streetcomplete
Quest: How many power conductors are seen on this big power line?
General
Affected tag(s) to be modified/added: cables Question asked: How many power conductors are seen on this overhead power line?
You can get an idea of cables=* completeness worldwide on this map (select a country and see the cables=* completeness KPI under the quality score section)
It should not be confused with wires, intended to count the amount of wires in a conductor bundle (we only count conductor bundles here).
Checklist
Checklist for quest suggestions (see guidelines):
- [x] 🚧 To be added tag is established and has a useful purpose
- [x] 🤔 Any answer the user can give must have an equivalent tagging (Quest should not reappear to other users when solved by one)
- [x] 🐿️ Easily answerable by any pedestrian from the outside but a survey is necessary
- [x] 💤 Not an overwhelming percentage of quests have the same answer (No spam)
- [x] 🕓 Applies to a reasonable number of map data (Worth the effort)
Ideas for implementation
Focus on power=line not power=cable, power=minor_line (invisible or too many items)
The goal is to encourage counting cables on such situations:
- Overhead transmission line => expected answer is 6.
- Overhead transmission line => expected answer is 9.
- Overhead transmission line => expected answer is 3.
- Overhead transmission line => expected answer is 2.
No metadata needed, counting power conductors is the same worldwide
I imagine several possibilities:
- Most simple one: a counter, mappers set the count and that's it
- Average one: 3 or 4 most usual situations are set as pictures and users can set a counter for any other less usual situation
- Most graphic one: Every situation is set as a picture, but there could be many of them
We don't count earth wires, intended to prevent lightnings to disrupt overhead lines, as conductors.
How do we prevent users from counting such lines?
On big lines ( >= 50 000 volts), electrical conductors are bound to towers with insulators but ground or telecommunication wires aren't. The guideline could be to count conductors that are insulated from towers.
Is it really always that clear? Here an example from sweden: https://c7.alamy.com/comp/BHHA72/a-power-line-sweden-BHHA72.jpg I believe the top cable is insulated right?
The attachment is particular and designed to prevent damage in case of surge caused by lightnings. It has a tiny insulator to let the surge go through a more robust connection on top of it.
But it has noting in common with the long insulation chains of the live conductors and that's a clear visual difference. Not to mention power conductors use to be thicker than ground cables.
But it hasn't noting in common with the long insulation chains of the live conductors and that's a clear visual difference.
Do you think an average users would be able to differentiate those two? Also sometimes towers can support two different types of cables right? With different voltages? Would these not also have different sizes of insulators? How would a user differentiate?
Also: I feel like this quest would could be very spammy: since if the user is traveling on a road along a power line, they would be repeatedly answering the same answer. Though I guess this is already the case with the existing power line quests (material and insulator type) which is why I personally have mostly disabled them.
Also sometimes towers can support two different types of cables right? With different voltages? Would these not also have different sizes of insulators? How would a user differentiate?
The quest doesn't require to distinguish different voltages. It only consists in differentiate live from dead cables. Big lines will show > 1m insulation chains for any voltage and that's the point.
That would be different if we ask for voltage but that's not the plan currently.
Also: I feel like this quest would could be very spammy: since if the user is traveling on a road along a power line, they would be repeatedly answering the same answer.
No because the quest covers the power=line, not power=tower nodes. We add cables=* on the line. So if the user goes along a road along a power=line, the question will be asked once. The spam could occur when the same road crosses a lot of different power=line, which is unlikely.
TL;DR: IMHO probably too complex for average SC user (SCEE might be a better match).
To be added tag [...] has a useful purpose
Out of curiosity, it might be useful to mention what is the useful purpose in knowing number of cables=*, as it is likely not obvious to many people.
Question asked: How many power conductors are seen on this overhead power line?
It should not be confused with wires, intended to count the amount of wires in a conductor bundle (we only count conductor bundles here).
If we are to only count conductor bundles, then perhaps the question should ask about "conductor bundles" instead?
Do you think an average users would be able to differentiate those two?
Yeah, I would say xkcd #2501 applies here, as I do not think it would be "Easily answerable by any pedestrian" (given all the explanations given above) (IME, many people wouldn't even be able to correctly name whether some thing you show to them is wire or cable, much less make determination on high voltage peculiarities, not to mention OSM tagging intricacies!)
The risk here is that unless the quest is perfectly clear[^2] even to, say, people who have no idea how to name 3 wires going to the socket from their home appliances[^1], putting such quest (in StreetComplete of all places!) will likely result in poisoning the cable=* tag -- thus making even values entered by domain experts unusable, as you wouldn't know which values are fine and which are not.
IOW: not having data is much better then having incorrect data.
