OpenStreetMap

DeBigC's Diary

Recent diary entries

The view from the summit of 200,000

Posted by DeBigC on 28 August 2021 in English (English).

The helpful webstats that are provided by Pascal Neis measure map contributions per sovereign territory. In Ireland we still have a relatively small community, and would seldom have over 400 mappers appearing in his ‘kinda random’ 60 day view of the mapping contributions. It would also be rare to have over 100 map changes recorded by more than 100 of those who make it on to the chart. The pascal records these excludes Northern Ireland, who are part of our community, and where our active mappers also map a good deal.

Anyway, a few weeks ago myself and my fellow Traveller, AK challenged each other to get to the 200K mark. AK has been to that summit before, and me never, though it wouldn’t be uncommon for me to have 100K map changes. I am so thankful to have her comradeship and support. It became less of a race, and more of us egging one another on to map every day for the 60 days, which we duely did, and here is the proof. 200K

What I have learned from this is that mapping is too often an isolated activity. Someone mapping alongside you is a lot more challenging, a lot more fun and certainly leads to more map completeness. The only difference in experience in mapping between myself and AK was she mapped Laois and Offaly, which are mainly rural, whereas I mapped urban North Dublin. I managed to make good contact with 2 or 3 mappers around the city who I didn’t know existed, but we ended up chatting over changesets.

What next? A rest, but back to my more customary 50-100K changes every 60 days. That would be Neis indeed.

Triskaidekaphobia in Dublin

Posted by DeBigC on 23 August 2021 in English (English). Last updated on 24 August 2021.

I map a lot of buildings in northern Dublin City. I also interrogate housenumber interpolations, view mapillary for little peeks of the housenumbers that aren’t blurred, and do my own StreetComplete captures to add as many address details as possible. Lately I have noticed something, born from a concern that I was doing something wrong, or making a mistake.

In a few housing developments around Dublin housenumber=13 doesn’t exist. I have heard of such indulgences before, for example some airlines dropping row 13 from seating plans, or the Irish car registration authority famously dropping its numeric system in 2013 to avoid the unluckiest of car numbers.

So out of curiousity I hit OverPass Turbo and launched a series of queries like this into the existing housenumbers. I couldn’t do loads of these so I thought that maybe 19 would be a sufficient sample (42,000 building objects) to make comparisons and establish patterns. snip1

I plotted these values and observed, as expected, that number 1 is most frequent, since short streets and long streets all seem to have a #1. The numbers after #3 have a simple arithmetic decline. An unexpected result was that there are a few more #3s than #2s, but I will return to that in a comment if anyone wants to ask me :)…. Back to #13! There were around 60 less than I expected. This means #13 doesn’t appear so much as it should, even taking the decline in numbers and the finiteness of space for houses to be built into account.

3Capture I took this a stage further. I used an R-square test to establish if the quicker decline from #12 to #13, and slower decline from #13 to #14 are significant, in other words - is there something else going on with this blip in the decline pattern other than chance. And with an R-square value of .99 it seems there is indeed something going on!

All joking aside the lack of #13 may not be down to superstition, and may rest on a dominant architectural tendency to place an even number of houses on a site. It could be that mappers are superstitious, though I doubt it. So while this itself is not proof I give you Ashbook housing development in Dublin, where #12 and #14 are present but #13 is not.

Good Luck!

PS talk about this on twitter

The State of the Map Conference 2021: feedback

Posted by DeBigC on 13 July 2021 in English (English).

My intention

It is what it is. This is some feedback for the organizers, speakers and community at large. I’m just picking out my main impressions and experiences. I sat through about 80% of it except for a couple of the Sunday things. The conference, being the second remote one had a dead atmosphere about it, even judged by the remoteness yard-stick. I have always been remote, for various reasons, and I enjoyed Japan, Milan, Heidelberg, and even the shell of Cape Town last year from afar. In those years small glitches occured, but despite these things did improve and did get better with the passage of time.

My intention is to give positive feedback, without being a cheerleader because cheerleading is just noise, not specific and not credible. Positive feedback is honest, and helpful with its honesty.

The content

I think the speakers, their topics, the content decisions, the preparation, the enthusiasm and the concern for the rest of the community and the project shown by the speakers was exemplary. As a viewer of several SOTMs I can certainly see interests in the community maturing, specializing and upping its game in terms of evidence behind ideas, and explanation of these ideas, and calls for engagement. I don’t dare to pick out my favourites as I have 4 friends who presented, but there were talks that were surprisingly interesting, beyond how they were billed. This needs to be preserved, firstly by naming it for what it is, and maybe the process of selecting and focusing talks is working well.

The format

The format seemed to be stretched out to the point of being energy sapping with all the long downtimes between segments. The talks themselves, if they were related to one another to begin with(?), are more difficult to relate to one another with those gaps. This was not hard to get right if the talks were pre-recorded. I can’t see any logical argument for the variety of audience time-zones creating a need to spread the whole event out over half a day for very few talks, and very short actual talking time. Big suggestion: if there have to be gaps put a countdown timer to the start of the next one. No more fiddling around with UTC conversions.

