Linking velocity matters mainly at the extremes, as an anomaly signal, not as a dial you tune for reward. A sudden, unnatural spike of low-quality links can look manipulative and draw scrutiny, but there is no blessed pace that earns you points for hitting it. So velocity is real at the edges and irrelevant in the middle, which is why both popular positions, that you should build links at a steady cadence to please the engine and that velocity is pure myth, miss it. This is observed behavior rather than a published formula.
Take the spike end first. When a page that has been quietly accumulating a few links suddenly gains a flood of them from spammy, irrelevant, or obviously purchased sources in a short span, the pattern itself reads as engineered. The engine is good at spotting link growth that does not match how genuine popularity tends to accrue, and that mismatch is the thing that can hurt. The damage is not from speed in the abstract, it is from speed plus low quality forming a recognizable manipulation pattern.
Now the other end. There is no evidence of a rewarded ideal speed, no cadence you can schedule that earns ranking credit for its rhythm. A page that genuinely becomes useful and gets cited will gain links at whatever rate its real popularity produces, fast when something takes off, slow when it does not, and both are fine because both are natural. Trying to engineer a steady drip to satisfy an imagined preference is solving a problem that does not exist.
So the calibrated reading sits between the myths: velocity is an anomaly detector, not a tuning knob. Avoid the unnatural spike of junk links because that is the only place velocity bites, and stop managing the pace of legitimate links because there is no setting to optimize.
For your link building, stop trying to schedule a “safe” velocity and instead just keep the links you earn relevant and real, watching only for the kind of sudden low-quality surge, often from negative SEO or a bad purchase, that forms an unnatural spike worth investigating.