A Mastodon tip argues web request rates should use Bq (becquerel) instead of Hz (hertz), because Hz is reserved for periodic phenomena and HTTP requests are random events.
Key Takeaways
Hz is formally defined for periodic frequencies; the argument is that stochastic arrival events like HTTP requests do not qualify.
Bq (becquerel) is proposed as the correct SI unit because it covers stochastic processes, not just oscillations.
The distinction matters for precision: labeling a random-arrival rate in Hz implies periodicity that does not exist in real web traffic.
The source preview frames this as a practical naming tip for engineers reporting server throughput.
Hacker News Comment Review
Near-unanimous technical pushback: Bq and Hz are dimensionally identical (both reduce to s⁻¹), so substituting one for the other changes nothing about the measurement itself.
The BIPM SI brochure does formally distinguish them – Hz for periodic phenomena, Bq for stochastic processes referred to radionuclides – but commenters note Bq’s radionuclide scope makes it semantically wrong for web servers too.
Practical consensus cuts the debate short: “rps” (requests per second) is already the standard in production monitoring, carries the right context, and requires no SI pedantry.
Notable Comments
@manuel-rhdt: Notes Bq doesn’t fix the underlying issue and that real web traffic is bursty, not memoryless Poisson – undermining even the stochastic-process framing.
@maxnoe: Quotes BIPM directly confirming the Hz/Bq split, but highlights that Bq is scoped to radionuclide activity – so neither unit is technically correct here.
@animuchan: “If it’s OK to measure this sound in Hz, isn’t it OK to measure the HTTP requests in Hz” – a concrete speaker-click thought experiment defending Hz.