I'd usually suggest SCEE (StreetComplete "Expert edition" fork) instead for such cases, but I worry whether this might be at the fringes of its reach too -- it would seem to me that one should really read both wiki pages before even attempting to tag either of them. I'm not sure the whole story could be condensed in small amount (i)nfo button text?
[^1]: who are much bigger percentage of population then one might assume, as shown by that xkcd truism above [^2]: oh, and don't count that all (or even most) people would read the instructions on how to solve the quest, experience seems to suggest otherwise 😢
Hey everyone, I'm also part of MapYourGrid and supporting @flacombe
Counting numbers of power towers cables is actually a pretty easy and simple job to do. The example @paulklie provided is really an edge case and you can find similar edge cases for almost every SC quests I guess. And even if they include the earth wires, it is not that problematic as we can automatically detect such errors using our QA platform / osmose: https://mapyourgrid.org/quality/
Alternatively, we could only allow even numbers for AC lines, which would make it even easier.
Out of curiosity, it might be useful to mention what is the useful purpose in knowing number of cables=*, as it is likely not obvious to many people.
The transmission grid data in OSM (power=line) is the only global applicable GIS standard for transmission grid planning and is used by many organisations, such as the World Bank. If you know the voltage and the number of circuits, you can estimate the capacity per line roughly. This is important for modelling, simulating and planning new grid infrastructure. You can read more about the impact of this data here: https://mapyourgrid.org/impact/ We specifically use this data for
- https://github.com/pypsa-meets-earth/pypsa-earth
- https://github.com/pypsa/pypsa-eur
- https://github.com/PyPSA/PyPSA what is the most used open source framework for energy system modelling and planning.
Also: I feel like this quest would could be very spammy: since if the user is traveling on a road along a power line, they would be repeatedly answering the same answer.
Most transmission lines actually do not follow streets. Distribution lines do but this is not what we are interested in. I would recommend to apply this only to power=line not to power=minor_line
Counting numbers of power towers cables is actually a pretty easy and simple job to do
I still think that is a heavy case of xkcd #2501...
To test that hypothesis, given the suggested question "How many power conductors are seen on this overhead power line?" translated into Croatian ("Koliko je energetskih vodiča vidljivo na ovom nadzemnom dalekovodu?")[^1], here is an random Croatian user answer:
Overhead transmission line => expected answer is 6.
The actual answer is "13"
Overhead transmission line => expected answer is 9.
After zooming in twice and some serious multiple counting - the actual answer is "about 26? or was it 25?"
Overhead transmission line => expected answer is 3.
The actual answer is "4"
Overhead transmission line => expected answer is 2.
The actual answer is (after zooming in and puzzled expression) "uhhhh, maybe 12?"
Sure, sample size of 1 is anecdotal, but that is what I have at the moment.
Would those answers be acceptable and fixed by your QA tool? It might perhaps improve the quality by some ~20% if there were good explanation added in the (i)nfo button (and maybe up to 50% if https://github.com/streetcomplete/StreetComplete/issues/6208 gets implemented first), but even in best-case scenario I would still think that is terrible data quality if that is representative sample at all.
Could you try asking just that question "How many power conductors are seen on this overhead power line?" to few dozen random people (not related to electrical engineering field) and showing them those pictures in isolation and without any further instructions (telling them there is no right and wrong answers to avoid anxiety), and then see how they answer?
That should create bigger sample size and give you the idea what data quality to expect (also more pictures than 4 would be even better, given by random, but that is much less important if you interview all people in isolation, of course blind test -- so they cannot see your reaction - but make sure they don't ask chatgpt or search for this question or similar sites which explain which answer is correct -- people will often "cheat" in such "tests" just so they don't come out as "dumb", even when you instruct them not to 🤷♂️ )
Alternatively, we could only allow even numbers for AC lines, which would make it even easier.
That would fail 50% of the OP's examples (where correct answers were odd), wouldn't it?
Perhaps we need a new StreetComplete "quest" type "Please take a picture of this <description of the element>" which only allows to take pictures, and leaves the rest for domain experts to handle? 🤷♂️
(But then, it is somewhat outside of what StreetComplete is supposed to be doing....)
[^1]: I admittedly might not be the best Croatian translator in the world, but it seems mostly accurate to me, and I have no idea how I'd translate it more closely to original English wording.
The transmission grid data in OSM (power=line) is the only global applicable GIS standard for transmission grid planning and is used by many organisations, such as the World Bank.
Thanks for your explanation though! And much thanks for your OSM mapping of the power infrastructure of course (I only do the little on that front)!
It makes me worry though; don't people who build those power towers IRL have more accurate database of they've built?!
