Category Archives: English

English OpenStreetMap

Remote mapping – lets do more

Recent postings by Erica Hagen and Gwilym Eades about the colonialism of mapping places remotely have led to some discussion about OpenStreetMap going local only – eschewing remote mapping in all but very rare cases. I find this view of this incredible global project to be in error, furthermore I am creating tools that will make it even easier to find remote places to map – hopefully increasing the number of remote mappers and the amount of remote mapping work done.

 

Colonialism

I find the premise of these articles to be incredibly narrow, focusing on the emergency response aspect of the map and also somehow managing to make a polygon that represents a building into an imperial construct of oppression.

Now for those who belong to nations that were or are empires it is probably ingrained to be wary of becoming too imperial. For those of us that belong to former colonies no such mental barrier exists, we have been where many nations are currently, they are mere decades behind us in infrastructure and not centuries, as they are in comparison to the empires of past and present. We know the hunger that drives us to get better as quickly as possible, to build infrastructure that aids in the development of our nations and we had no or few adversaries then that tried to hold us back for cultural reasons, fearing we were overextending ourselves and losing ourselves in the process.

To be able to assist others to go further faster, that is something we would have liked to have (and did indeed sometimes receive such help) and so we are puzzled to be accused of imperial land grabs for our efforts.

The articles in question focus on the humanitarian mapping that has been in the news and manages somehow to present what was a boon to people in the field, both locals and those that arrived to lend aid, as an affront to locals and their culture.

The HOSM efforts and missing maps will end up, instead, homogenising that same map forever, westernising it, colonising it, and in effect coopting the last vestiges of autonomy in its creation that remain(ed).
Gwilym Eades

How do you manage to colonise a country by displaying its basic infrastructure on a map, a map that makes it easier for locals to get involved and add better local data? How can a building in Botswana be westernized when it is a polygon with the same values as a building in any other continent, if a local adds the detail of its material as mud or straw does it then become Africanized and no longer a colonial subject?

This premise of the articles is indeed the most imperialistic aspect. The self-expression mentioned sounds like any local is a noble savage, and the remote mappers are the decadent civilization encroaching on their territory.

 

The blank slate

One argument put forward is that by not presenting locals with a blank slate of their home we rob them of their self expression. That is a ridiculous notion as the map can be edited by anyone at a later date, it is also a total opposite of how most people interact with the digital world.

Through my work on OpenStreetMap I have been in contact with locals in many places, both in my native land and abroad. Presenting them with a blank map of their home and telling them they can map it now, using their “own expression”, has not given much result, the percentage of people that see a blank slate that then find the appropriate tool and read up on documentation and experiment is low, evidenced by the low number of original OSM mappers.

The best results in getting locals to contribute has been to map their area remotely and then get them to contribute data, it is easy for most people to see their village on a map and point to where the tavern, healthcare and shop are, they are not creating anything from scratch but adding details to a slate that makes it easier for them to lend their knowledge. By mapping remotely the map is given a leg up.

Some examples are Khawa in Botswana and Eilao in Spain, in both cases I was in contact with a local, showed them that it would be easy to map, got interest but no effort from them due to a technical barrier, mapped the place remotely and showed them again – then receiving feedback that allowed me to add details to the map.

 

Remote mapping works

The success stories from Liberia and Nepal and past global responses have been many, less known are the umpteen successes that occur every day with the aid of remote mappers on OpenStreetMap in conjunction with local people, foreign and native.

The Peace Corps have thousands of volunteers all over the world and they use OpenStreetMap and remote working to aid in their work as they aid in healthcare for example. An example is the Stomp Out Malaria program, one report detailed how a village in Botswana’s Chobe District was visited by volunteers who checked which houses had been sprayed by using a GPS device and handheld records, then using a base map from OSM to see which places had not yet been sprayed and then with this knowledge they managed to make a plan to revisit places to spray, thereby going from 61,4% coverage to obtaining the 80% threshold that is the minimum recommended coverage.

The base map had been created by remote mappers just days before the visit of the volunteers. This added procedure of mapping was found to be cost-effective and more vitally, enhance the odds of success in combating malaria, a disease that kills hundreds of thousands of people every year, mostly in sub-Saharan Africa and overwhelmingly young children. Just now task #1057 is finishing, Mapping Mosetse, with several benefits for the locals in aiding the locals.

 

The spirit of OpenStreetMap

More poetically perhaps, I also see maps as representations of who made them rather than a place per se.
Erica Hagen

I’m curious as to how remote mapping interferes in the expression of locals. For a map to be global some minimum standardization is required, we use a polygon in the outline of the building to mark it on the map, or a point. We put addresses as attributes of the polygons or as a seperate point inside, or without, a polygon. We do not generally mark addresses as a line between two points although that is possible.