The Platform

I was late booking, and didn’t get a ticket. As a consequence I missed out on the main viewing platform, and found asking questions impossible. Why were tickets even necessary? Why choose a web space or application that is limited technically to such a ridiculous extent that it has to have a ticket system. Some of the moderators/convenors were facing dead air for engagement and resorted to shouting for twitter questions, which were studiously ignored because the moderators weren’t communicating with the twitter department. Doesn’t it go against the grain of the history/identity of the community to force people to have tickets, or use mastodon (no thanks) to be able to meaningfully engage? I think it obviously does. If you decided this, and you are reading my reaction please reflect upon this as a lost opportunity. Sadly some talks got no questions, and there was little engagement evident for those of us without tickets.

Conclusion

In balance the interesting talk content and speakers outshine the weaker points, and I hope that comes through here. If there was a big thing to suggest about remote conferences I would say to the organisers at the start “make it feel like a conference”. Hopefully the next SOTM is face to face, but even if it is that will only be an experience for a few hundred people, while thousands of highly committed mappers will still depend on the SOTM organisers getting the balance right, and should not be excluded from viewing all the talks, and all the questions and comments.

StreetComplete in Ireland: part 2

Posted by DeBigC on 5 July 2021 in English (English).

This diary post is a follow up on the one I did last week here where I was looking at how much, and for what the Ireland OpenStreetMap community uses the StreetComplete(SC) app.

The StreetComplete creator Tobias replied showing an interesting worldwide trend for SC usage. The trend here shows that there has been an upturn in SC usage worldwide in the last couple of years in terms of the users and edits contributed:

changesets.jpg

In particular Tobias shows that co-incidental with the Covid-19 months that there was a big upturn in the changesets contributed, starting gradually in May 2020 with a doubling on the previous month, but continuing to rise until the changesets remained at about 4 times the normal level for the forthcoming year. Most users are in the Northern hemisphere, so to see the higher level of changesets sustained through the Northern Hemisphere’s winter months is impressive.

In Ireland the changes, as in the big upturn are also reflected in a little bit of analysis done here.

changesets-IRL.jpg

While this does also reflect a rise of an equivalent magnitude given the lower base number of users in Ireland, it was delayed until December 2020, and lasted only for 3 months. The number of changesets being contributed settled down to around 4000 per month.

As I said in the first post we have the summer, our community in Ireland could be making much more of this app. Tagging “depth” is uneven in a lot of places, and StreetComplete is a great leveller. A lot of building=yes tags still exist, and if there was nothing but that to do it would still justify us taking this app seriously. There are missing buildings, (we are no more than 25% complete), and the vast majority of buildings do not have addresses still in Ireland. Loads of potential for usage of SC.

Review of StreetComplete mobile app usage in the Ireland osm community

Posted by DeBigC on 29 June 2021 in English (English).

StreetComplete was designed by Tobias Zwick and launched in 2016 as a handy “low-bar entry” field capturing application for OpenStreetMap. In the following 2-3 years it rapidly rose to have a share of 3.9% of contributors favoured application, since it requires no real prior expertise in OpenStreetMap. The beauty of it is that it exploits the opportunity of a mapper being in situ where some features have been mapped, but lack details about their attributes, and tags on the OpenStreetMap database.

The UX is good, even for the experienced mapper it masterfully detects the absence {not to jump down a rabbit hole, but very like my favourite Pink Floyd Album } of tagging details, and then it supports the insertion of the correct tags with menu systems that are visual rather than everything being list driven. Perfect for being on the move, a car passenger or a leisurely walker. This is not supposed to be a review of the app, but you can probably detect that I am a fan.

With some help from my friend Amanda who downloaded the .pbf, and extracted all the editor traces of StreetComplete here I was able to get hold of a csv of all of the Ireland edits and then use that to make a heatmap, with some help from RustyB. The heatmap shows, what we would expect, greater use of the app in Ireland’s cities and large towns.

heatmap.jpg

However, my expectations of the number of StreetComplete contributors, and the number of edits contributed in total were higher, and it was disappointing to learn that after five years was only 108,500 edits, and this done by 105 different contributors meaning an average of only 1,000 each. On the other hand, and typical of OpenStreetMap contribution statistics, there is a “long-tail” of contributions, where the vast majority are small contributions and 80% of the work is done by the top 10 users.

User Edits
cart0 26878
ag-nius 11627
DaCor 10555
DeBigC 8612
b-unicycling 7628
Florent Ch 4024
tshedy 3531
kenguest 3105
Dr Breandán Anraoi MacGabhann 1627
desfitzgerald 1428

Being very involved in the #osmIRL_buildings project I found that the app works in the perfect context to help tag newly added buildings, where we are still using satellite imagery to capture buildings it is unavoidable that some high percentage of them will be building=yes. Looking deeper into into the extracted data is was interesting to discover that the objectives of #osmIRL_buildings were relevant to what a lot of StreetComplete activity was focused upon, with the following findings:

  • 23,221 edits related to the “building” value key
  • 24,773 edits were to define a roof type, or levels within the roof
  • 15,163 edits were to add the building levels
  • 6,887 edits related to adding the address in either house number or housename
  • 3,559 edits related to the addition of a placename or streetname to a building address

Overall I think StreetComplete is a useful and very handy tool, but I do think the Ireland OpenStreetMap community could embrace it more. In particular there are 700,000 or more building=yes tags on the island, and this challenge needs to be addressed to get towards a useful and scalable building dataset. More StreetComplete surveyors doing more walks and cycles, and contributions into the black fog of absent attributes is needed in the future.

