OpenNTPD hit the OpenBSD CVS in late 2004. Now, 20 years and a few milliseconds later, it's time to look back, how it was received, and what changed.
We'll briefly look at the world 20 years wrt having systems' clocks synchronized (or not), the design of OpenNTPD and how it fundamentally differed from everything around back then, and how it was - intentionally or not - misunderstood by some with very little knowledge but a lot of opinion, and how FUD from 20 years ago is still around. We'll cover what changed in OpenNTPD after it went public, and what changed in OpenBSD to increase accuracy substantially. We'll also cover the later added constraint feature to further defend against getting fed incorrect time.