OpenStreetMap

1 million is a big number

Posted by Harry Wood on 6 January 2013 in English (English)

1 million users! Woosh! Welcome to 2013. This is going to be a good year.

osmstats.altogetherlost.com screenshot

1 million is a big number. My uncle is a retired school teacher, and he has this great story. He told a classroom full of school kids "1 million is a big number. Between all of us, d'you think we could draw one million dots?". He showed them how to draw a block of 100 dots on a piece of paper in a neat 10x10 grid, then they raced to see who could draw 1000 dots as a row a 10 of these blocks. Some kids did it very quickly. Some kids took a bit longer and did them in colourful stripes. "Not bad" he said "If we add them all up, working together, you've drawn a lot of dots, but if you lot can draw 1 million dots I will eat this banana whole, including the skin!". Excited by this prospect, the kids all feverishly carried on drawing dots, but at the end of the lesson they added them up and found they had only done about 200,000 dots. Good progress, but they had a long way to go to reach a million. Lesson learned. 1 million is a very big number.

...but a couple of girls went home that night and carried on drawing dots. Two months later they presented my uncle with a box piled high with dots on paper all carefully arranged in colourful checkered grids. My uncle (now retired) showed me this box which he had kept ever since. It's a true story.

...and yes, he had to eat a banana whole, including the skin.

I'm not sure what the lesson is from this. Maybe "never underestimate the power of bored kids", or "if you ever thought editing OpenStreetMap was a pointless waste of time, people do more pointless things", or how about this lesson "A massive task might just be possible if you have lots of people to helping out and a few very dedicated people getting really into it".

Anyway 1 million is big. The above graph is from osmstats.altogetherlost.com. It shows us blasting over the 1 million user mark some time in the small hours of yesterday morning. This morning's raw stats report clearly show us over the line. The graph shows a downturn at the beginning of the week, which isn't usually possible in the count of registered users. This is caused by the sysadmins doing a clear out of some spam user registrations. They've done this before in the past, and they wanted to do it again in advance of the 1 million users announcement.

Whenever we do announcements about the number of users we have, people are quick to complain that this is misleading. This number includes people who signed up, but didn't make a single edit to the map data. That is no small component of the total, in fact it's a very large component. About 2/3rds of signed up users never get as far as clicking the 'save' button on a map edit. 2/3rds! This is based on Pascal Neis' analysis of this from a couple of years back, but I'm assuming this ratio hasn't changed much.

So we might say a less misleading figure to look at, is the number of users who made a least one change, but this quickly leads to the next question "Should we really count users of OpenStreetMap who have only made a single edit before disappearing?", and then we're into different ways of measuring activity. People have done this analysis. Andy Robinson has been re-plotting this graph since the early days, showing a "rolling week" measure of node editing activity, currently fluctuating around the 8000 mark. We have to measure the size of a community somehow. There's no doubt this 8000 figure is a more down-to-earth realistic figure than the over-blown 1 million figure we're singing and shouting about. But why take a 1 week rolling count? Why look at node editing? There's lots of different ways of aggregating these stats, all with different quirks. The fact is, the least arbitrary measure we have available to us is the cold hard simple count of user sign-ups, and today this is showing a big fat 1 million! Yes we're singing and shouting about it. Guess what? We're trying to promote OpenStreetMap. Nothing wrong with that, so please join in!

But the revelation that 2/3rds of sign ups do not edit, is a frustration for people who work on improving OpenStreetMap. I like it when the discussion comes around, because it leads to inward-looking self-criticism. What are we doing wrong? I would urge people to think of the zero-editors as part of the long-tail distribution of our community. I'll explain what I mean some other time, but basically we shouldn't beat ourselves up about it too much. But self-criticism is good. It leads to conclusions like:

  • OpenStreetMap editing software needs to be easier for beginners,
  • Documentation needs to be better for beginners
  • Editing on mobile needs to be easier
  • The website needs "social" features for welcoming and helping new users
  • We need simple bug submitting and workflows around fixing them.
  • We need to make contributing fun, like a game, with microtasks.

The good news is that we're tackling things on all these fronts. We saw some great progress in 2012, and we're set to see exciting announcements in 2013. In particular...

  • Development of the iD editor is full steam ahead. A version stable enough to operate on the live database with full editing functionality will be a huge moment. This promises to bring editing onto javascript capable mobile platforms and (I think) attract a lot of developer interest in customised editors.
  • We may soon go live with a "Report a problem" link on the OpenStreetMap homepage. 'notes' branch
  • In 2012 we've seen the PushPin OSM app brings easy OSM editing to iPhone.
  • In 2012 we've seen MapRoulette challenges breaking TIGER fixup in to interesting little microtasks. Now we're seeing the 'Kort' presenting KeepRight bugs as a fun little game.
  • switch2osm.org launched at the beginning of 2012, providing targeted documentation on using OpenStreetMap for web developers. RichardF was the main man behind this, and he's itching to do something similar again. Meanwhile learnosm.org is set to receive a facelift very soon.