The evolution of adding buildings in Ireland

Posted by DeBigC on 31 March 2021 in English (English). Last updated on 2 April 2021.

The tag #osmIRL_buildings is a kind of handy brand name our community has developed for our mapping campaign to map all of Ireland’s buildings. The campaign started with a brief test at a single mapathon in Galway in September 2019, and kicked off in earnest in Kilkenny at the end of November the same year. I say “in earnest” because this is when tasks were opened and the tasking manager started to be used every day.

The campaign has a number of effects which merit discussion beyond just the obvious upturn in mapping. But let’s mention that upturn first. The upturn started at the same time as the task manager being opened at the end of 2019. The number of buildings went from 810k to 1.5million in those 15 months, which means 47 thousand buildings were added each month.

buildings
picnic images

At the outset there was a big concern that the building=yes tag would be invited by dint of using the hotosm tasking manager source code. However, this was completely unfounded as the building=yes contributions have remained static (in fact it grew by 20k but this is a miniscule increase compared to the surge in building=yes tags being added from 2016..

Finally, tags like building=garage, building=shed and building=farm_auxiliary have started to grow only a few hundred to having approximately 60k instances of each. This shows that contributors, once their mindset is focussed on buildings are very willing (and able) to define and capture what can often be passed off as insignificant.

Long may our success continue!

osmIRL_buildings merch

Posted by DeBigC on 15 March 2021 in English (English).

As a Director of OSM Ireland (osmIRL) chapter I occasionally get to do nice jobs for the community. Our chapter received a microgrant from the OpenStreetMap Foundation which we are using in conjunction with the agreed community task to map all 5+ million buildings on the island. With the support of this microgrant the chapter deployed the concept of rewarding mapping landmarks with a small token of appreciation for reaching the landmark of 10,000 buildings.

In this case we are giving our contributors a handy mouse mat, which is made with foam and nylon and carries a nifty design. Bias declaration: I designed it! mouse-matt.jpg

The background image is simply one part of suburban Dublin, showing the importance of adding buildings to the map. It also shows the campaign name #osmIRL_buildings. We have a task manager here to co-ordinate all this work.

Why is our community mapping buildings? Well after years of focussing on other things with more historic value we had a big discussion about what mapping focus would be scalable, and stimulate most map completeness. We picked something that would help us add more addresses and landuses (especially the human ones). More progress with these will, we hope, allow more commercial and environmental re-use of the database.

How are we doing with this task so far? Well we have almost 1.5 million buildings added, which is 25% completeness. This is why osm communities need to have multi-year projects. Do you want a mouse mat? You can have one if you map 10,000 buildings with us.* The first 18 are on their way to their new owners.

*while stocks last :)

Location: Woodlawns, Kilmore A ED, Dublin, Dublin 17, Leinster, Ireland

The year past and the year to come for my mapping

Posted by DeBigC on 31 December 2020 in English (English).

Deja Vu, again :)

This is now the 3rd annual instalment of a crazy naval-gaze about my mapping. Blame it on whoever called this part of the osm space a diary. If there is a chance to speak about my personal reflections I’m grabbing it. And what better time of year than now, since I see other people reflecting and goal setting.

My Targets

I wrote about measuring my own commitment to mapping in 2019 and then again in 2020 for the years ahead. It was fun to see my objectives shift around and good for me to test myself against them. In 2020’s diary I then had to be like the Roman God Janus; to look to the previous year by way of review, and to the coming year by way of promising some targets.

My Hot Spot

Last year I undertook to have my hot spots look less like the measles and more like a large boil, and for that boil to be in Ireland. This is definitely the case now, as the graphic shows (use the slider to see the expansion of mapping in Ireland). Having achieved this I’m not concerned with continuing to make a target for this in 2021.

My Ireland Contribution Rank

Again, the target was to stay in the top 10 mappers in Ireland. It was sustained in 2020, but there was one occasion in July where I dropped out of the top 10. I am presently the 5th biggest, without too much effort. Since I technically didn’t keep this one I will go forward with it and try to stay in the top 10 in 2021.

In the past year I realised that Pascal Neis’s “Ireland” is in fact only the Republic of Ireland. A lot of my mapping is in Northern Ireland, and as a consequence shows up in the UK. This about 20% of my present mapping.

As an extra kick I’m going to try to develop a tool that shows all the mapping on the whole island of Ireland, and this itself can be a target since it would be of benefit to everyone involved in mapping Ireland.

