Hi Lasse,
I remember having received a message earlier from you, although I probably didn't find time to reply. At the moment I have time again, and there is indeed very much a could tell. Now looking back, all this is a long story, with a lot of experiences about user-contributed content on the web.

I began modeling in 2007 when SketchUp was just bought by Google. They had just built 3D Warehouse (that SketchUp didn't have it before). There was no organized process to accept buildings to Google Earth and there were no guidelines (acceptance criteria). The Google Guides that I later learned to know personally just found good models in the haystack and accepted them, often happy that someone had created a building at all (even without textures or roughly estimated dimensions). My first mode took 7 months to be accepted, and my motivation was just to learn modeling a (random) monument in my proximity and to try out the process. After a year the review process was improved with the so-called "3d pipeline" [announcement], and models where accepted within 1–2 weeks (and around 10000 human reviews per week). That was when I actively began modeling, and walking through my town to systematically take photos of important buildings to model.

Motivation
My motivation was the joy of modeling (or of achieving to have finished a building) and documenting historic buildings in my home town/region and showcasing them/making them accessible in 3d to the world (often with detailed description of the building's history).
For buildings outside of my region (elsewhere in the country or world), I was motivated either by the huge size, or by the lack of an appropriate, good model in Google Earth (for example if the previous building was untextured and totally in wrong scale, eg. I replace [1] by [2] and [3] by [4]).

Creational Process
Whenever I saw a massive/characteristic building or stumbled over a footprint of a historic building, I was thinking how great it would be to have it in 3d. The creational process is (in contrast to other processes like sport or music) characterized by a material output. As an artist I like to collect what I create and wish it remains for ever (no vanitas), and I feel very touched when loosing something. Virtual 3d models do not degrade or become dirty over time (as physical paper/clay/wooden models would do). I
However we have learned over time that virtual goods also "degrade" by becoming irrelevant (loss of interest or harder to discover), or because progress in technology allows to create better virtual goods (higher quality photos etc.).

User-contributed content
Over time, Google learned to "use" its community and began to systematically organise it, by introducing Community Managers (Nicole, Mason, Jordan, Craig, Alex, …), competitions (Model Your Town), conferences (first conference: Zürich 2010, since 2011 a conference on every continent per year) and a community website [announcement] and featured modelers on their blog. The conferences where always together with Google MapMaker (2d maps editor) and Google Panoramio (geo-located photos) communities, although the communities had not so much overlap. Modelers who attended only had to pay their travel, but where provided with accomodation and food in a 4 star conference hotel. The event was professionally organized by Google's James Therrien and SketchUp's Product Evangelist Adian Chopra, who let us at the end shout "UGC! UGC! UGC!" (user-contributed content).
Google also introduced "badges" with amount of achieved models in 3D Warehouse to honor modelers and they had a special "Supermodeler" group of 50-100 selected modelers. A Googler (Craig) worked in his 20% time to allow selected users to adjust the Google Earth Terrain to fit their models and integrate these patches back to Google Earth. This non-public "Terrain Edit" project was offered to frequent modelers but it was complicated and error-prone and did not scale to larger user numbers.

Changes
Google also introduced new and alternative technology in steps, and it was always very controversially percieved, but never as bad as the end, because people became used to it as long as they could do their human-made models.

The first change was the introduction of BuildingMaker [fan blog], [shut down], and online tool that uses 45° aerial photography and photogrammetric algorithms to create simple 3d buildings from matched points. Its advantage was that it was fast and efficient and its imagery (where available) covered backyards and roofs that you couldn't photograph from street level. On the other side introduced many new people to geomodeling who had never used SketchUp before. Many more people tried it out once and only a few became permanent modelers. Due to this BuildingMaker had a lot of poor output and flooded 3D Warehouse with undetailed blurry partly black boxes, which in turn gave a bad impression of 3D Warehouse content quality.
Me and other SketchUp modelers tried it out and saw that it was possible to create good buildings. But since the aerial imagery was too blurry (ca. 0.5m detail) and not available for rural areas we turned back to SketchUp for our major work.
Building Maker became available in ca. 100–150 cities.

Other changes where revised acceptance criteria. Newer acceptance criteria was supposed to make reviewing easier for Google stuff and reasons for rejections more transparent to users [official blog]. In general they were stricter to avoid an inconsistent user experience for Google Earth users (and many modelers complained about their models being rejected), but I personally had no problems because the criteria seemed to me logical and my models very realistic enough to meet the criteria.
A side effect of the new criteria was also that user-contributed models were treated equally to models from other sources, ie. human-made models had no higher value, but the better model wins.

