Round 2 / Filecoin Spark

This project is important for the Filecoin Network because it is focussing on a huge pain point for Filecoin - the retrievability of data. Verifiable retrievability is a massive hurdle that Filecoin needs to solve in order to find product market fit. However, this is a hard problem to solve in the trustless setting. Spark is currently the most advanced initiative in trustless verifiable retrievability.

The retrieval success rate of public data stored on filecoin through the FIL+ program was hovering around 1% when Spark measurements began in April 2024. Since April 2024, and in particular since Spark data started being used by allocators in data cap allocation pathways, Spark has measured a more-than-10x improvement in the retrievability of FIL+ data stored on Filecoin. (https://blog.filstation.app/posts/spark-sees-10x-improvement-in-filecoin-retrievability)

In April 2024, no SPs had a weekly Spark Retrieval Success Rate above 90%. At its high water mark so far in October 2024, 77 SPs had a weekly Spark Retrieval Success Rate above 90%, with a handful of SPs over 99% on the weekly score.

Once clients can be confident of the retrievability/availability as well the durability of data stored on Filecoin, we believe we will see many more customers use Filecoin as their storage layer.