The Thingiverse – Model Reuse in the 3D Printing Community

meckler14m223D Printing is still a hot topic. More 3D printers are being developed which can be seen for example in the amount of Kickstarter 3D printer projects. Having an opportunity to use 3D printer is not enough though. Digital 3D models are needed too and sharing the models is said to be common among the community members. For long time Thingiverse has been the de facto service to store digital 3D models for others to use.

According to Wikipedia Thingiverse.com is a website dedicated to the sharing of user-created digital design files.” It is said to be “the most widely used 3D model repositories in the Maker community“. Some say that “Thingverse is built on the concept of giving back. In other words, as a responsible community member, you should not only download models but also upload and share your creations to repay the community”.

Other digital 3d model sharing services

Yet, not all share the same positive image of Thingiverse, which is mostly due to issues that has been tangled around Replicator2 and Terms of Use for Thingiverse. What connects the two items (Replicator2 and Thingiverse) is that both were owned by same company, Makerbot Industries, before it was acquired by 3D Printing industry giant Stratasys in June 2013. The topics were so hot that even a ‘occupy’ movement was born and some took their 3D models down from Thingiverse.com. Around the same time, an alternative for thingiverse was born. It is based on GitHub and is labelled as githubiverse described by Gary Hodgson. Another alternative candidate for thingiverse is cubehero.com. Neither githubiverse or cubehero.com seem to be popular. Their impact, if none, on digital 3D model sharing remains to be seen.

Inspired by the discussions about the future of Thingiverse and other activities around the topic, I decided to do some analysis on ‘things’ in Thingiverse.com. Is it really about sharing? What kind of licenses are used?

Methodology

Analysis is based on metadata information from 117,450 ‘things’.

Data collection and statistical analysis. Scraping data, there is API. It lacks useful examples, so I found it easier to build scrape-script with ruby than use provided API. All data was collected to MySQL database. Analysis is based on metadata information from 117 450 analyzed ‘things’. It must be noted that the data represents snapshot view (16.8.2013) to thingiverse 3D models. It is not ‘live’ situation. By running the same process again after a few months could provide some information about timely changing features that can not be captured with snapshot. One such feature is the amount of private ‘things’.  From 3d models in status of “private” metadata collected was limited to thingid.

Collected metadata for 3d models having ‘public’ or ‘in progress’ status contained following items:

  • Thing id,
  • status (private, in progress, public)
  • creator handle,
  • create date,
  • downloads count,
  • likes count,
  • tags used,
  • collections count (how many have added model to collections),
  • makes count (how many times this object has been manufactured),
  • license used and (if given)
  • remixes count (how many derivates have been made from this model)
Data available for further analysis in CSV format in GitHub (15,6MB, ~117k lines)

Results

Here’s some key results in brief. More detailed charts with interactive features can be found fromhttp://ossoil.com/thingiverse/. I will not go to details or discuss the results much. The results from this research will be used in co-authored academic article which will hopefully be published some time around early 2014.

Only half of 3D models are shared

Nearly 42% of the thingiverse.com 3D models analysed were private.

As it was mentioned earlier, Thingiverse is often understood as a place to share 3D models for others to manufacture but also to enable derivate work. To us results indicating that thingiverse is not so purely about sharing came as a surprise. Nearly 42% of the 3D models analysed were private. Private is a ‘thing’ that only owner has access to. Even metadata about the model such as license and creation date are hidden.

 

thingiversepie

tnumbers

Figure 2: Amount of uploaded model by status. Source: http://ossoil.com/thingiverse/#overview-in-numbers

 

tnumbers

Tags

The popularity of ‘customized’ tag might indicate high rate of derived work

Some other tags such as ‘model’, ‘useful’ and ’3d’ can be seen more or less useless or at least seem to have very little informative value for end-user. However those tags are most common and tell us more about the users and in this case about their tools than about the models they upload. Most commonly used tag is ‘customized’, which is quite general. The popularity of ‘customized’ tag might indicate high rate of derived work. Also the usage of parametric indicates that some 3d models are intended to be scaled easily. Another observation is popular usage of ‘reprap’ and ‘openscad’ which both refer to open source/hardware thinking since first one is open hardware printer and latter open source CAD tool.

It seems that some of common tags are ‘invented’ from tools with which model is created,  name of printer it’s part of or targeted at.

ttags

 

Tags count in uploaded models is rather low (figure 5). Nearly 11 000 3D models (public) don’t have any tags. Most common practice is to put one tag, apparently one of the above.

ttagscount

For more detailed list of licenses and percentages, see http://ossoil.com/thingiverse. 

Derived work and remakes

Most remixed 3D models are licensed under Attribution 3.0 Unported (CC BY 3.0), which allows commercial usage and is not ‘sticky’. However, when it comes to making (manufacturing), Attribution-ShareAlike 3.0 Unported (CC BY-SA 3.0) is more used. The latter license is ‘sticky’ and requires that “If you alter, transform, or build upon this work, you may distribute the resulting work only under the same or similar license to this one.”

 

chart(8)

 

Uploads and downloads

Makerbot industries and thingiverse was under microscope in media for altering terms of use in Thingiverse during spring 2013. Apparently media attention has possibly given a boost for thingiverse, not the opposite since upload counts seem to have taken a steeper curve around the time media focus peaked. It must be noted again that the below charts contain just ‘in progress’ and ‘public’ things.

chart(7)

The 3D model download amounts (Aug 2013: around 25,5 million) have grown steadily without dramatic changes even though the amount of uploaded 3D models has started to grow in steeper curve since Feb 2013. Reasons for this might vary and can not be found from the data. One can only speculate. Perhaps the community has become more picky and require high quality models and publicity has lowered the barrier to upload lower quality models.
Another reason might be that community has not yet found the new models due to lack of proper tags or model descriptions.

chart(6)

More detailed interactive information about the statistics: http://ossoil.com/thingiverse/

###

Jarkko Moilanen is an experienced community builder in realms of hackerspaces and MeeGo networks, several successful project management positions in software development and system design. http://fi.linkedin.com/in/jarkkomoilanen

This article originally appeared here. Republished under creative commons license.