Map on the move

I am still Ireland’s biggest Mapillary contributor with 1,395,539 street images, which is an increase of 100k. The target was to get them where they were needed most of all, and I have done that in Kilkenny, Newry and Clonmel in 2020. I will continue until I reach 1.5 millions images which will hopefully be this year, but its not a hard target.

I am also Ireland’s 6th biggest StreetComplete contributor, and I want to improve this position until I am 3rd. This will be hard, as lots of great mappers were using SC before me, and lots of big contributors are getting more and more interested in it - watch this space!

Mapping Gaps

I had a qualitative target to join up a specific gap area in North Dublin, which wasn’t mapped at all. I nailed this, and I am now going to do the same in Blanchardstown (there are two large gaps with no buildings) and Harold’s Cross/Terenure (Dublin’s last large very unmapped area) in 2021.

My Success rating

I am awarding myself 3.5 out of 4 this year, with 0.5 deducted for being out of Ireland’s top 10 in July. I will be back at the end of 2021 to report on whether I can hit all my target. Stay with me and encourage me if you can, and if you see something I should improve give me that feedback when you can. Happy New Year!

Cartographic Poverty - the grounded truth

Posted by DeBigC on 29 October 2020 in English (English). Last updated on 30 October 2020.

Background

The following assertion was made by a blogger around 18 months ago:

"..mappers as a whole are mapping more features (buildings, houses, libraries, schools, green spaces etc) within the wealthier areas of the Dublin region than in the poorer areas." {53degrees}

It was made here and in a follow up here

Reading the two posts together with the twitter promotions for them it is clear that the hypothesis rests on the perception that OSM contributors, specifically those contributing to OSM around Dublin, suffer from some kind of collective unconscious bias, to the extent that they have neglected the deprived areas of Dublin and mapped the affluent areas to a higher level of completion. This assertion, if well founded, would reveal that not only does the map of Ireland, specifically Dublin have gaps, - but that these gaps are patterned in such a way as to further exclude and marginalise the residents of these areas and the second blogpost makes no doubt of this consequence.

We have a great name for the band!: "The Spurious Correlations"

The author refers to 3 or 4 examples of poor areas, sparsely mapped, compared in turn to affluent areas mapped to a higher extent using visual assessment. There is no transparent methodology or apparent logic for these comparisons other than they were what the author appears to want to choose. Perhaps they are conveniently selected to suit the conclusion.

The author repeated the claim on twitter [now deleted] on the 23rd October 2020, presenting the evidence of these three or four comparisons as conclusive proof of a “startling lack of completion of working class areas in osm, a citizen science project”. Some charts were shown, demonstrating a very selective method of counting tags. Whether this represents traces of a deeper investigation remains unclear. This needs to be tested here and now.

Transparent Method

An inquiry with some more rigour than the two blogposts is much needed now. The first step is to refer to the deprivation index available here, and identify 10 outlier electoral divisions that are the most affluent, and 10 outlier electoral divisions that are most deprived1. If the assertion holds weight it should be detectable at the outlier values.

The ED boundaries were downloaded from the CSO website and the deprivation index was downloaded from the Irish Government’s official data portal and matched to these boundaries to enable comparison2.

depmap.jpg

The OSM data was extracted as a .pbf tile from Geofabrik on the evening of the 23rd October and parsed in a PostgreSQL database to exclude areas outside the boundaries of the electoral divisions. Three queries detected the three types of object geometry which is what the blogger’s assertion rests on and simply counts them. A fourth query detected In the case of ways and relations it included those that were partly inside the boundary3.

The calculations are simply the aggregate count of all types of objects, further aggregated into an affluent total and deprived total, and then divided by the size of the area to fairly reflect the fact that the areas vary from 1.06 square km (Avonbeg) to 5.96 square km (Priorswood).

The Result

Let’s get down to it…

The first result produced is a count of all objects in the 20 extreme affluent/deprived areas, grouped into those categories, aggregated and normalised by the area size of the total electoral division area. Ultimately this counts and compares the density of mapped objects in the two groups of areas

  Affluent EDs objects per Sq KM;         Deprived EDs objects per Sq KM;
  696         1,173

This result disproves the blogger’s conclusion about deprived areas being neglected by mappers. I will provide a separate diary post showing changeset dates to demonstrate that from 2016 a big upturn in mapping of Dublin’s deprived areas took place, and a rationale for why that began to happen, but that is for another day. And now for the tagging depth result…

  Affluent EDs Tag3+ objects per Sq KM;           Deprived EDs Tag3+ objects per Sq KM;
  652           186

This shows that the affluent areas, though having less object geometry are tagged in somewhat more depth and detail. The figure may be skewed by two of the Affluent EDs which has peculiar characteristics, these are Mansion House and Rathmines East. The former has several city centre streets with specialised Government and Commercial services, however the latter is mainly suburban with a few shopping districts. This does support a totally new hypothesis of more nuanced mapping effort in Affluent areas, but more nuanced does not mean more “complete” as the original blogger’s claimed. The ages of the changesets suggest the original object geometries were typically provided around 2010-2012, and more osm contributors using on-the-go editors like StreetComplete, Vespucci, maps.me and OSMand.

