Last night before bed, I was checking on-chain data, and even though blocks were being produced, the dashboard seemed to freeze for two or three minutes... Many times, it's not that the chain is "slow," but that the layer you're viewing is gasping for air: Subgraph indexing needs to chase blocks and parse events, and when encountering node hiccups or reorganizations, it will roll back and rerun; the RPC side may also be rate-limited, and if you make too many requests, they get queued or dropped, so the frontend caches a bit, making it look like "suddenly no updates."

That's why I now get used to opening two sources to cross-check when viewing fund flows, so I won't be misled by a single chart.
By the way, L2s are arguing daily about TPS, fees, and subsidies; if the data pipeline isn't properly maintained, no matter how high the TPS is, it can only run on posters.