> And what the hell is the “revised standard” that j&j lawyers feel they can now meet? Sounds to me like it could be the engineering equivalent of reducing SLO

> And what the hell is the “revised standard” that j&j lawyers feel they can now meet? Sounds to me like it could be the engineering equivalent of reducing SLO thresholds and admitting defeat either due to not enough traffic or not enough capacity to deal with the problem properly [1]: > You might also want to reconsider if the impact of a single failure on the error budget accurately reflects its impact on users. If a small number of errors causes you to lose error budget, do you really need to page an engineer to fix the issue immediately? If not, users would be equally happy with a lower SLO. With a lower SLO, an engineer is notified only of a larger sustained outage. Without thinking of: > Lowering the SLO does have a downside: it involves a product decision. Changing the SLO affects other aspects of the system, such as expectations around system behavior and when to enact the error budget policy. These other requirements may be more important to the product than avoiding some number of low-signal alerts. The "product decision" is the court ruling and the "other aspects of the system" is how many people will die. [1] https://sre.google/workbook/alerting-on-slos/ https://sre.google/workbook/alerting-on-slos/