Conclusion

These results, despite the finding of more detailed tagging in the affluent areas, do not support the main hypothesis that Dublin’s socially deprived areas have been deprived of mapping. It would appear that the mapping for these areas is at a higher level of completeness and will be easily converted into tagging depth. There is nothing in the original idea that provides much insight into either the OpenStreetMap community and its mapping preferences, neither is there much to increase the cause of Dublin’s deprived neighbourhoods. A further diary entry will deal with some of the extraneous claims and wrong assumptions made, but that is for another day.

Footnotes
  1. This is an outlier comparison often used in large datasets to eliminate “middle to middle” statistical noise. Twenty areas were chosen to increase processing speed. 

  2. Neither dataset was uploaded to OpenStreetMap since they do not have compatible licence. The analysis was done in QGIS. 

  3. DeBigC is grateful for the support of Rusty Broderick, who mentored me through PostgreSQL six years ago, and patiently ran the database queries to correct counts 

StreetComplete Discovery

Posted by DeBigC on 25 October 2020 in English (English).

This is a fright that turned to fun.

In my morning walk today I travelled along Shanowen Grove, and I took StreetComplete with me, adding building levels, addresses, building types and so on. On the last section of road I discovered one of the buildings to appear on StreetComplete as a tower.

Posting in the #osmIRL Telegram group I shared the screenshot. Dónal came to my rescue It turns out the building was specified to have 21 levels. Looking at the object history on JOSM later it turns out the levels were set by Tshedy see here. However, knowing Tshedy’s level of accuracy and the likelihood that “21” is a typo that was intended to be a “2”. This is very much in keeping with a similar “small”(but big) error in Melbourne.

StreetComplete is a great tool, especially where the basic building objects have been mapped, in a slow stroll along a street one can add serious detail, and the better the initial mapping the easier it is to see which objects require further details.

Mapping in Clonmel for Heritage Week - Ireland

Posted by DeBigC on 8 August 2020 in English (English). Last updated on 20 August 2020.

OpenStreetMap in Ireland has launched a project for National Heritage Week 2020, with some preparatory work now underway by the community. A town was chosen, deliberately large to have a decent number of heritage sites, historic buildings and well documented past celebrities and people of repute, and so we arrived at Clonmel.

There is a bit of a plan in place, set out here on the project wiki. The first step was to set down a base map, with the roads, street layouts, landuses and as many buildings as possible traced off Bing. This got done here with 176 squares mapped and validated in just 3 days by the community! The Power of the crowd……

The next part was to do some mapillary, to harvest as much about missing streetnames, building names and details as possible. This was done on a joint visit with Annekaro a.k.a b-unicycling. Admitedly what was captured is about 20% to 30% of the area of the town. So perhaps more is needed and hopefully Waterford Dave’s tweet will rescue a few more of the missing details ;)

The next phases are probably a more advanced population of building detail, and way more pin mapping. It is hoped that by liaising with locals on the ground we can activate more new users and make the project more visible, especially the benefits #OpenStreetMap offer to community based heritage mapping.


View Larger Map

My Podcast Interview on nodes and ways with Austin

Posted by DeBigC on 15 May 2020 in English (English).

In my professional life I have been on radio a whole bunch of times, and on TV a couple of times. My mother used to say I had a face for radio ;) .. but I am experienced enough at this point to handle most media. For the first time ever I was involved in a podcast, and I was super glad that it was the new osm related podcast, “ways and nodes” which is done by Austin Bell who is known as itsamap! and maps a whole lot around South Carolina, in the USA.

The podcast is right here so listen and enjoy.

I learned a lot observing and listening to Austin. He hangs out beforehand and develops his questions by talking to you, and it was great that we mapped together beforehand looking at the same stuff on the map of Ireland. There was a bit of a theme about migration. By the end of the podcast he was determined to map based on his own family’s historical origins in Scotland.

You can get in touch with Austin by simply mailing him austin@nodesandways.com and look at this mapping account here I think it is important that osm has a regular podcast, and Austin gives up a lot of time he could use mapping to make these episodes.

What I learned: #StayHomeAndMapIRL

Posted by DeBigC on 29 April 2020 in English (English).

This is a about what I have learned most recently by being involved as a contributor to, and organiser of the “mapping sprint” or “mapathon” called #StayHomeAndMapIRL.This is written in my own voice as an individual mapper.

The mapping sprint was a week long, and really grew out of a spontaneous confluence of two things:

  • The mapping campaign #osmIRL_buildings for which we use a task manager

  • The necessity of everyone in Ireland to stay home, as part of the Covid-19 emergency

By combining these two we were creating an immediacy and dot-joining exercise between something people were all experiencing, and something useful to do with the time at home.

Our community agreed at the 2019 AGM to have a mapping task, one where participation was voluntary like always, but it would be the “common ground” for mappers. Some deep thought went into choosing buildings as the focus, not least being that openstreetmap for Ireland seems to lag behind in the completeness of the buildings. Typically, a large Irish town or city could have a small number of buildings mapped, even in the town centres, almost as if buildings were not a significant human feature and not a major part of the detail of any map at the end-user scale. Of course the wider point is that working at this scale adding a feature like buildings allows our community to scale up and down, and by that I mean add useful end user things like details about the buildings, or scale up and create better de facto landuses - ultimately to increase the completeness and usefulness of the map.

