• July: Leo finds out SDR in the cost model is more resilient than we thought
• New analysis
• Corollary: you can remove a few top layers without losing anything
• August: Matteo’s audit/summary of Leo’s proof
• available here
• From there on we targeted better constructions:
• in the latency model
• faster decoding
• we targeted (with limited success) the problem of SealStacking that comes with faster decoding
• September-November (main highlights)
• Looking more at DRG (for latency model):
• Starting conversations with Susanna de Rezende
• Alternative constructions for fast decoding:
• We looked at DRG-PoRep from Proofs of Catalytic Space (also in Ben’s paper)
• These constructions are not good enough.
• On Single-Layer DRG approaches
• Other approaches we half-explored:
• Experimental approaches to analysis of graph based PoReps
• Source here
• We used it to verify Leo’s results. (Done)
• But this approach can be promising for future designs and their exploration.
• New techniques Leo for analyzing SDR in the latency model
• we never went too far with this. They still have potential. We just did not explore them enough.
• November/December:
• Improved rational analysis
• Leo started adding storage to the rationality-based analysis of PoRep
  Better Cost Analysis via Adding Costs of Storage and Computation
• Motivation on analysis of studying DRG/Pred robustness
  Why Better Predecessor Robustness or Depth Robustness are Crucial

Current status (as of January 2024)