”path” and “highway”=”footway” apparently using data sourced close to when the game launched in July.

January 2017 - Pokemon GO launched in South Korea. For apparent legal reasons, in South Korea Niantic uses OSM data to generate game map visuals, and also adds in OSM copyright/source information in the app’s “About” section. Globally, nesting and blocked areas are updated to OSM data sourced from some date between late December 2016 and early January 2017.

Early/Mid November 2017 - Pokemon GO begins to transition globally to OSM for its visual game map, this rollout happens over several hours. Visual data appears to have been sourced earlier in August. Other OSM features remain unchanged.

Late November 2017 - EX raids are hinted at occurring in parks and green areas, it is quickly realized that these parks are based on OSM data. Later, it is discovered that the OSM data used for EX Eligibility is from July or August 2016 - meaning that EX raids are possibly using the oldest OSM data still in Pokemon GO. As of mid-2019, this data has never been updated.

January 2018 - The Pokemon GO visual map is updated to more current OSM data from earlier in January. As with the previous visual update, other features remain unchanged.

March 2019 - Nests, Blocked areas and spawn points distributed along paths are all updated to OSM data that was seemingly sourced in late February. Map visuals remain unchanged.

How OSM impacts Pokemon GO

Visuals:

As of 2019, the Pokemon GO visual game map is now entirely based on OSM data, though only a limited number of features render at all, and these can be broken down into 4 basic categories:

  1. Dark Green “Park” areas
  2. Blue “Water” areas
  3. Grey “Highway” lines
  4. Semi-Transparent 3D “Building” areas

The following tags are known to visually render in Pokemon GO:

Parks: “leisure”=”golf_course”, “leisure”=”park” and “landuse”=”recreation_ground”

Water: “landuse”=”basin”, “landuse”=”reservoir”, “leisure”=”swimming_pool”, “natural”=”water”, “waterway”=”riverbank” and the ocean (beyond coastline ways) all render. Un-closed water lines such as most streams and drains do not display as anything in Pokemon Go.

Highway: Most highway lines render, highway areas do not. Footways and similar tags tend to render as very thin lines, whereas primary highways are displayed in-game with a slightly thicker weight than residential roads.

Buildings: Any building area tagged with building=* will render, building:part and similar tags are also supported and can have quite a nice effect.

Nests:

A nest is an area in Pokemon GO in which a certain species of Pokemon will appear about 25% of the time from all active spawn points contained within that area. Nests “migrate” or change species every 2 weeks and can also change as a result of in-game events.

There is some evidence that nesting tags may modestly increase the number of spawn points in an area, but there are plenty of nesting areas without any spawn points at all.

The following tags can create nests if they are present on either closed ways or closed relation based areas:

leisure=park, landuse=recreation_ground, leisure=recreation_ground, leisure=pitch, leisure=garden, leisure=golf_course, leisure=playground, landuse=meadow, landuse=grass, landuse=greenfield, natural=scrub, natural=heath, natural=grassland, landuse=farmyard, landuse=vineyard, landuse=farmland, landuse=orchard, natural=plateau, natural=valley and natural=moor

Some nesting Pokemon are highly desirable, however, (in my opinion) Niantic has somewhat backed away from the nesting feature with recent additions to the game, and many newer Pokemon do not appear in nests at this time.

EX Eligibility:

Certain POIs that are known as “gyms” are capable of providing trainers with special invitations to obtain a rare Pokemon after specific interactions. In order for the gym POI to be EX Eligible, it must have the center of the level 20 S2 cell that it occupies centered over a mapped area that meets particular criteria.

EX Eligibility shares the same list of tags as Nesting areas, however, it does not support relations, only closed ways count for EX Eligibility. Additionally, the data used for EX eligibility is old, with eligibility being observed to require the mapped area’s presence on the map in July-August 2016 in order to count.

Blocked Areas:

Certain areas can block spawns, additionally spawns will not occur within the proximity of certain ways. These include - school grounds, military, wetlands, many water tags used for areas as well as basins, some religion-related tags, higher level highway tags, railways, and powerlines.

