Architecture Debugging - Brainstorm
3/14/22
Based on customer feedback, we’ve improved the rubric for Architecture Debugging to more-consistently reward quality of ideas relative to quantity.
Previously, an extreme number of lower quality ideas like “maybe the internet is down” could add up to a score that rivaled a smaller number of very high quality, realistic ideas like “Since only the beta service seems to be impacted, the problem is likely related to infrastructure exclusive to that. Maybe we’re being throttled by the recommendation service”.
We’ve also made some slight changes to the scenario prompt to be more explicit about the system behavior that is broken, specifically that the non-beta service is not having any problems.
Moving forward, you’ll notice more nuanced scoring and new observations on candidates in the analysis.