The week through insights taught me the following things:

  1. The #StayHomeAndMapIRL mapping sprint evolved somewhat spontaneously. OSM meetups in Ireland seems to have a number of gears, in which we focus on one aspect of activity, training, talking, showing new stuff, or mapping. We rarely do all of these together, yet the week allowed this to happen. Working together, and changing gears together, as talks were delivered can only increase the glue and focus between everybody, and will help osm to recruit new contributors. These two outcomes are vital for a healthy community activity, especially an activity done by volunteers.

  2. The community seems to work well using the task manager, where people see each other marking a tile in blue with their presence they do stuff like map tiles near each other, and have a sense of progress - not measured but visible across the space we are working on. Mappers can see unfinished tiles by others, and finish them. Mappers can split tiles where they feel bogged down. The tasks where a larger number - at some points in the week we had 7 at one time in Cork - were working simultaneously seem to make the mapping task disappear quickly, and even big and complicated Cities like Derry, Drogheda, Newry and Limerick had big chunks of detail added to them and they are all approaching or above the 50% completion mark. We should reduce down the number of tasks that are open, and not dilute our ability to work together and motivate one another, through gentle competition, validation and interaction.

  3. I think community interaction would increase if it were to establish itself on a more face-to-face (or voice to voice) basis and with regularity online. At the end of the week we learned that we already have a dedicated Big Blue Button space, and the board members should organise this a bit to have a weekly hangout or chat and the community members could each volunteer to take the lead and talk about something. Perhaps a format will evolve in time which allows this happen with regularity.

  4. The week showed me the output of the community if we try to map together, learning from the three previous points. The community was able to get from 902,000 buildings on the 1st January, to 1,110,000 buildings on the eve of the mapping sprint. The mapping sprint added almost 40,000 buildings on its own. This would represent a speed multiplication factor of 4 times for adding buildings on what has gone before this year. We cannot sustain probably this effort using the same call to action, so the organisers and community will have to think about what can be sustained about the week that would keep us focused and contributing to the #osmIRL_buildings campaign.

Thanks to everyone who got involved in this, the osm community from far and near for every building they mapped. Thanks to my fellow board members of osmIRL for supporting the sprint in loads of ways. Thanks to Heikki Vesanto for making the gif, which shows activity over 3 hours chunks and captures all the mapping action. action

Hitting (and missing) DeBigC's personal targets

Posted by DeBigC on 23 December 2019 in English (English). Last updated on 31 December 2019.

My Promised Targets

Last year I set down my mapping new years’ resolutions well before January. So wouldn’t it be fun, and also embarressing :) to review these and see if I had the year I hoped to have. My diary entry a year ago is here. Set out the various targets.

My Hot-Spot

I wanted my heat map to shift to Ireland. While it is true that I developed a number of heatmap centres in Ireland, I achieved these by following the dispersed regional community initiatives (remote tasks and face to face meetups) set by #osmIRL. This meant that I didn’t have a particular area into which the aggregate effect of highly intense mapping could be channeled. So my big heatmap remains in Lesotho, and because it built up over 4 years it may stay that way for a while. However, considering I wouldn’t have made my largest hot-spot shift anyway, I regard the case of measles I have caused on the linked map to be a success.

My Ireland Rank

Next, I want to maintain a top 10 place in the Ireland 2 monthly contributor list. I was able to do this exept for twice, as in June I fell to 20th and in August I was 25th. This was caused by a lovely summer where I went outdoors a bit more plus my old trusty laptop had some issues and I couldn’t afford to replace it immediately. I have learned that to stay on top (top10) in Ireland one can have 2 week breaks, but no longer. In 2020 I will not have breaks longer than 2 weeks.

Map on the move

I wanted to hit the Mapillary millionare status by the end of January, and then I wanted my total in Ireland to hit 1 million by the end of June. I comfortably made it with both these, and my Mapillary is now 1,292,702, and I don’t want to set more targets here, except to capture images where osmIRL needs them most of all.

Mapping Gaps

My target was a little bit too qualitative to know whether I made it or didn’t make it. However, subjectively I am very happy that the areas that were unmapped that I mentioned, Coolock, Raheny, Drumcondra, and Beaumont are far more complete in terms of highways, buildings and street furniture. Pretty much I made it here.

My success rating

I’m giving myself 3/4 with the added benefits of hindsight and learning. I am a stronger mapper now, but because I have large contributions my new targets should be quality based, and relate to osmose and keepright detections. I would also love to have a target for people I teach to map, maybe the number of them that hit 10k nodes or something like that.

Newry meetup yesterday

Posted by DeBigC on 17 November 2019 in English (English).

I was at the #osmIRL and #digitalNewry meetup yesterday.

It is wonderful to see the people of a town coming together and speaking about the challenges they have in situ, boundaries, licences, old buildings clumped together in their town core. There was a diversity of interests, yet a common idea that getting Newry more mapped than it presently is.

