Hello everyone 👋 I'm developing an engineering strategy methodology, and to publish it, I need a few proof-of-concept examples of it applied to real systems (that are not under NDAs). So I went looking for open source projects to audit — and since I've always known about the famous Solidus vs Spree drama, I decided to make Solidus my first public report. It's built entirely from public evidence: the repo, git history, PRs, the Open Collective ledger, blogs and podcasts. The data sweeping and writing is mostly LLM work, but the reasoning is mine; that's the lever that makes a report like this possible in a day. What the public record shows, in short: the project is in excellent technical shape (CI tested against unreleased Rails and Ruby versions, machine-enforced deprecations, automated backports), but the new admin has absorbed most of the funded development since 2023 and is still at v0.4 with its main screens switched off. The July 2025 status update asked for "another developer to pick up this work" — and since 5 July 2025 no expense has left the collective, while ~$132k sits in it. My bet is that the admin is the most important surface in an e-commerce framework: merchants can rebuild a storefront they dislike, but they can't rebuild the admin — they use it as delivered. So while the new one stays half-built, everyone stays on the jQuery-era one. Also, your public blog has no recent info on future decisions or roadmaps. Before I publish, I'd much rather be corrected than confidently wrong. So, three questions that would fix or confirm the most load-bearing findings: 1. Why was the funded engagement not renewed after the July 2025 status update — no candidate, a deliberate pause, budget caution, or simply nobody proposing it? 2. Does the core team consider the new admin alive — intentionally paced, or stalled? 3. One of my conclusions on the new admin is that Nebulab started from the easy admin screens and then stalled on hard ones, but instead they should've started with the hardest ones to prove it is viable. If it works, it would be much easier to decide on the easy ones or even automate/speed up their migration with LLMs. I'll hold publication for a week in case anyone wants to comment, and I'm happy to share the full draft here first — every claim in it is tagged with where it came from (repo / web / inference). Thanks, and genuinely: the delivery machinery in this project is some of the best I've seen, which is exactly what made the rest of the picture so interesting. You've shipped two big replacements before: the storefront and the promotions engine. So I can't quite figure out what's different about the admin. Here's the draft report (75 min of reading) and a short complimentary slide deck (5 min). You can read them here or download them into a single dir so you can use cross-links from the deck to the report.