What catches my eye is the implication that in OSM features are "baked" into the map. More than anything OSM is a database. It is not so much a map drawing client. The client that you are referring to is probably the one on the view tab on the OSM page. That particular client's main purpose is merely to visually support contributors in making updates to the database.

In the database, nothing is baked in.

That said, the OSM foundation doesn't own the data. The contributor owns their contributions agreeing to license them under CC by-SA. Couple that with the fact that the specific OSM client you are referring to is open source, anyone is free to clone and extend it to their purposes (layers or what have you). If desired, their improvements to the client can be submitted to OSM for inclusion in the common if they are beneficial to map making. Or, an altogether new client can be created using the OSM database.

Nothing is baked-in in OSM. Contributors are free to add whatever (non-malicious content) they want, if there is no tag for a novel feature, then from what I've read OSM is open to additions.

OSM offers more freedom on all fronts.

It's like you have 400 books in one book just because you have 400 pages in it…

I think you have to keep that definition inside the company… When you are talking to general public or even not-so-educated GIS users, 1 map means 1 map… That's just a marketing mumbo jumbo you are playing onto them… That's my 2 cents…

CyanBlue

Tim's tweet specifically referred to the number of "user-generated maps on the site", this is why I chose to use Unique Visitors as the main metric. I agree that Compete does not have the most accurate data. That being said, their data is good enough for the purpose of this post which is to show the general difference in traffic magnitude.

You have a good point, ESRI's definition of the map might be very different from my own. In my opinion, a user-generated map is simply a bounding box explicitly created by a user with or without associated geometries…

Cheers,
Andrei

I think you make a good point: it is the definition of "a map" that is key to understanding the figure of 400 million. I don't know what definition has been used, but I am pretty sure it is not likely to be just tile requests.

So, what is a map? I make the point in my crude analysis that I take one pan or a zoom to equal one map request (I have no idea if this is what ESRI are quoting of course). In ArcGIS desktop, ArcGIS Explorer or any of the ESRI web API based clients, this would perhaps equate to between 1 and about 10 separate tile requests (for new areas of base map not previously cached in the client) plus perhaps one or two dynamic map image requests (for any "business" data overlays).

My back-of-the-envelope calculations were just to highlight that the average GIS user map usage I use when sizing mapping systems, 6 pans and zooms per minute on average, could account for that number of maps (and more) requested in a month with the number of unique visitors quoted by Compete.com. So yes, I stick to my original statement, it doesn't seem unreasonable to me, based on those assumptions. Of course I have no idea how accurate my assumptions are in this case!

My bone of contention with Andrei's analysis is merely that Unique Visitors is the wrong metric to use when challenging the quoted figure of 400 million maps. Compete.com defines it as follows:

"The Unique Visitors metric only counts a person once no matter how many times they visit a site in a given month".

My point is we don't know how many times those 33,000 visited in a month. I am just theorising that given the userbase of ArcGIS.com it is not unreasonable to think that some or many of those visitors are working in the 6 maps a minute, all working day, 20 or so days a month pattern I described. If so, they could request a very large number of maps indeed.

I have no better idea than anyone else what the 400 million figure includes but I don't think it can be entirely made up of tile requests. As far as I am aware, not all maps published through ArcGIS.com are tile caches, some are dynamic map services where you request a new bespoke image each time. It gets more complicated when you look at ArcGIS Server map services registered with ArcGIS.com but hosted by other organisations elsewhere; I have no idea if requests to these services via ArcGIS.com are being included or not either.

Ed

Learon, ESRI has 'confirmed' that this was not a misquote. http://apb.directionsmag.com/archives/9155-Esri-C…

Andrei, Awesome Blog post, thanks.

If one unique visitor actually visits every working day for 20 or so days a month, then 12,000 maps over 20-odd days is about 500-600 a day, 80 or so an hour in a 7 hour working day, or less than two a minute, per unique visitor.

My experience tells me that GIS users tend to average about 6 pans and zooms (=maps) per minute when working so this doesn't sound unreasonable to me.