It's things like this which will improve our "conversion rate", moving people from signing up to making edits, and allow us to really tap into our 1 million strong army of registered OpenStreetMappers. That's why 2013 is going to be a good year.

Comment from Geonick on 6 January 2013 at 22:03

Hi Harry

Nice blog. One million is a big absolute number indeed - and also in relative terms: it was you who reported almost exactly one year ago about the 500,000th signed up user of OSM in this thread: http://lists.openstreetmap.org/pipermail/talk/2011-December/060829.html .

Yours, Stefan

P.S. Thanks for mentioning Kort which is a thesis work of two students of mine! There will be data for whole world soon.

Hide this comment

Comment from Rovastar on 6 January 2013 at 22:15

A great summary of where we are at so far.

Actually (from osmstats.altogetherlost.com) it is more like 4/5ths or 80% don't make an edit at all (208,000 from 1 million).

[http://osmstats.altogetherlost.com/index.php?item=members under the graph No. of members who are the last modifier... (Node, Way or Relation)]

But there more as in the past since the license change 6 months about 350,000 new users have joined and only about 25,000 of those users have made edits. So the current rate is about 7% of new users in the past 6 months actually commit an edit.

But there are many reasons for this (I suspect even more spammers than we didn't catch, etc) and didn't want to be negative just to get a better understanding of the stats so we know if we improve the "conversion rate" - but lets shout about the million "users".

Also there is a new group the Welcome Working Group that is helping address how to welcome new users and encourage more editing and identifying the reasons why so few do edit.

http://wiki.openstreetmap.org/wiki/User:Mvexel/Welcome_Working_Group

Hide this comment

Comment from Rovastar on 6 January 2013 at 23:53

Ummh my comments seem to have disappeared I'll posted again......

A great summary of where we are at so far.

Actually (from osmstats.altogetherlost.com) it is more like 4/5ths or 80% don't make an edit at all (208,000 from 1 million).

[http://osmstats.altogetherlost.com/index.php?item=members under the graph No. of members who are the last modifier... (Node, Way or Relation)]

But there more as in the past since the license change 6 months about 350,000 new users have joined and only about 25,000 of those users have made edits. So the current rate is about 7% of new users in the past 6 months actually commit an edit.

But there are many reasons for this (I suspect even more spammers than we didn't catch, etc) and didn't want to be negative just to get a better understanding of the stats so we know if we improve the "conversion rate" - but lets shout about the million "users".

Also there is a new group the Welcome Working Group that is helping address how to welcome new users and encourage more editing and identifying the reasons why so few do edit.

http://wiki.openstreetmap.org/wiki/User:Mvexel/Welcome_Working_Group

Hide this comment

Comment from dcp on 7 January 2013 at 14:01

"We're trying to promote OpenStreetMap. Nothing wrong with that, so please join in!"

I must be one of the oldest contributors to OSM and of course I will try and promote the concept but I refuse to brag using manipulated statistical data!

We just don't have 1 million users/contributors and you all know that to be true. If we promote OSM using the 1 million statistic we leave ourselves open to ridicule and I see nothing to be gained from that!

I welcome the "Welcome Working Group" and just hope that they will go about their System Analysis in an applicable manner.

If they want to know why some sign on and do nothing then they must make the suggestion that the sign-out procedure be simplified. Something like this:

-sign-on

-don't like it

-sign-off with a short multiple-choice questionnaire.

Nothing more is required.

PS As one of the oldest contributors I should sign-off days/months before I die, but you know, there is no simple way to do it, and I am, I think, an experienced contributor. That cannot be right can it?

Hide this comment

Comment from Harry Wood on 7 January 2013 at 14:47

"I refuse to brag using manipulated statistical data." Well of course you're are very welcome to brag about OpenStreetMap with whatever supporting facts you like. But 1 million users is a good news story this week. It's not a manipulated statistic. As I've explained, it's the least manipulated of several possible statistics we could look at.