I've always worried that it was next Carrington event that would completely destroy our civilization (unless AI did us first), but now I worry it could be random annoyed OSM vandal whose Great Tagging Renaming Schema community didn't like that could easily single-handedly accomplish the same 😰😵
@mnalis The shocking truth: Many utilities don't have good, accurate data, particularly in low- to medium-income countries in Africa and Asia we are working with.
OpenStreetMap is widely used, especially when it comes to regional studies involving multiple power grid operators, because it is very time-consuming to compile the individual data sets that are created with little / no common standards. We are currently working with some Canadian groups to create the first high-quality grid data map of Canada. Also the US government needs 25% of OpenStreetMap data to create a nationwide map of the transmission grid. OpenStreetMap is widely used, especially when it comes to regional studies involving multiple power grid operators, because it is very time-consuming to compile the individual data sets that are created with different standards.
I absolutely agree with @mnalis and @paulklie here.
I respect your fervor to map all the power line infrastructure, but surely no data is better than wrong data, isn't it?
The two other power pole related quests are already somewhat borderline on what can be expected to be answered correctly by a non-expert, this one clearly exceeds it, in my opinion. (I would have totally answered @paulklie 's example wrong).
To be honest, no offense, but that you didn't immediately see that it could be too difficult to answer correctly by a layman after the first replies leaves me with a bad feeling regarding the two other quests. Are they really as easy to identify as came out of the discussion back then? Aren't they also prone to wrong answers by laymen?
Anyway, this one I will close as not planned.
I respect your fervor to map all the power line infrastructure, but surely no data is better than wrong data, isn't it?
Really sorry to here that @westnordost @mnalis @paulklie. It would have been a great addition to StreetComplete and the whole OpenStreetMap ecosystem.
I don't share your doubts about the quality of the data regarding the number of circuits, since counting them is really easy. Three or six cables or nine cabels it is in 99% of the cases. If someone enters four or seven cables because lightning protection was included? We would immediately see this discrepancy with Osmose and our global quality analytics. You used a borderline case to make your argument.
For your interest, here our quality dashboard that help us to identify such issues: https://mapyourgrid.org/quality/
This is a real missed opportunity to significantly improve the quality of important data in OpenStreetMap. Unfortunately, the existing Poles' quest is completely irrelevant. The quality of the distribution grid data on poles is very poor and it will always be in most place. However, for transmission lines on towers, we have the opportunity to obtain high-quality data. We will look into other ways to get this done.
However, for transmission lines on towers, we have the opportunity to obtain high-quality data.
Mapillary should be usable often, as you only need one good image per section.
I don't share your doubts about the quality of the data regarding the number of circuits, since counting them is really easy.
So, if you @Ly0n saw that some user counted number of cables as 12, 4, 38, 13 (at 4 different places; real-life user answers); would you really be able to tell us reliably what should be the actually correct numbers there? (and how did you calculate them from user answers)?
Mapillary should be usable often, as you only need one good image per section.
Good point! I'd encourage users to contribute street-level imagery to Panoramax as more open alternative to Facebook's Mapillary (Baba app is nice choice for contributing to Panoramax if you use Android on your phone).
Street imagery would be quite useful there IMHO, as it would help mapping power lines, but also many other things (powergrid-related or not) - without taking any extra time from the person on the ground. And dedicated mappers could then much more safely extract correct numbers from those in areas, without relying on layman mappers knowing the difference between wire and cable (much less having other high-voltage transmission knowledge)
@mnalis As I said. Because we talk about AC transmission lines the answers can only be 3,6 or 9 in almost all cases. I'm mapping power lines for quite some time now and this are the answers that a technical possible for 99% of the cases.
@HolgerJeromin Mapilllary is technical possible but the coverage of this data is really low and most of the data is very old.
I agree on the help that Street imagery can provide but isn't this the case for almost all quest in StreetComplete? Why StreetComplete was interesting, is because it is working on mobile and we have many people in low to medium income countries that would like to contribute, but they only have a mobile phone.
@westnordost Any chance to reconsider this decision for power=lines and remove maybe some other power related quests? This would be really important to us because it would people that just have phones to fill this important gap that we have in the power infrastructure data.
Have proposed question been updated from initial one:
How many power conductors are seen on this overhead power line?
How exactly it would be phrased to stop people from answering 7 or 13 or 12 to https://wiki.openstreetmap.org/w/images/0/01/Power-tower.JPG ?
(I would definitely not answer 6)
How exactly it would be phrased to stop people from answering 7 or 13 or 12 to https://wiki.openstreetmap.org/w/images/0/01/Power-tower.JPG ?