Newry still misses a lot of details on buildings

Old ground and new frontiers: OSM meetup in QUB

Posted by DeBigC on 20 February 2019 in English (English).

I used to live in Belfast, for around 5 months in 1995. I went there to do a piece of research, some of which was ethnographic, and the rest of it was a few visits to the Northern Ireland Statistics Office. It was an amazing time for that city in terms of peace breaking out, and a City literally stretching with a sigh of relief. And it was an honour to be there at that time. Since my return to Dublin, I only went back three times. Which is very little for a place that I felt very much at home in.

Heading up with my mapping buddies Lineo and Tad we talked about how streetsigns and road markings have subtile differences across the now infamous brexit hard/soft border.

Lineo compared the border to that between Lesotho and South Africa, while Tad resolved to come back this way on his motorbike and capture as much imagery as possible while it remains open.

Arriving at Queens

When we got to Queens’ University and found our kind event host Conor Graham we realised how ready a university can be to hear about openstreetmap things. Their labs, wifi, and the interests of the staff are all really very OSM things already, and opensource is the way forward.

Queens Department Banner

Queens asked two of us to participate in a GPS improvement experiment which he describes here using this thingy.

The event was enjoyable, mostly because of the social side of mapping we get when we have a meetup. I always learn loads, which is why I mention new frontiers. Also, I had a chance to map the building I lived in, on Oldpark Road, old ground indeed. Old Maps

2019: the year ahead in mapping for DeBigC

Posted by DeBigC on 1 January 2019 in English (English).

Setting Goals for DeBigC’s mapping

In the year ahead I want my mapping to be more strategic and pointed. I have set the following goals, to assist it in achieving this. These goals will be hard to achieve in total, but if the majority of them are, I will be very happy since they will mean I am moving towards my goals. I am a board member of OSM Ireland CLG LTD, so just for clarity sake these goals are nothing to do with any position taken by the board or on behalf of the community.

My Hot-Spot

Firstly, I want my heat map to shift to Ireland. I have been a large #MapLesotho contributor, but for my own reasons I want to map Ireland much more. At the moment the largest concentration of my nodes is Lesotho as shown on this heatmap

My Ireland Rank

Next, I want to maintain a top 10 place in the Ireland 2 monthly contributor list. I’m all for other power mappers flexing their muscles, so this is a relative target to whatever they contribute, I will stay up with them.

Map on the move

Ireland doesn’t have a “millionaire” Mapillary user, and there are more than 50 of these worldwide. My target has three aspects,starting with me reaching 1 million images by the end of January 2019 (right now I’m on 913k). I want my Irish image haul to reach 1 million by June 1st.

Mapping Gaps

Regarding my own contributions Dublin is priority. For some reason large parts of North Dublin are not well mapped, with some exceptions like Castleknock and Clontarf, due to the stewardship of large locally focussed mappers. However, there are huge swathes of urban Dublin barely mapped at all aside from landuse and road networks. I’m aiming to map Coolock, Raheny, Drumcondra, Beaumont much more than they are done, to the point where all the buildings in these areas are complete.

OSM Ireland's Mapillary Workshop - Tog Dublin 24-11-2018

Posted by DeBigC on 25 November 2018 in English (English).

I was asked to do a workshop on mapillary yesterday by the OSM community.

I delivered some slides to kick off. These are here I was focussed on discussing how best to manage the mapillary app on an android device.

Two of the teams

At the end of the talk we went walking after breaking into groups of 2 and heading out with our phones held high. We decided simply to walk around the blackpitts area of Dublin and capture some of the uncovered areas. In general main roads in Dublin have a high chance of some coverage, but when you look at side streets and a lot of residential roads there simply isn’t the same level of coverage.

Before the mapillary teams walked

After the walk images were uploaded

The mapillary users learned a number of lessons when they returned to the workshop. These came out in the discussion:

1.Walking on narrow streets the optimal mapillary capture angle is not straight ahead.

2.If walking on the left side it is better to point right, and the opposite if walking on the right. The reasons is that the extra distance allows a whole building be captured

3.Old phones you have laying about the house can be pressed into service

4.A more nuanced view of the coverage in Ireland is needed, maybe something that highlights areas where there is a highway, but has no proximate imagery.

All in all it was a good day, and hopefully the members keep up what they learned.

Making lots of new data for the OSM.ie community: Mapillary

Posted by DeBigC on 21 May 2018 in English (English).

At the end of 2017 I was the biggest Mapillary contributor in Ireland, without really trying too hard and just using a front facing Samsung camera in my car and short diversions on the commute to and from work.

At some point around March 2018 my good buddy and OSM Ireland contributor Dave jumped ahead by about ten thousand images. I made a solemn (ahem… competitive, but sporting) promise/threat to get working on closing the Mapillary gap from me to him. I know, this sounds like General Buck Turgidson from the movie Dr. Strangelove….

mapillary_me3.jpg

I could see that other Mapillary users get more images out of a trip than me, so I set the phones to distance capture at 3m. Distance capture of course needs to be reviewed before being uploaded. Thinking about what I had learned about using Mapillary in Lesotho (mainly that you may not pass this way again) I decided to collect together my phone, the spare phone and the older android phones laying around the house and press them into service, pointing in various angles, which is another lesson from trying to map. A front facing camera on a road where there is a road clearance is next to useless unless all you want is road-signs.

