The goal of fairness algorithms is to be unseen. Benchmarks that sit on send obviously hit them, but we'd hate customers with real work loads to have to think about them, and perf goes up silently with every rev of our Nitro updates.
For workloads that are not sitting on send, there isn't a flat "baseline". The behavior will depend on how many credits have built up and are remaining. Ideally, as for the vast majority of workloads, nothing will happen!
-
-
To mention again, we're always refining these, improving the parameters slightly to accommodate more marginal workloads without sacrificing good fairness.
-
Together these make it hard to document; the behavior is sophisticated, and when we make the specifics public, it becomes very hard to improve for everyone because some folks code or alarm against the specifics.
- 7 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

Your characterization isn't correct; we're transparent with what's going on; it's fairness through a credit system, similar to a token bucket in principle ...