Initially, I tried fetching website stylings straight from archives like the Wayback Machine. Then I tried creating an algorithm to interpret frontend elements in terms of older design standards. However, both of these approaches meant that new websites would take a while to load, and there would often be some website or nested feature that the extension couldn’t interpret well. These were issues I encountered with other retro-style extensions as well. To get around both of these issues, I decided to use opus 5.5 as a tool to search for past screenshots and individually implement designs for high-traffic pages over the course of a month, with me acting as an iterative tester to further review each page against its 2019 captures and iron out any aesthetic or functional issues.
I mainly targeted the 100 most-visited websites in the US, as well as the top social media, news sites, Wiki pages, Google apps, and ex-US-dominated websites. I excluded banking websites and websites that look similar to how they did in 2019. I know 2019 might not be peoples’ favorite year for many sites, which is why I think it would be cool if motivated people made extensive layouts for other years as well. Though this extension works well for my daily use, there’s bound to be issues and inaccuracies I missed that I’m optimistic can be patched with outside feedback.