The spirit of OpenStreetMap is not in my mind to stay in your own garden, to mind your own business and not aid in other areas. No project I have participated in has had that expectation of its volunteers, amongst them are Distributed Proofreaders, Project Gutenberg and various Wikimedia projects in a variety of languages.

The spirit of OpenStreetMap is to make a map of the world, including those places that are not commercially viable for other providers and whose population lacks the technology and monetary power to do it themselves. No village left behind.

 

The mobile world

The number of people that are reaching middle class status in the developing world has been growing and will grow further still, in tens of millions in the coming years. Most of these go mobile and mobile is where editing OpenStreetMap is very much lacking still – not to mention that data charges are still very expensive and lacking in distribution in the developing world, even the developed world.

Try drawing the outlines of buildings on a mobile editor, it will take you an order of magnitude longer to map a village on it rather than using a more powerful desktop editor. Adding smaller items is easier and less of a pain on mobile editors, being able to add points of interest is much more feasible. Totally eschewing remote mapping and telling locals they need to get by on their current hardware using their current connectivity is another example of imperialism in action – leaving the noble savages to fend for themselves.

 

Remote mapping made easier

I myself did what most others do when they discover OpenStreetMap, I first looked at my local area and improved it. Currently it is pretty much in maintenance mode only, adding or moving POI and on the lookout for buildings being built or torn down. Suddenly my hobby needed more outlets and so I expanded, and expanded more and now I’ve touched the map in dozens of countries.

That was all very random though, zooming in randomly onto the map to find something that was missing, blank slates often and adding something to them.

So I’m building a tool, Askja, that makes it easier to improve data where it is lacking. I split it up into two areas, remote and local. Remote mapping is traditional, roads and buildings – the base map. Local mapping is needed for addresses and street names, amenities, paths and Mapillary images.

For each country I add into the tool Overpass is used to get a list of all cities, towns, villages and hamlets. Those are then assigned to regions and subregions using Nominatim. Finally the quality of the current data is evaluated using Overpass and Mapillary APIs, the bot that runs the evaluation checks for roads, buildings, amenities, paths and Mapillary images and marks a scorecard for each settlement.

scorecard
Scorecard example

The scorecard has 4 different values generally, grey for unknown, red for absence, yellow for partial coverage and green for good coverage. The bot can upgrade network to green status and the others to yellow status automatically. To upgrade anything to green a human is needed to look the area over and pass its judgement. Imagery needs to be evaluated by humans at this point, perhaps that can be automated as well.

Some examples:

So how does it help us to have a list of places that are lacking data – or have data? It makes it easier for the community to find weak spots in their area of interest, whether it is nearby them or not.

The tool is in alpha status – many more features will be added to make it easier for new mappers to find suitable tasks and for niche mappers to find tasks that interest them. We have niche mappers that only like to do the road network, others prefer to map the streets of a village and yet others just want to add buildings or forests or trees.

In addition this approach makes it easier to find quality assurance problems, triplicate settlement nodes (imports gone wrong), settlements that lack names, similarly named settlements within a few hundred meters from each other and more. Not to mention places that are marked as villages but are a tourist campsite, a tavern or otherwise.

Good housekeeping is a part of making a map. Approaching the task of making a world map from many different angles makes the map stronger, it makes the map better.

 

Everyone needs a map

Today the Western world is used to maps being easy to come by in digital formats, they listen to their smart devices as they use location services from OSM, HERE, Google, TomTom and other providers. They are the base that companies like Uber and FedEx depend on.

The growth potential for communities who are lacking these maps is phenomenal, to withhold that power from huge areas of the world is colonialism, aiding them isn’t.

In addition the locals are not the only people who will need to use a map. Tourists, passers-by, people moving their goods or offering their services also need maps of places where they go. Everyone needs a map even if not everyone wants to make it or use it.

English Leikir

Cities:Skylines – the public transport system

My latest gaming fun is had in Cities:Skylines, by Colossal Order and published by Paradox Interactive. A very nicely done city simulation game where you build the infrastructure (roads, power, water) and amenities (school, police etc) and assign land use – hoping that residents will move in, build their business and enjoy what you have created for them.

Fantastic game

The game is fantastic, the support of 3rd party mods and assets is great and overall I highly recommend it.