Biomes/Habitats:

Biomes remain poorly understood, but OSM data certainly has some influence. Tags related to water are, in particular, commonly observed to have a strong correlation to particular biomes. Some trainers have observed lines of water associated Pokemon appearing along underground streams as an example, or have been disappointed when unmapped lakes have yielded no water Pokemon at all.

OSM’s influence on biomes has perhaps become less important over time, as Niantic has taken steps to diversify and add complexity to the biome system through the addition of features such as in-game weather.

Path Spawns

When Pokemon GO launched, Pokemon spawn point distributions were reliant on a global database that Niantic possibly sourced from Google/Android location pings while they were a division of Google. This disproportionately favored dense urban locations, and in response to this problem Niantic made some changes to how spawn points were distributed.

One such change involved the addition of spawns to certain highway tags such as highway=footway and highway=path (and also, oddly, highway=unclassified, but not highway=track). While this was likely intended to bolster spawns along hiking trails and in other low pedestrian activity but still high “value” locations such as parks and open nature reserves, it had the side effect of adding spawns too many sidewalks and similar features as well.

Outside of rural locations, the addition of spawns along paths was modest. Creating paths does not turn a location into a Pokemon paradise by any means, but after being updated in Pokemon GO, they do have the effect of making many previously unplayable locations much more accessible in the game.

Trainer-Mapper Motivations

With the above in mind, we can define a number of archetypes for Pokemon GO inspired mappers

The Bad:

The Self-Serving Vandal:

Their first edit will often involve adding a park or several other nesting features over their house or entire neighborhood. Often this will be accompanied by scribbles of footways and possibly water features. Often you’ll see no more than 6 to 9 edits out of this kind of mapper.

More persistent Self-Serving Vandals may attempt to fabricate evidence and play into OSM’s expected benefit of the doubt/”on the ground” culture.

The Malice Inspired Vandal:

This mapper will add spawn blocking areas or remove nesting areas en masse. They are often inspired by petty rivalries from Pokemon GOm and can be more tricky to deal with, These mappers can be very persistent, with some cases stretching across multiple accounts and hundreds of edits.

The Hyper-Altruist Vandal:

Attempting to help their local gaming community, these mappers are like a Self-Serving Vandal who has had way, way too much caffeine - expect entire residential areas changed to parks, footways everywhere and very messy, truth stretching edits.

The Graffiti Artist:

Not that unusual for OSM, but in an attempt to get their name on the map, this mapper will draw text or pictures using features that specifically render in Pokemon GO

The Good:

The “My Street Wasn’t on the Map” Trainer:

This mapper will add one or two features, maybe with a couple of errors, maybe not. Once their small set of edits are complete, they will vanish never to be seen again.

The Empty Town Trainer:

Often these trainers become mappers because their local areas just simply isn’t on the map. These kinds of mappers are worth a bit of patience as they come to OSM from Pokemon GO with good intentions but often very little experience or guidance.

The Map-Savvy Trainer:

Well informed on how Pokemon GO interacts with OSM. In some ways, this mapper is similar to the Hyper-Altruist Vandal but will attempt to play “within the rules” of OSM. While it’s always worth watching in case they go rogue, these mappers are typically very careful, trustworthy and make many worthy additions to the map.

The Former Trainer:

They may have started as another kind of Trainer-Mapper, but they have started to spend more and more time editing OSM… “and what was Pokemon GO again? I forgot”.

Discussion/Conclusion:

I hope that the above is useful to mappers who are struggling to understand the motivations behind these strange Pokemon GO inspired mappers who have been appearing on OSM over the past few years.

I know that sometimes, their actions may seem frustrating and even confusing, but in my experience, the vast majority are well-intentioned and in many cases, you wouldn’t even know that they came to OSM as a result of Pokemon GO.

