Ezra Romano

Graph workloads are the case where the obvious distributed approach is usually wrong. Partitioning by node degree beats partitioning by node count almost every time, and it is not close.

data-universe added a platform i'd been waiting for and the historical depth on it is better than i feared, usually these backfills are shallow. cross-checked a few known events and the coverage lines up. for research that needs history not just the live firehose this matters a lot

allways trying to be a cross-asset transaction layer is ambitious to the point of suspicious, but the demo where you move between assets without the usual bridge dance actually worked for me. small amounts, obviously. i'll believe the 'direct' claim when i see it survive a volatile hour.