Re: Thinking about how best to leverage our work so far
toggle quoted messageShow quoted text
Dear Chris and all,
thank you for raising these points, I have the impression this is an important discussion for bonsai. I need to say that I was not so active in bonsai because in my view it suffers from two things:
The name is easier, “bonsai” is (again in my view) a procedure to basically torture a tree to prevent that it becomes as large as it should become but keep it in a micro state throughout its life. I did not really see the link to the idea that we want to pursue (of course, “cdlci” is also not too catchy and remembers me more of a Korean microprocessor name).
The dogma is more difficult. I believe that there is a supply chain for LCA models, which goes from data, to tools, and models; and I think these datasets need to be prepared with care and should be important for decision making thus have value. I never understood why it makes sense to declare that one step in the supply chain must be free of costs; it is as if saying one step in an LCA must be emission free, while for the rest environmental impacts can be as they are. I see that this concept is attractive for those who provide expensive software tools (sorry Eric..) but for end users, the overall package, the full life cycle costs if you want, count more. Further, if one step in the supply chain is free, it is more difficult to work on it more than for fun. I think this is especially true for dataset development, which is more “scattered” than tool development. You need more insight in different branches and regions, while for software development, it is ok to develop one generic tool that can calculate and model all kinds of products. For openLCA, we manage to provide it for free but also because we have projects paying us for further development, including our contributions to the various initiatives Chris has cited (the lciaformatter, the reference data, of course the openLCA ontology, and we even provided training to the people who are now developing the new database in Canada).
I am not too aware of many reasonable initiatives for working with data in a smart way (not the micromanagement-heavy, manual work with claimed correctness and claimed consistency in the remodeling project and in PEF, e.g.). And I believe that there are not too many persons needed to really develop something meaningful. Maybe for bonsai an important role could be a governor and middle-man, by developing and promoting a code of practice for smart data development, focusing on the approach and not on the fact that data must be free. For the approach, it could be updates, maintenance, how data is collected, that it is all about modeling and that there is not absolute truth, interoperability, verifiability & transparency, doing it not as stupid as PEF, asf. Even, maybe, the current work in bonsai can be maintained and then be one example of applying the bonsai rules, and other initiatives could be “certified”.
What do you think?
All the best,
Von: email@example.com <firstname.lastname@example.org> Im Auftrag von Massimo Pizzol
My impression is that right now we are still in a very pilot stage, due to various reasons: volunteer work, relatively recent initiative, few meetings, very decentralized and slow decision making structure… (and I don’t mean to say these are bad things, on the contrary). We are trying to get stuff to work. So naturally we lack a final product and the documentation is still chaotic. We haven’t yet reached critical mass yet, but once this happens then “translating our knowledge base into material that is easy for other to understand, benefit from, and use” will be easier - I believe. The other groups in this space seems to be in a more advanced stage. So perhaps instead of waiting to be invited we should be the ones inviting others to our hackathons, probably we have much to learn from their experiences.
From: <email@example.com> on behalf of "Eric Mieras via Groups.Io" <mieras@...>
Although I have not been involved in the discussions lately, I wanted to respond to these observations as I think they make a lot of sense. As you know, we as PRé also started a Community Driven Initiative (CDLCI) together with some of the major database providers to make it easier to make sector and region specific data available. For that, we also collaborated with BONSAI and because we see a lot of value we also sponsored the hackathon. With CDLCI we want to focus on making the core data and methodological/procedural guidance available that everybody needs to build a database. Next to that, we're currently working on a number of projects to make regional databases available by unlocking data that is now only accessible within research institutes and other organizations. So, we're making progress but from time to time I have the same feeling that you express. There's so much to do and only limited time and resources. In my opinion the key issue is that everybody has the same "dream" and tries to achieve that (too much) on its own. If we can combine all the existing initiatives and identify where everybody's resources would be most effective I believe we can make much more progress. For that, we probably have to think about how to organize this ecosystem and what role everybody wants to play in that. If there's no attention to organizing this, the risk is that BONSAI identifies where it can have the biggest impact but that the direction is not aligned with the other actors in the ecosystem that can benefit from it.
I realize this is not a direct answer to your question, but I believe it's important to consider this when thinking about where you chose to focus the limited free time you have available. Hope this helps the discussion forward.
Sent: Saturday, October 26, 2019 23:44
Subject: [bonsai] Thinking about how best to leverage our work so far
Disclaimer: The following could all be incorrect, it's late, I'm tired, people make mistakes, etc.
As far as I know, there are currently three initiatives started recently to build large, open LCI databases: The PEF in Europe, the effort to harmonize LCI data generation in the US (led by the US EPA, but involving many federal agencies, see previous mailing list post), and a new database to be created in Canada.
The PEF was led, or at least steered, by people with a lot of experience in the LCA community, and they followed their habits (and EU political instructions), by buying data from commercial databases (I include ecoinvent here), but requiring data delivery in the ILCD format. The PEF was a bit strange in that each product category went through a separate bidding process, as so each product category database lives on its own. Aggregated data is available, but not under an open license (see examples here: https://ecoinvent.lca-data.com/).
The US EPA has a history of collaboration with Green Delta, built on their use of the OpenLCA software. OpenLCA is attractive both because it is open source and free (not the same thing :), but also because it is easy to deploy to many partners without too much training. As part of this collaboration, they developed the OpenLCA JSON-LD format (https://github.com/GreenDelta/olca-schema), which they use extensively, both in exchanging datasets and databases, but also in their tooling, elementary flow lists, etc. OLCA schema is different than our ontology, but it is a semantic web ontology, and it has great documentation and examples. Development by both US EPA and GreenDelta take place on GitHub, with a core group of a few people contributing most of the work.
The Canadian effort is still under development, but is leaning towards working deeply with ecoinvent, which means not just integrating well with ecoinvent, but also adopting their toolchain for data input and validation, system modelling, etc. One person who is part of this effort told me (and I paraphrase): "I am sympathetic to working more closely with BONSAI, but I don't have anything hard to bring into meetings with others who appreciate open data as a practical good instead of a philosophical one." And they are right! Currently, we can't really point to a finished tool chain, though we are obviously trying to change this as fast as possible, thanks again to the Aalborg crew for the mini-hackathon this weekend.
So BONSAI has effectively zero contribution to any of these three databases, even though all three are or will be open data.
What I want to say is simple: There are a million problems, including a lot of fun ones. BONSAI really suffers from the fact that most of us are volunteers, and as such we want to work on fun problems, and not necessarily doing the second 90% of the work
(https://en.wikipedia.org/wiki/Ninety-ninety_rule) that is needed to bring a tool from proof of concept to something that can be used by many people in actual practice. We have a lot of experience, and have made a lot of decisions as a group in a systematic and organized fashion, but we need to do a better job of translating our knowledge base into material that is easy for other to understand, benefit from, and use as arguments to build systems compatible with BONSAI.
I contend that, after completing the 2019 hackathon toolchain, the most effective use of our resources would be to help others make systems that build on and work with BONSAI. If this supposition was true, how would this change where you chose to focus the limited free time you have available?
Technology Assessment Group, LEA
Paul Scherrer Institut
5232 Villigen PSI
Telefon: +41 56 310 5787