My personal journey has taken me from adding a cemetery while Pokemon GO was in beta in an attempt to bring out more ghost type Pokemon, to helping remove vandalism and guide new mappers on an almost daily basis. I personally find both Pokemon GO and OSM very rewarding in their own right, and it’s been quite a journey walking the middle road with both of them.

I’m always happy to answer any questions or write some more on the topic if this is well received (this is my first diary entry so I’m not sure how it’ll turn out). Feel free to comment or shoot me a message at any time.

Comment from Tim Couwelier on 21 June 2019 at 08:22

Nice post. Being both a mapper and player, I can agree with most of this.

I’d like to add that the mapping community has a role to play aswell. Personally, I have an RSS feed for any changeset in a 20 km radius to track what changes. As an active player, usually on username alone, I can spot the Pokemon Go players.

There’s two approaches: * If it’s obvious vandalism, revert and inform the mapper that he’s vandalising a ‘greater good’ project for own benefit. * If it’s Pokemon-inspired, but not inherently incorrect, try to give them a gentle nudge.. stating: okay, I see what you did there, and it’s not a bad start.. but what about the benches? and the trashcans? and you’ve even missed a few paths there…

Paths in particular are relevant: 1) they are also said to help generate spawn points. 2) there’s few usergroups walking around THAT MUCH while looking at a rendered map. They are very much more likely then your average armchair mapper to spot where they are missing.

Also, while the effects of certain tags/landuses is fairly well-documented (the Berlin ‘meganest’ with the plateau tags), there’s also a bit of an urban legend about node density having an effect. If that leads people to add qualitative data to OSM, let’s please not try and debunk that.

Comment from freebeer on 21 June 2019 at 11:06

Feel free to comment or shoot me

and then
and then you
and then you open
and then you open a
and then you open a location
and then you open a location like

https://www.openstreetmap.org/edit?editor=potlatch2&way=469178602#map=20/33.19825/-117.29602

and then
and then the
and then the pain
and then the pain the pain
and then you realise the brain surgery had no effect
the pain the pain omigodomigodomigodomigod the pain the pain
i’ve only been awake 48 hours without a break
my first caffeine in years
it’s only a map, why am i shaking so
maybe i am not seeing right, i close me eyes
eyes closed click

hello? is anyone out there??? is this qa purgatory?
SNAP eyelids stronger than multiple layers gaffer tape carefully applied and welded to skin
mebbe if i stare at it, it will go away. show it who is the dominant canine in this doghouse.
eyes watering. NO BLINKING. the tears come. sobbing, heaving, weeping, wailing, moaning, gnashing of teeth.
I AM AN EDITOR. I AM STRONG. I AM NOT A LOONY.
i can fix this.
i can.
i can.
can i?
the uncertainty. is this art? is this sleep deprivation?
this is a masterpiece.
ii am hearing visions and seeing voices. iinn sstteerreeoo. with reverberberberberb.
all hail the mapper.
learn to love the mapper.
we are all the mapper.
we are here to serve the mapp.
no. i cannot. i close the tab and go somewhere else.
i have failed. i am no longer a mapper.
RIP

Comment from Tim Couwelier on 21 June 2019 at 12:04

Well that’s one way to be overly dramatic…

Granted, some of that is ‘poor mapping’ (naming it ‘walking path’ and not setting the baseball field as a pitch, …) but well.. there’s good intent. How a community responds to such things, helps define how much influx you’ll get to the active editing userbase, and how the osm-community is perceived overall.

While I’d love to have a glass of whatever you’ve been drinking - please tone it down quite a bit if you’re going to reach out to whoever did that.

Comment from joost schouppe on 21 June 2019 at 13:11

As inspirational as way 469178602 was, I couldn’t help myself and fixed it. Sorry.

Also, I thought I was the last person to abandon the Potlatch2 ship.