You seem to be confused about "sign-out procedures". You can of course very easily "sign out"... being the opposite of "sign in". The opposite of doing a "sign up" to create a new user account, would be to destroy your user account, an operation which people rarely have any sensible reason to want to do, and which presents all kinds of problems in terms of IDs cross-referencing of contributions. If a user has some sort of unusual privacy concern and they want to remove all identifying information, they can very easily change their 'Display name', and remove their 'Description', 'Home location' fields. Beyond that sysadmins can label as a user account as 'deleted', but non of this is a regular operation that any normal user would want to worry about. It's not something a user who is no longer interested in OpenStreetMap would want to go through, and it's certainly not something we'd want to make easy to do by accident. If you think I'm saying something deeply mysterious or nefarious here, please have a look at... well any other sign-up website on the internet.

The issue at hand is all those users who have signed-up on the site, failed to make an edit, and then not come back. "Signing out" has got nothing to do with it. So what we could do (and this has been suggested before), is email a questionnaire to those users. Or we could just email them just asking them to come back!

Hide this comment

Comment from dcp on 8 January 2013 at 08:38

To Harry,

I love your enthusiasm. Just the way it should be. Please carry on.

Maybe I am confused/wrong or just stupid. True is: The majority of registered users have not contributed anything to OSM in the form of data or services (i.e. the back-room guys without whom none of this would work. They must get the most praise from me).

We are not discussing those users who wish to hide their identity by destroying their account, but those who register for the first time without knowing what is expected of them. When realization kicks in and they decide OSM is not for them (many reasons here) they cannot easily close a registered account.

I far as I understand it and I may be wrong, the only way to close an account is to contact an administrator and ask them to do it for you. Do you really think a newly registered user would know how to do this? I, myself, do not know how to do it, and surely the administrators have better things to do than close user accounts!

Are, therefore, the statistics being interpreted correctly and do they truly mirror our community? Let every man decide for himself. For myself I don't need hundreds of thousands of redundant users to admire the efforts of the OSM community. It is amazing what has been achieved in a few years and that is something worth promoting.

Hide this comment

Comment from chriscf on 8 January 2013 at 08:44

Out of interest, where did 2000 accounts go across the new year?

Hide this comment

Comment from Harry Wood on 8 January 2013 at 10:02

@chriscf as mentioned, sysadmins dropped some spam users

Hide this comment

Comment from Rovastar on 8 January 2013 at 12:33

@dcp

As Harry said most websites don't work this way where you can delete an account.

Try deleting an account from hotmail, facebook, any forum/message board, etc. There might be an exceptional way to do this but if far from common.

I cannot imagine many, if any, people request to close an account unless they want they contributions deleted. The overhead to the admin is minimal, in fact I would say it would be more work to create a system to delete accounts then the do them manually.

Hide this comment

Comment from dcp on 8 January 2013 at 19:39

@Rovarstar

Let us not emulate other sites behavior culture. Let us do it our way!

I know I am repeating myself when I say that we are not discussing those uses who have contributed data to OSM. My main point is those who register, don't do anything and want to unregister. LET THEM DO IT in an easy and simple manner.

As @Harry Wood mentions above the system administrators were able to delete 2000 accounts and that must have costs them a lot of time just to define them as spam accounts.

How much effort went into the licence change which culminated in the deletion of many accounts.

Obviously Harry Wood does not agree with my arguments so let us leave it at that. Harry you doing a good job. Keep it up. I have great respect for you.

Hide this comment

Comment from Skippern on 9 January 2013 at 15:10

I saw you mentioning a few editors, just to inform, I am part of a small selected team to betatest OSMiOS, a full scale editor for iOS, which according to the developer will be released for public free download in App Store within a month or so.

This is probably thebyear we see editing goes mobile, with hundreds of mobile devices contributing with adding buildings, POIs, roads, addresses and much more.

OSMiOS at current state are able to create, edit and delete nodes, ways and closed ways, though cannot merge or split ways, and can only read relations. It have the most complete tagging preset I have seen in any application (only JOSM with additional loaded presets can overgo it), as well as free hand matrix tagging. If you are an iOS user, look for it in the App Store in near future

Hide this comment

Comment from SimonPoole on 10 January 2013 at 18:59

@dcp

we actually have something over 300'000 contributors that have made edits (the numbers on alltogetherlost only count the last editors of objects). As to the 1'000'000 accounts, I believe reporting the number is quite valid (well at least in marketing terms). Particularly since we know that we have a substantial number of contributors that register an account, go away and come back a long time later and start editing.

I am myself such a case with a good year between creating the account and my first edit.

Hide this comment

Leave a comment

Parsed with Markdown

  • Headings

    # Heading
    ## Subheading

  • Unordered list

    * First item
    * Second item

  • Ordered list

    1. First item
    2. Second item

  • Link

    [Text](URL)
  • Image

    ![Alt text](URL)

Login to leave a comment