I’m going to be grumpy today and complain about something trivial: I’m frustrated by how much AI is touted as the ultimate multiplier to any business without explaining what it actually takes to get good results. It takes hard work, work that has been established as good practice and followed by the top software shops for decades. And yet from what I gather, this is exactly the kind of work most eager AI buyers have been avoiding since well before generative AI was a thing.

I would like to invite everyone to my talk at the next Wrocław Java User Group meetup in Wrocław today. My presentation titled “9 niepokojących publikacji o architekturze oprogramowania. Numer 7 zaskoczy nawet Ciebie!” (“9 disturbing publications about software architecture. Number 7 will surprise even you!”) is a literature review on some aspects of architectural planning is IT software projects. I hope to see you at the meetup!

A single, well-considered choice in one component of a distributed system can significantly impact its efficiency and behavior. At Storj, a decentralized object storage platform, this principle is demonstrated through the use of Bloom filters for garbage collection. This post is based on my notes and observations as a small-scale contributor and active community member. I found this case study interesting not just for its set of trade-offs, as is typical in system architecture, but also for demonstrating the necessity of intentional design.