Comment from freebeer on 21 June 2019 at 15:45

if you’re going to reach out to whoever did that.

i would never contact anyone about so-called poor or allegedly sub-standard mapping. there are plenty of other socially-gifted mappers who have chosen to fill that role on occasion. when i fix mistakes like this (aligning to best imagery, fine-tuning, de-overlapping or joining ways at present) i do so quietly unless i am trying to debug a systematic error. this is a user skill/knowledge issue that one cannot expect of beginning independent pokemon mappers and no reason to attribute it to any one person.

i only „reach out“ (what an awful phrase, sounds like an unwanted physical approach) in cases of technical issues (as in https://www.openstreetmap.org/changeset/71385918 ) or vandalism (such as https://www.openstreetmap.org/changeset/68080542 )

I couldn’t help myself and fixed it. Sorry.

sorta. you depart from the ramp earlier than mapbox shows it meeting the pavement, and i hope the pokemon players who take your directions literally hold you personally responsible for their injuries, immersed in their vr like lemmings.

i expected something like that :-P luckily or not, i have plenty of other examples up my sleeve. traces i spend more time trying to understand than it takes to re-draw. with this thread, we might be able to clear up california… (how easy to get mappers to carry out my unspoken bidding, muahaha)

https://www.openstreetmap.org/edit?editor=potlatch&way=469560542#map=19/33.98787/-118.43639

the usually best mapbox imagery hides too much reality here.

Also, I thought I was the last person to abandon the Potlatch2 ship.

in early 2019, there were 1’673 users, down from 6’451 for all of 2018. however, i have changed the above url to reflect the status of potlatch 1.4 as my default editor, the number of users of which would not even fit into my flat, uncomfortably or otherwise, for a potlatch party neither for early 2019, nor the 233 in all of 2018.

i need to get a bigger flat.

Comment from Horza on 1 July 2019 at 14:47

Seems I should be adding more sidewalks, which probably would actually be useful in my neighbourhood as they aren’t especially common in some areas around where I map.

Comment from Tim Couwelier on 1 July 2019 at 14:55

Use common sense with sidewalks though: if they are just part of the road, don’t map them seperately if the map is ‘basic’.

A sidewalk that has no seperation from the road other then a gutter/curbstone combo shouldn’t be mapped seperately in my opinion, as it’s needlessly cluttering the map.I’d avoid it, unless the entire area is highly micro-mapped.

Comment from Horza on 1 July 2019 at 21:39

That’s why I’ve been holding off on it. I figured I should do none of it or all of it and there are some areas that are kind of borderline where some sections of the sidewalk are directly on the road and some are separated by a yard or so of grass.

Never assume common sense is common XD

Comment from WoodWoseWulf on 1 July 2019 at 23:56

there are some areas that are kind of borderline where some sections of the sidewalk are directly on the road and some are separated by a yard or so of grass.

My local mapping area is quite similar. Footways here can join with the road as a sidewalk for a short distance, then veer away across a wide green space only to turn back and cross to the other side of the street before traveling up a gap between two buildings and popping out on an entirely different road.

The pedestrian “street” experience is quite different from the vehicle one in those kinds of cases. It can be difficult to determine when exactly you should swap from a “footway as a separate way” to “sidewalk as a property of the road” schemes of tagging, especially when the latter might only be used sparingly due to the complexity of the footway “network”.

As a side note though, the accurate mapping of pedestrian paths in larger parks is something that few would discourage. Quite often, features in parks can be obscured by trees or other features in satellite imagery, and Pokemon GO/Ingress/Harry Potter WU players are uniquely positioned to run GPS traces and survey these kinds of places while they play. If done right, games could benefit OSM, which could, in turn, could benefit the games in the future. There has been a number of cases where I’ve found parks drawn over houses quite close to unmapped (real) parkland… I wish there were more ways to encourage certain game inspired mappers to think a little less selfishly.

Comment from LivingWithDragons on 2 July 2019 at 12:53

This is a good an informative post, thank you for collecting the data.

I’m interested in some of the points (In Wizards Unite they are “inns”) that have photos. What is this data source? Could it be from the Ingress game that Niantic have? I remember at one point (1-2 years ago) they added more. They get as outdated as the base map, but are more important - such as one is a Post Office that has closed.

