I have a slightly different perspective on this part. The work within vprogs (especially @hus_qy and I) is well-coordinated, although it's still in a somewhat "exploratory" phase, resulting in low coupling (e.g., he's building a comprehensive cross-vprog execution runtime; I'm exploring the underlying Cairo to understand the specific forms of virtual machine instantiation). Integrating elements from previously known Sparkle implementations with vprog research concepts is indeed the harder part, and coordination is still lacking, but we're working in that direction. We had a great discussion about this yesterday on Discord and privately. I'm happy to say that the two perspectives are rapidly merging, and we share the same grand vision: building a fully composable ZK program space.