Supply Chain Data Exchange Standard (SCDEX) is the joint project of Mapped in Bangladesh, Wikirate and Open Supply Hub established this year to collectively build an open exchange standard for global supply chain data. In September, they launched the SCDEX podcast: a series of public research interviews the world’s foremost experts across policy implementation, technical standards writing, supply chain transparency, open data platforms, and other stakeholders, to examine the different perspectives and incentives critical to creating a standard that is useful, usable, and currently missing in the supply chain data exchange ecosystem and together discuss and dissect how to best fill this gap.
Curious? Here’s a taste of what you can expect from this series.
Shuya Gong, ecosystem and partnerships manager of the SC DEX, sits down with Open Supply Hub’s Chief Technology Officer, Kate Chapman, sharing how her 15 years of experience in open data led her to supply chains and how complex webs are behind even simple products.
The episode touches on the topics like:
- the right level of complexity for supply chain maps
- why the exchange standard is like an electrical plug adapter
- deep dive into the technical infrastructure needed to co-create the global supply chain data exchange standard.
- why having a dog is helpful when mapping open data
Read the podcast highlights below or listen to the full episode on Spotify.
In one sentence, what do you find interesting about supply chains?
They shouldn’t work, but they do.
Let’s unpack that. Francizca [previous podcast guest] said, “If you haven’t found the risk in your supply chains, you just haven’t found it yet”. Tell me more about why they shouldn’t work and what’s keeping them together right now.
Well, the other way to think about it is, are supply chains the original distributed system? The fact that it’s all these one-to-one relationships that create a web and then you end up with this really complex product. How does a car even manage to exist if you think about how complex it is? And even simpler products still have that web. It’s seldom one-to-one-to-one-to someone who bought it.
From the perspective of Open Supply Hub, which has been collecting and growing this database of production locations, what is the interest in an exchange standard? How does it start to extend the impact that is already going on because of the existence of OS IDs?
I think the big thing is not duplicating the work. I don’t know what the percentage is, but the majority of the information, the base level information about supply chains, is on some corner of the internet, but how hard are you gonna work to pull it out?
The amount of work that would be required for all of us to do that, whatever that data looks like, it just really seems like a waste of effort versus if Open Supply Hub has really good production location information for the apparel industry and someone else has very good for electronics or agriculture, or maybe just a single country’s worth or some other aspect, we can get to having that full picture a lot sooner.
What are the misconceptions that people or organizations have about data standards or even just sharing a data standard?
I think it goes to both ends of the spectrum. Sometimes people want to oversimplify it, which, I mean, you can have a simple standard, but sometimes the process of getting whatever the type of data is into that standard can be really difficult. And then the other side is making the standard so complicated that no one will ever want to use it. There’s the quote about a one-to-one map of the world and we spent all day folding it because, you know, it covers the world. And when I think of standards, I worry that we’re trying to create a one-to-one map of reality from our standard because those to me become unusable.
The systems are always complex, but the thing we can try to control for is where that complexity is. At a base level, where things are made and who owns the places making those things, are really the two crucial parts. Then there’s the discoverability and the exchanging of data.
That’s not to say that it solves any of the data needs that are specific to things like forced labor or deforestation, but that can serve as a base. Personally, I think one way of moving some of the complexity around is to have specific modules or libraries or add-ons for some of these other pieces. If you’re entirely focused on deforestation, you’re not going to probably care about the data that doesn’t relate to that.
If you were in that seat of technologists advising or seeing this policy implementation being put into place this year and over the next couple of years, what are some pieces of advice you have for data teams? What would you share with them?
I think you need a reference point of where the data came from. For the example of whether this field has been deforested, are you using imagery to draw the boundary, or did someone walk it with a GPS? How accurate is that? Then, if you’re summarizing it in some way, what does that mean?
I also think about the ownership component. In terms of accountability, there’s that location component, but then there’s who owns it. Because if you don’t know who owns it, then to what end have you figured things out?