Comment from Tim Couwelier on 2 July 2019 at 13:30

@LivingWithDragons :

Niantic has a ‘POI’-database they’ve collected over the years, and both Ingress / Pokemon Go / Wizards Unite draw from that database.

The data for this database is crowd-sourced: * Ingress players with a level of 10 or higher, get the option to nominate ‘portals’, at a limited rate of 14 every two weeks. * Every proprosed portal goes through a review-process, called ‘OPR’ (Operation Portal Recon, iirc). Players who’ve reached level 12 or up in Ingress get to review the proposals.

Portals that get accepted though OPR are added to Ingress, and synced to Pokemon Go a day after (at a fixed time), if the portal falls within the rules for pokestop creation. The process for this relies on the use of S2-cells. If there’s no existing pokestop/gym in a lvl 17 s2-cell around it, it’s eligible to come over, if there is already an object in it, it doesn’t come over. Based on how many stops there are within a lvl 14 s2-cell, one of the pokestops will be converted into a ‘gym’. Upvotes in Ingress can help determine which.

Now you’ve realised that Ingress and HP:WU have different ‘rules’ to determine which things are shown in-game. But the photograph, title and possibly additional info are always sourced by Ingress-players…

UNLESS… it’s in one of the countries where (possibly due to a lack of portal density/ingress players) the portal nomination is opened up to Pokemon Go-players aswell. Mostly south-american countries are in this case, ( https://www.reddit.com/r/TheSilphRoad/comments/b79e62/current_map_of_countries_that_can_submit/ for reference).The pokestop submissions are treated similarly through OPR, relying on Ingress players to judge it was ‘worthy’ portals or not.

Comment from Tim Couwelier on 2 July 2019 at 13:45

Part two of the reaction:

You claimed ‘they added more’ -> must’ve been an Ingress player that pushed for a few extra portals. They get outdated -> Yes, the world changes. Ingress players can also opt to nominate portals for removal, or to update a picture for them, or move them slightly. Pokemon trainers (or folks playing both) tend to be less prone to ‘remove’ them (as it can negatively affect their gameplay), but every now and then some disappear. There’s been a tendency to remove any portal that’s in the center of a roundabout for example, as it’s pretty much a guideline the stops needs to be accessible to pedestrians. Over 3 years, we’ve lost two gyms in my city, and a few pokestops. On the other hand, I must admit we still have a few (both stops/gyms) for things no longer there.

As far as the basemap getting outdated: Niantic isn’t exactly keeping up nicely with the tiles they generated as in-game basemap… to avoid having to regenerate tiles for the entire world. The visual basedmap is (I think) based on december 2017 data. The difference in my city would be fairly significant in some spots by now.

But the slower update cycle makes it at least appear that ‘vandalism for own benefit’ has a minimal chance of actually being able to kick into effect.

Comment from Horza on 2 July 2019 at 13:49

A lot of the time the stop that becomes a gym is the oldest one in the cell, which is annoying in my area as that tends to be a church.

Currently, there is little incentive for players to remove inaccurate POIs because less POIs is nearly always a bad thing for them. The only ones where are incentivised to remove them is if they’re on private property/hard to reach areas, especially if that one is a gym, but when I made one removal request Niantic asked for proof I was the landowner (and they have no-trespass signs I could submit as alternate evidence)

Comment from Tim Couwelier on 2 July 2019 at 15:02

The deciding factor appears to be ‘whichever portal has received most upvotes in Ingress’. Older portals have of course had more time to accumulate these, and especially if it’s a ‘landmark’ thing. But that said, for a majority of cases around here (63k population city) it takes a very limited number of upvotes to ‘steer’ the choice towards preference.

There’s indeed not much motivation for Ingress folks to vote to remove existing portals, but it does happen, unless that affects either their in-game medals (need some achievements in those along with AP to be able to level up), or they have a grudge against the Pokemon Go players (not entirely uncommon).

Login to leave a comment