Traffic Problems

Once my town hit 70.000 inhabitants I became fixated on fixing the mounting traffic problems, so I started to look at my public transport system in more detail. Noticing hundreds of people queuing up at some bus stops seemed like an impossible task to decipher, looking at random samples of the people to see where they were headed seemed inefficient. So when I learnt that I could assign different colors to different bus lines (and rename them) and see how the crowd split between those bus lines, by the color in transport mode, it was a game changer.

I started to create hubs, connected to other hubs with express buses or metro lines. At each hub there would also be local buses, trying to spread the load and make sure no stop had dozens of people that never got to take a bus (and despawned back to their homes after some time). Sometimes I’ve had to create a complementary line to aid a congested one, particularly where big tower flats are numerous.

cities-skyline-bus-system

I’ve by now spent hours on fine tuning the system to eliminate gridlocks, both in general traffic and at bus stations and metro stations. Things seem to flow nicely but I’m still missing several things that would make my life easier!

 

1. Bus stop information

I want to be able to name a bus stop and see which lines use it, how many people exit there and enter on average per bus. From this information screen I would like to be able to delete the stop from a line, using the current interface to do this is very hard as if several lines use the same stop there is no way to pick the correct one. Currently metro stations can be renamed but bus stops not – a generic Bus Stop 1 suffices for default values. My mockup is thus.

cities-skyline-bus-stop

2. Bus line information (and metro)

I want to be able to view the bus line (as is today) but with the stops added and there see average number of entries and exits (and maybe number of people left behind). Similar format for metro lines, although most of mine only go between 2 stops.

cities-skyline-busline

3. List of all bus lines, metro lines, metro stations, bus stations

A list of all metro stations for example and what lines they contain. I now have about 30 metro stations and 60 metro lines. Easy to lose track.

4. Name streets

Whilst I would probably not name every little residential street, for the main arteries of my transport system I would like to be able to add names, to aid the in the naming of bus stops, bus lines and metro stations. Again a simple Street 1 as default would suffice.

5. One other thing – the power grid fragmentation

I have an ample power grid now that I’ve got a nuclear plant in addition to the solar and wind gadgets. But a message informing me the power grid had fragmented was of a limited help as I could not see how much power each separate power grid had nor could I color code them to differentiate between them. This would be a welcome help because I’m pretty sure I’m overspending on electricity in some areas.

English Leikir

Map issues in World of Tanks 0.8.3

I’m an avid player of the MMO World of Tanks where I roam as Hagbardur and have kept a close eye on how the map changes have affected the matches. Following the tremendous work done by Phalynx, where he collects data from players on matches and makes them available on his website, I managed to get together some statistics on team performances in the maps because quite frankly there were some seriously flawed maps.

It was good timing as the day after I finished compiling the percentage chances the developer Overlord posted a topic on maps in WOT. Since my findings could use some visual aid I’ve decided to present them here.

All numbers taken from Phalynxs website with Tanks with over 100 battles included. I’m sure Overlord has access to much accurate numbers but until Wargaming releases them this is the best approximation we can get.

 

The Percentages

Here I list, in order of worst skewed maps to least skewed, the maps, the number of times Team 1 (green) won and Team 2 (red) won. You can see at the bottom of the post which team is Team 1 Green and Team 2 Red. The skew number is the absolute percentage difference between the two teams. I also highlight maps which are prone to draws.

 