After ransacking the house that gave me four working phones, so I needed to go to adverts.ie and I was pleasantly surprised that if you bargain you can easily get a reasonable second hand mobile phone for 15-20 euro (obviously caveat emptor). I bought two of these. I then went down to the discount shop and bought two cheap charging cables of 2m length for that they can reach from the cigarette lighter to any window in the car. I also realised that my emergency car starter has a USB port and two 12v cigarette lighter type ports, so without any additional purchase I also have a way to charge up three phones that have a depleted battery off an independent cell and without running the engine of the car.

mapillary_me.jpg

I then deployed five of the window cups supplied by Mapillary and set off. I have to say this didn’t last long, and in my opinion didn’t constitute safe driving. So I figured out that taking another mapillary user to make sure that the phones were all capturing and set correctly is the only advantage here. I have been lucky enough to involve Tshedy and Fifi in several of the longer and more detailed trips.

I find the android Mapillary app sometimes has issues, two of my phones have effects where the captured images don’t delete and then the capture randomly stops, sometimes crashing the app and sometimes not. I know this is a work in progress and will improve so that it can be more reliably deployed across a range of different devices. I had issues last year where a new phone basically didn’t capture the exifs containing the co-ordinates and this was a big loss. The other thing is the more I use it the more work-arounds I figure out.

Now I am back on top, having recently passed the half million images mark (see below). All the way through this I have to remain focussed on the value that a street level image is to a mapper in terms of applying the correct tags to what’s on the ground, be that a building, street furniture or a natural feature. Being top is only a temporary thing, there have got to be more clever solutions to the challenges I have overcome, and when that rocks in I will hand over the crown.

mapillary_me2.jpg

I sometimes have 3 but other times 7 angles on my trips. As a parting nugget of wisdom I have also realised a blurred image is better than no image if you are a mapper. In the land of the blind, the one-eyed man is king. Ireland is a very small community when it comes to mapping and Mapillary. The way I look at it images taken now should be of ongoing value to mappers for the next few years, as Ireland doesn’t have the urban areas and built environment mapped in detail yet on osm.

We have very bad weather here, which would mean at least 6 months of the year give the only option to be armchair mapping. When the sun comes out it is great to get out and about and capture it for others to enjoy, and map.

Location: Santry, Whitehall C ED, Dublin, Dublin 9, Leinster, Ireland

#MapLesotho the long tail of 824 mappers

Posted by DeBigC on 12 July 2016 in English (English).

Martin Dittus normally does this sort of analysis!

Just looking at our present counts of everything in #MapLesotho

We count nodes being created, edited and deleted. We have set up the count to feed out of the osm database, so anyone working within the polygon of Lesotho is picked up. Of course GeoFabrick only gave Lesotho “spatial sovereignty” in February 2015, so we have a year and a half of editing to look at here. This shows 824 people involved.

I like to cut off at 100 mappers. When you do that you see that the top 100 have done 97% of the mapping. The remaining mappers listed tend to have very modest contributions to #MapLesotho. The modal value of nodes is in fact 1. I have no idea what causes this, but when I click into their usernames I tend to see a mixture of once-off mapping, dormant accounts, while others are do appear to be active. Indeed, there are even some that I see involved in other HOTOSM and missing maps tasks. So regarding humanitarian mapping I guess #MapLesotho is like a holiday resort, a bit of a hotel to park your single node. And its not a high priority task, so that’s ok isn’t it…?

Between 50 and 100 is an interesting space. Most of the mappers here I have met and taught to map. They are mostly Irish schoolkids from Portmarnock or Basotho Planners. Mapping didn’t really grab this group as their favourite activity, of course they can have up to 10 mapping days and over 10,000 nodes. But those that are not new are attenders, they sit back… others will organise the Mapathons. Others will map alone. They feel safe in the big group and have mainly helped the mapping by making the feeling of a flock. But when we get to the 50th mapper mark we still have have 94% of the mapping done by those above.

The mappers in the top 50 are a blend. There are far more Basotho people in this group. In fact I can see that 36 are Basotho, and 14 are not. Of the 14 around 7 are from Ireland, while the other seven are international crowdsource contributors who simply like #MapLesotho. However, these 14 are more likely to be at the upper end of the register. For example in the top 10 provide 75% of all the mapping, and as many as 5 are international, leaving 5 Basotho. Tshedy is a Basotho Physical Planner and is at the top of the tree for all of us all, but the next biggest Basotho mapper Lineo is 6th.

While its always good to know that such a large number of people are involved I am learning that quality control is best exerted with the top 50 people. The others help out - of course they do. Is every node cherished - of course it is. Is it all a caste system where mappers not in the top 50 can’t get into the club - of course not! However, the end result is best affected by talking to the top mappers, engaging and even debating things with them.

Want to play with the top 100 data?