Maybe a number select where the user can only select numbers divisible by 3. And we instruct them to "round down" to the next integer divisible by 3?
Could you try asking just that question "How many power conductors are seen on this overhead power line?" to few dozen random people (not related to electrical engineering field) and showing them those pictures in isolation and without any further instructions (telling them there is no right and wrong answers to avoid anxiety), and then see how they answer?
Yes I will, and it takes time to properly plan and execute it.
Maybe a number select where the user can only select numbers divisible by 3. And we instruct them to "round down" to the next integer divisible by 3?
I think something like such visual can help anyone to understand what is expected and to feel safe about their answer. That's the point of discussing it here: find relevant solutions to make complex things simple.
It should be refined and the poll I plan to build will contain both annotated proposals like such and not annotated ones to see the difference.
Because we talk about AC transmission lines the answers can only be 3,6 or 9 in almost all cases.
Well, then perhaps the quest should only offer those 3 answers (and force mapper to leave a picture for remaining 1% of the cases). E.g. something like surface quality quest.
Picture and longer explanation for each why that is correct answer. If done well enough, that might actually work for SCEE. (but I'm still quite dubious that it would pass "👨💻 Users are no experts" StreetComplete guideline)
stop people from answering 7 or 13 or 12 to https://wiki.openstreetmap.org/w/images/0/01/Power-tower.JPG ?
Maybe a number select where the user can only select numbers divisible by 3.
Isn't a 12 divisible by 3, yet 12 is completely wrong answer for that picture?
That's the point of discussing it here: find relevant solutions to make complex things simple.
Sure... although https://c.osm.org or mastodon or somewhere else is likely much better place for that then the SC issue tracker -- which should be used for actionable items; i.e. suggesting quest only when a working approach has been found (i.e. one that produces good results even for people completely unfamiliar with the subject).
It should be refined and the poll I plan to build will contains both annotated proposal like such and not annotated one to see the difference.
Perhaps one can create a secret theme on MapComplete, make an introduction and javascript exam somewhere on the https://mapyourgrid.org web, and only give a link to that mapcomplete theme to people who have successfully proven to be able to answer relatively correctly (e.g. at least 95%+ correct answers)?
Or have an app with integrated poll like e.g. MapSwipe that people must pass correctly before solving some quest type?
Dear all
I found some time to set up this poll to assess how accurate answers could be on such a question proposed for a quest here. https://framaforms.org/how-many-electrical-conductors-are-seen-on-this-line-how-openstreetmap-mappers-are-used-to-count
It's an attempt. Actual options and improvements for this quest could be found once enough answers will be provided.
For reference: This is the discussion about the poll in the OSM community forums: https://community.openstreetmap.org/t/tagging-poll-how-mappers-are-counting-cables-on-power-lines/139436
While I appreciate the initiative taken, unfortunately (as I noted in more detail in that community thread) that poll results would be unusable for assessing whether such quest might be suitable for StreetComplete. 😿
Because the poll:
- explains to users in detail (with extra links to the wiki!) how to correctly answer the questions (something SC would not be doing!), and
- also uses extremely leading questions like "Does this picture show 4 cables?" (with 4 cables marked extra bold in the picture!) instead of behaving like the proposed SC quest would actually look like.
also uses extremely leading questions like "Does this picture show 4 cables?" (with 4 cables marked extra bold in the picture!)
There may be a misunderstanding here: pictures with extra bold labels are actual possible answers. There are only 4 questions asked (one per page) and annotated pictures are possible answers!
There may be a misunderstanding here: pictures with extra bold labels are actual possible answers.
Oh, ok, that would make more sense -- that was definitely confusing in combination with "yes"/"no"/"I'm not sure" for multiple of them. But even with that clarification, it is also stil confusing as available answers differ per question (in SC all instances of the quest would always get all answers, as it has no way to preselect only some of them) . And also some of them seem to have no correct answer (e.g. "situation 3" should clearly be "12", but that answer is not even present in that situation?)
If I understood correctly then, you would like to do something like What's the surface here? or What kind of fire hydrant is this? image-quests? Where there would be exactly 6 images: pictures of power towers with emphasized cables (and overlaied number of cables: 3,4,6,8,9 or 12) and Uhhh... answer to leave a note if it is any other number, right?
Still the issues with the poll remain:
- the pictures need to be much smaller if they are to represent what would be visible on phone (even on my oversized Samsung Galaxy S23+, the "answer" pictures in poll are much bigger than what would be shown by SC. So they should be about 1/4 of their current size to represent what SC would show.
- the poll should not have had extra instructions beforehand on how to solve it correctly (as neither does StreetComplete give such instructions before solving quest).