Map Team 1 Team 2 Draw Skew
Westfield (assault) 66,00% 34,00% 0,00% 64,00%
Fjords 57,70% 41,60% 0,70% 32,43%
Malinovka (assault) 58,00% 42,00% 0,00% 32,00%
Sand River (assault) 57,50% 42,50% 0,00% 30,00%
Dragon Ridge 56,70% 42,20% 1,10% 29,32%
Province 54,70% 42,00% 3,30% 26,27%
Mines 55,20% 44,40% 0,40% 21,69%
Arctic Region 44,00% 54,50% 1,50% 21,32%
Highway 54,50% 44,60% 0,90% 19,98%
Siegfried Line (assault) 45,00% 54,60% 0,40% 19,28%
Murovanka 45,00% 54,40% 0,60% 18,91%
El Halluf 52,10% 43,60% 4,30% 17,76%
Prokhorovka 45,90% 53,80% 0,30% 15,85%
Serene Coast 49,40% 42,70% 7,90% 14,55%
Ruinberg 46,00% 53,10% 0,90% 14,33%
Erlenberg (assault) 46,50% 53,50% 0,00% 14,00%
Ruinberg (encounter) 47,00% 53,00% 0,00% 12,00%
Malinovka 50,40% 44,70% 4,90% 11,99%
Fisherman’s Bay 52,60% 46,70% 0,70% 11,88%
Widepark 47,00% 52,50% 0,50% 11,06%
Karelia (assault) 52,60% 47,20% 0,20% 10,82%
Steppes 47,20% 52,40% 0,40% 10,44%
Redshire 45,70% 50,40% 3,90% 9,78%
Erlenberg 46,80% 51,10% 2,10% 8,78%
Live Oaks 51,60% 47,40% 1,00% 8,48%
Himmelsdorf (encounter) 47,90% 52,10% 0,00% 8,40%
Sand River (encounter) 47,90% 51,90% 0,20% 8,02%
Karelia 47,70% 51,40% 0,90% 7,47%
Murovanka (encounter) 48,30% 51,60% 0,10% 6,61%
Airfield 51,10% 48,10% 0,80% 6,05%
Steppes (encounter) 51,40% 48,60% 0,00% 5,60%
Himmelsdorf 51,00% 48,30% 0,70% 5,44%
Port 50,50% 48,50% 1,00% 4,04%
Lakeville 48,70% 50,30% 1,00% 3,23%
Abbey 50,20% 48,70% 1,10% 3,03%
Ensk 49,30% 50,70% 0,00% 2,80%
Ensk (encounter) 49,40% 50,50% 0,10% 2,20%
Siegfried Line (encounter) 49,50% 50,40% 0,10% 1,80%
South Coast 49,60% 48,80% 1,60% 1,63%
Cliff 49,90% 49,40% 0,70% 1,01%
Siegfried Line 50,00% 49,50% 0,50% 1,01%
Sand River 49,50% 49,20% 1,30% 0,61%
Malinovka (encounter) 49,10% 48,90% 2,00% 0,41%
Westfield 49,80% 49,60% 0,60% 0,40%
Mountain Pass 49,30% 49,20% 1,50% 0,20%
El Halluf (encounter) 48,70% 48,70% 2,60% 0,00%

 

The Worst Performers

Most skewed map
Westfield – Assault

It is no great shock that Westfield – Assault mode is the most skewed map, the defenders have ample sniping spots in the lower right corner and on the other side of the valley very well protected spotting locations that can also shoot at anyone travelling the edges of the map. With the defenders (Green on this map) winning 66% of the battles this needs a major re-balance.

 

Fjords
Fjords

Fjords is skewed towards the Green team which have a 32% edge on the Red. Their protected sniping spots both in the north and south that cover the field in the north no doubt are a big factor for this as well as the natural defense positions they get in the middle and south roads.

 

Malinovka - Assault
Malinovka – Assault

Malinovka in assault mode is skewed 32% towards the Green defenders who not only enjoy higher ground but also can get to ample forest cover faster than the Red attackers. Playing this map as a T95 on the Red attacking team is nothing short of the biggest joke played on players as your lack of speed and lack of cover while attacking mean that if you are not killed by artillery then you are too slow to make a difference in the attack. Other slow tanks on the attacking side suffer heavily from this as well with the Physics update punishing their uphill climbing. This needs better balance and this is why I have personally disabled Assault and Encounter modes on my account.

 

Sand River - Assault
Sand River – Assault

The Green defenders get a 30% advantage against the Red attackers. Crucially the Green team can snipe any Reds trying to go north so those that do make it to the north are badly beaten and pushing them away is quite easy usually. A fast medium rush from Reds to the north can combat this but this is a huge weak spot for the Reds.

 

Dragon Ridge
Dragon Ridge

Dragon Ridge is skewed heavily towards the Green team in the south who are 29% more likely to win each game. First of all they get, from their own spawn, a superb sniping spot to shoot any tanks in the Red team moving from their spawn towards the town in the middle. Secondly from the mountain in front of their spawn they get good sniping lines on the town in the middle, something that the Red team do not get from their own mountain safety. Thirdly the map funnels all the play into the gorge in the east, where Green team have better sniping lines into it both from their spawn and the path in the middle, and into the town where Green are again better positioned. This map is thankfully being removed in 0.8.4 and hopefully will come out of it not only more enjoyable to play but more equal for both teams.

 

Province
Province

Province is a special case as it is mostly played in tier III and lower these days. Here the Green team that start on the left side of the map enjoy 26% more wins than the Red team on the right side of the map.

 

Serene Coast
Serene Coast

Special mention must go to Serene Coast which is undisputed king of Draws, 8% is an incredibly high number but not surprising for those that have played it. Green have a 15% advantage over the Red team, no doubt due to the sniping hill and the B1 corner for spotting. It will be removed in 0.8.4 and hopefully adjusted to make gameplay more smooth.

 

The Maps

Mouse-over to see the advantages each side has.