 Mozilla has a guide to electronics called Privacy Not Included where they review various different products every year. And you can see this little happy face in the bottom right getting sadder and sadder as we go towards the creepier electronics on the market. Super creepy. So if you look at one of these products, they've done a very thorough write up about why it's creepy, what you can do to protect yourself, etc. And you can also see the scores in terms of level of privacy, security, or AI. What I would love to add to this guide is the products use of energy and its carbon footprint, as well as how repairable it is. So if we think about sustainability of electronics, really the best case scenario is that the device is able to stay in use longer without ending up in a landfill. So I learned about iFixit and the right to repair movement through a podcast and contacted iFixit to see if they had an API. They said that they did, but that the most comprehensive guide to repairability is the French Index de Repairabilité. I'm not sure if I'm saying that right. I'm pretty sure I'm not. And in France, it is law to report the repairability of a device, but this is self-reported by the manufacturers as far as I know. If we look at iFixit's API, it seems like it would be comprehensive, but it hasn't been touched in 10 years. I found it pretty unusable, so I did turn to the French Repairability Index to try and extract data. Now they don't have an API, so I had to scrape the data and I will show you how I scraped that data, but in an attempt to get iFixit data as well, I found a developer who had posted a link in 2021 where he managed to work with this antiquated API to essentially scrape iFixit in order to get repairability scores. And he reported on that. I got in contact with him, his name is Manuel, and I asked him if he could make his code available and he actually did publish it then on GitHub. So if you want to work with iFixit data, take a look at this GitHub repo that I've got showing here. However, I have worked with the Sparica Repairability Index. Sparica is a company that has been supporting consumers to repair appliances with tools and sales spare parts and tutorials. This is their page and I've highlighted a few things here. In France, in 2020, only 40% of electronic devices are repaired. Repairability Index is mandatory by law. There are certain devices that actually just don't have products reviewed. The other problem with this index is that the score is only visible in this image. So if we look at this image, I was hoping that to be able to scrape it, I might be able to see the score added to an alt tag, but you can see our alt tags here are just blank, which is not only bad for accessibility, but it also means that we can't grab that score. So I decided to try and get the score via computer vision, essentially reading this image, looking for this say 6.9 score and translating that into text and incorporating that into some sort of data scraper. And to do that, I used a product called Bardin because I am also interested in looking into emerging low-code and no-code tools to automate tasks like this for things like machine learning or dataset generation, etc. So here we're going to set up our scraper in Bardin. You create what's called a playbook where you use the scraper template. We pick this site we want to scrape, we give the scraper a name, and then we pick an element in the list that we want to compare to another element. So here I'm just picking that title. And now I'm picking the pagination button so it can go page by page. And then we set up the things we want to scrape. So I'm getting the product name, the image which has that repairability score, that's the thing I want. And then we set up a limit and it's going to ask us what our limit to scrape is. And then we want to save it somewhere. So I'm going to click on this Google Sheets. So I've authorized it to access my Google Sheets. It's going to add a row each time to this repairability index sheet. I click done, give that playbook a name. And now I can run it. And I'm going to say I want 20 pages. I didn't actually run all of these in this screenshot. I changed it. It's going to scrape. I'm going to click view. It's not actually that fast. I paused it. And these are our results. I've got the product name, image, category, the date. You can see this one came out a little funny. I had to refine this a bit to get the information I needed. But I eventually worked it out. And this is a lot faster way to scrape data than trying to set up some headless browser and run it using code. That's a nice visual interface for scraping. Okay, so this is the setup I have for getting all of the repairability scores from the images via the Google Cloud Vision API. So here I'm cycling through a bunch of URLs. The Cloud Vision API is returning an array of scores based on the number it detects visually in the image I pass it from the URL. And now in my original spreadsheet, I'm going to just paste that array data. So now I've got scores for all of the electronic devices from the repairability index. So this is a GitHub repo I've set up with all of that data as well as the code I use to generate it, generate the scores from the images, from the Sporica French Repairability Index, which is as far as I know, the most comprehensive index for repairability of electronic devices, making this repo here the most comprehensive open source data set of repairability scores that I know of. That file is set up here as a CSV. You can see it here in GitHub. It includes the brand, the model, the category, the page it was found on, and the vision score that the recognized score from the images from Sporica. And again, to note, Sporica slash French Repairability Index does not label the scores anywhere. They put them into an image. And so you have to run some sort of object character recognition on that image to figure out what the actual score is. So that's what I've got here. Hopefully it will be useful to people. If it is, please let me know. Feel free to use the code. And if you want to collaborate on anything that improves sustainability or repairability of electronics, please let me know.