Hi, this is Wayne again with a topic “Maker Faire 2013 3D Printing Stage: Open Innovation Environment Designers in Thingiverse”.
Thingiverse is a website for the 3d printing design. Community users of thing givers can freely download designs, they can modify for existing designs and upload their new their versions, or they can combine design, combine multiple designs and create something new. We looked at designs that were submitted to think averse over the course of more than three years. It was more than 16,000 designs and some of the success metrics we were interested in were the number of likes of any given design the number of makes meaning how many people downloaded that design 3d printed. It took a picture of it and then later uploaded that picture back to think averse and because of the Prezi logo. You cannot really see it so at the number of tile designs. So how many designs were inherited ideas from that design later on, and we were also interesting understand, measuring novelty in a more objective way than the ones currently used most of the scientists looking at how people innovate and create something novel. Usually, we focus on either analyzing the network attributes of the process, or they usually evaluate innovation from more in a qualitative way which sometimes can be somewhat subjective. So we we adopted a method primarily used in computer graphics and what that method basically does when comparing two given designs is a focus on their safe differences. So basically, the distance we would get from implementing the metal will give us an understanding of how different they are F, regul medical perspective.
So such a metal not only allows us to understand the individual contribution of someone and in thing givers and in the process of a adopting others, ideas but love, but also it allows us to compare any given Thingiverse designed with any design that was submitted on thingiverse. Before it so in a multi-dimensional space, where each dimension is any prior design, we can understand whether the design that is proposed or submitted is has been its relative neighborhood, any other designs. So that’s, let’s understand the foul novel. The design was at the time of submission and something that probably we expected was that novelty was associated with popularity, meaning that more novel designs, the way I described being as novel, were also more popular compared to imitative design or less novel designs, but somewhat unexpectedly. We also saw that practicality was associated with a novelty, a meaning that designs that were not more novel, also tended to be pretty more.
In addition, we were interested in the value of openness, meaning whether participants of fingers were actually deriving some value out of their participation. In a opening open collaboration environment like fingers, and to test that we categorize designs depending on whether they had inherited ideas from any other prior design in thingiverse, or they were standalone designs. Many that the whole creation process happened outside thinkers and again, as you can see and inherited designs, outperform standalone these and bouncing both in terms of lights and in terms of mix and going a step further on the in charity designs. We classify them depending on the total number parents, these designs head and, as you can see, designs that had more parents and tended to receive a higher number of likes, focusing on designs with only one parent.
It seems that the best strategy designers could follow would be to build upon prior creations of theirs, suggesting that a more modular process should be followed when in terrific, from only one design, and also we when we looked at designs with two parents, it seemed that the Optimal strategy that someone could follow would be to use one of his prior up. Sorry, please use one of his prior designs and won a design of someone else’s, given that his final design will be closer more similar to his prior design. Some some another part of work, we’re currently exploring is a it’s related to lead. Users lead users from the academic perspective. Are people that face leads that others will focus in the future and also they are typically people that will benefit more from proposed solutions and why these people might be interesting is that in open collaboration, environments like king, diverse people and the a fraction of the people Contribute the majority of the content so Wikipedia 1 % of the editors contributes more than half percent, more than half of the content and in thinking verse, 5 percent of the users contributes about five of all designs and usually what where people would focus in academic papers? At least, is there now the contribution, so how many designs do a just a number of design someone has submitted in total, which usually is a good metric, but, as you can see, it is not very accurate, so we we’ve done it to ition. We we realize that that might not be the best metric to have so, as you can see, some of the people keep showing up in the top 10 list in all categories, but not everyone is not.
There is no one that is always top, or so we wanted to find a more sophisticated, a metric for how we can actually evaluate the contribution of someone in an open collaboration environment like Thingiverse so to test whether contribution would be a good metric. We classified users depending on how often they would contribute so blue these people, but on average they would contribute the design every three weeks or less, and people with under orange are all the people that followed a much longer production cycle and, as you can see, people That would follow a much longer production cycle, we’re performing much better so but but but kind of that it justifies why a more sophisticated evaluation might make sense. So from these five attributes, if we go get them got down to 22, you can see that they’re a bunch of users but distinguish themselves from the pack, and also we were interested in understanding whether designers we would be able to identify lead users early on, and This is just their presentation or in 3d of well the metrics I showed you before and on the left upper left side. You can see that what we did was basically just take the very first design of a particular users and then classify them in three categories.
So people classify this green where people, but the very first design performed quite well yellow, was about average and red was and, as you can see, especially people that didn’t perform very well and easily we were, they started. Stop this little big quickly stopped contributing. So, as you can see, after friendly designs, we are 50 designed a people classified as green or high performance in the first design, basically take over the contribution. And finally, we looked at designers that had at least 25 designs and classified them again as with whether they’re popular design, the very first design, was popular or unpopular and, as you can see, although there it seems to be a pattern, but people are start cutting up.
So there’s a learning and involved in the community even after 25 designs and people that do you perform well in their first design, still didn’t cut up with a better designers initially. So I our work focuses on predicting design success early on so understanding just by the attributes of the design, whether it will be successful or not. We also want to build a system that will suggest remixes to potential makers and inventors and finally identify early on and or lead users. I think that would be all any questions: okay, so yeah! Thank you! Ah! Well, we oh yeah, so the question was: if we have used our metal did anything else besides thinking us, so then novelty metrics, because it’s based upon 3d designs and it is difficult to be applied somewhere else, because we need to also have success metrics.
So it’s kind of limited in any digital format of of an object. So for that for know, we we’ve been looking. Neither open innovation and a communities like github and some patterns exist, but we’re early on in our analysis. So I cannot really say yes or no yeah, it’s a good question so and today yeah.
So the question was whether we have checked other 3d printing communities like safe ways. So safe ways has some of their designs as freely download a so people can freely download them, but doesn’t support remix ink. So it’s difficult for us to keep track of what’s going on there, and on top of that, the vast majority of designs on segways are not even available. You just order the 3d part and you get the physical object so yeah that it was something we’re looking at, but we haven’t done it so far, yeah well, it has the yeah, probably their largest repository.
Although thing givers now with the customizer has a lot of designs, also, but think of us by far – is the most open largest, often the repository, so that was what we were interested. Okay, any other questions. Well, quality will be probably difficult to justify. That’S why we were interested in novelty and what wanted to compare it to any other given design.
However, there is a pattern in the complexity, meaning that designs designs that had more parts also tended to be highly likeable, but, however, there was they, they were less made. So people tend to also make those designs list because probably of the complexity of figuring it out later. So, thank you very much. You .