Later modelers discovered usernames (buildcities001…) of groups at Google who uploaded masses of building models (thousands, ten thousands). We assumed these were hired by Google to complete cities in densely populated areas, and we were confirmed that their models are judged by the same criterias as ours.

Finally Google introduced autogenerated cities (before the 3d-mesh era). They were based on the same data type (polygon models) and loaded on top of the Google Earth Terrain like other individual models. They were not hosted in 3D Warehouse, and it was harder to consider individual user-contributed models, because the autogenerated model needs to have the right "cuts" to replace a building by another model.
But it became clear that algorithmic modeling approaches are faster and more efficient. Only the quality was significantly different: We modelers could achieve a higher level of detail and precision whereas autogenerated models achieved more completion and better visual consistency (same shadow and light conditions). Our own models were uniformly lighted and with repeated textures because it was effort to take photos from all sides.
Like BuildingMaker the autogenerated cities were only available in few densely populated areas / metropoles. We modelers were not too much interrupted because we saw it progressed slowly and there were still 2000000 villages/cities on the planet.

Only in 2013, Google announced the new 3d mesh [link] which was so completely different that it could not reside side-by-side with traditional models.
The old technology was a terrain surface with models sticking on top of it (not allowing overhangs, tunnels, caves).
The new technology integrates all into one mesh where every vertex can have an arbitrary position in space (including overhangs), but it makes it impossible to distinguish buildings or objects algorithmically.

Relationship
Google provided free tools and is often seen as the "benevolentor" who gives all things for free to the people (which can be disputed, see adverts, data collection).
The modelers maybe accepted that without thinking much about it (I don't want to say naive) and concentrated on their artworks and sharing.
Compared to MapMakers or Panoramio users they were less associated to the brand of Google. I didn't ever feel as a Google Evangelist, but rather as an artist, and Google provided an easy platform for us that required almost zero effort other then making our art.
However since I use free/opensource software and operating system, I saw this more skeptically, that by contributing we gave Google an advantage over other companies. Also there were conflicts in the Terms of Use: The Terms of Use were intentionally hold fuzzy about rights of creators vs. rights of Google/Google users. I believe (without being a lawyer) models from 3D Warehouse are not allowed to be included in any other repository or mapping service (→ monopoly of GMaps), except if the creator himself uploads a copy to another repository.
But there was no alternative in sight, and even if there would have been, Google Earth would be more widely known, had more viewers to see our models, was easierly accessible to new users.

After October 2013
I was very bound by my studies and hadn't modeled for long and after the and of the 3d pipeline there was no motivation for me to start a new model. Two months ago I discovered the OpenStreetMap 3D pages in the wiki, and the reference to 3D Warehouse. It seems their efforts have so little advanced (and I wonder why they never found contact to the 3D Warehouse modelers) that I don't have high hopes, but it would be so interesting if the once engaged 3D Warehouse community could help advance this project. It is the only noteworthy alternative to Google Earth.
I thought about it and planned it theoretically (evaluating the chances and risks), but then I was suddenly surprised by the rapid launch of the new 3D Warehouse (that has no messaging system), that I had to quickly launch this project within a few hours and invite as many modelers as I could reach. It lacks definitely in practical things and good marketing.
The response is rather little (10–15 of more than 400 people have directly replied and promissed their interest). I believe more people are watching and would join when/if it advances. It seems, in comparison to other community projects, 3D Warehouse modelers are not used to the role of driving a project on their own (and contributing in planning/building) but rather to use what is provided by Google or to leave in mourning. They are not technical people and the lack of a big brand is not very convincing to them. But even contributions in words/opinions are highly valuable, I think.
The problem is: Currently there is no website where I could tell people "Instead of uploading to 3D Warehouse, we upload now to this site". OpenStreetMap does not have enough such infrastructure yet (because 3d model files are a different data type than the map primitives node/way/relation). It is a chicken-and-egg problem.

• The short-term goal is to find a way to convert 3d models into 2d OpenStreetMap data types with height tags (to see them immediately on a map).
• The long-term goal is to accelerate the discussions in OpenStreetMap 3D to finally result in usable products, ie. develop a repository and an attractive 3d viewer for end users/visitors.
  

I'm doing this because nobody else does. I'm not sure if we will ever get there, but I'd rather try out to prove whether this alternative works than directly giving up. If it doesn't work after several months, there is nothing lost.

–––––