If you’re observing the Earth, like from satellites, there’s a lot you can determine from them. I mean, you can look at buildings, but you can’t usually tell what’s manufactured in those buildings unless there’s a very particular manufacturing process. And the same with ownership. Unless someone spray-painted the name of the company on the roof, you probably don’t know who owns it either.
There are data collection methodologies you can use. Some of them are the same and some are different. The one commonality is having unique identifiers. It’s crucial no matter what because to me that’s a shortcut to “Are we talking about the same thing?” If done well, we don’t have to worry about names in different languages or different resolutions of data. We could just say, I know something about this ID, whatever the thing might be.
I’m curious how working in the civil society sector has changed your approach to building technology. What matters about building it in a way that’s really useful, and who are the people that are really being served as we’re thinking about the way to set up and structure the standard?
It’s really about being clear who your primary users are. I don’t think of a data exchange standard as something we’re building primarily for civil society. I think about products you would build that utilize the standard need to be usable.
What are the things that influence the standard? I think one of the hardest points is that things need to be multilingual. A lot of times in technology, we just assume we can use English for everything. With programming languages, for example, people learn English to get better at developing software because there is a lot of content available in that language. To be inclusive and effective with data, we just can’t do that. And I also don’t think that automatic translation is the right thing.
One of the projects I think about when it comes to how much data you need for the standard to be usable is when I worked on a program where we had a lot of scientific models about what would happen if a disaster happened—a tsunami, a volcanic eruption, something like that. And so we had these complex scientific models, but how do you make them accessible to disaster managers? And how does actual action happen? The other data that we didn’t have at the time was good road information or building information. One of the reasons we didn’t have good building information was the survey that civil engineers needed to do to determine if a building would fall down. It was like 100 and some odd questions. And an expert had to do it. So we went back and we tried to get four. What are the four questions to get an approximation of a city what would happen, not will this specific building fall down? In the end, we got to five. That felt pretty good. Then we could combine the model with this building data to say, OK, if this disaster hits here, this is what the situation is going to look like. And to prepare, this is how many tents you need, this is how much water you need. To make it usable, it was actually a simplification process versus having as much data as possible.
What are some of the back and forth about “we must be very stringent about open data” versus being okay with reporting private data sets to get insights from? That is quite a common misconception around sharing my supply chain data – if I’m transparent about where I get my things from, I’ve lost my competitive edge.
I think there are different angles to it. There’s the data that people don’t want to share because of what you’re saying, a competitive advantage. I think about when I used to work in Homeland Security-related efforts in the US and people would say, well, if we stick this oil dike on a map, then the terrorists will know where it is. If someone wants to disrupt something, I’m pretty sure they know where it is. There’s that type of concern about open data. But then there are the very real individuals having the right to not be public on the Internet. When you get down to the beginning of supply chains, a lot of times smallholders are based in their homes. We have to think about the privacy side of disclosing such information openly while still being able to report for due diligence legislation.
How do we talk about data exchange standards to everybody who’s involved in this process? A conversation about building this in a room full of engineers is very different from a conversation in a room full of policy implementers. What does it look like and feel like to bring all of those people together?
Well, I think it ends up feeling like it’s slower, but it’s better. So, is it really slower?
I think about the example of automatic soap dispensers that don’t work with dark skin because the whole team had light skin. So imagine the equivalent of you getting all the way to the automatic soap dispenser in the world, and realizing we left out the majority of the world. That doesn’t work. We just missed this giant thing. People think it’s slower to build more inclusively, but really you’re de-risking things. So you don’t make that terrible error at the end. I think it requires a lot of patience, but also thinking about that end goal of wanting to do something effective as well.
Access the full SCDEX podcast episode on Spotify. Learn more about this project on their website and make sure to follow them on LinkedIn.
OS Hub is a non-profit platform that relies on philanthropic support to sustain the world’s most complete, open and accessible supply chain map. Join us in powering the transition to safe and sustainable supply chains by making a donation today.
