**Parent Topic**: [[Software/README]] ## The Trade Nobody Sees Cook: a reliable system's low accident rate encourages change — especially new technology — to reduce the *high-frequency, low-consequence* failures. But "these changes may actually create opportunities for new, low-frequency but high-consequence failures." New tech used to eliminate well-understood failures "often introduces new pathways to large-scale, catastrophic failures," and these new catastrophes can have *greater* impact than the ones eliminated. ## Why It Hides The new failure modes are "difficult to see before the fact; attention is paid mostly to the putative beneficial characteristics of the changes." And because the new high-consequence accidents are rare, **multiple system changes may occur before an accident**, making it hard to attribute the failure to any particular change. ## The Implication Improving the common case can quietly buy a worse tail. This is the deep reason maintenance and upgrades are dangerous ([[Every Maintenance Activity Is an Experiment]]) and why [[Check the Change List First]] is the right reflex — though Cook's warning is subtler: the contributing change may be several deployments back. Pairs with [[Catastrophe Requires Multiple Failures]]. --- *Source: [[Web Operations]] (Allspaw & Robbins, O'Reilly 2010) — Ch 7 — How Complex Systems Fail*