Thinking in Tails

18 Sep 2026 - Conan Mercer

Avoid the Average

Of late, I have been thinking a lot about averages. A lot of my work is based on statistics, finding patterns in data, and using the knowledge gained from these insights to help improve the efficiency of software and hardware at scale.

Tail analysis is taking data and looking at the extremes, or the tails, of that set of data. In Site Reliability Engineering, this is of more value because we care about the customers who are not getting the best performance or experience. If we focus on them, the tail subset, we can improve their experience and, as a consequence, the tide will raise all boats. So the averages will also improve.

Because I also read and write a lot (text, as well as bytes), I have noticed that a lot of content online these days is AI generated. To me, it is quite obvious when I read an AI generated piece of text. Immediately, this devalues the text for me. AI will produce an average response, it always will.

Knowing this signals two things to me: what I am reading is of average quality, and that the author, did not put much effort into it.

When writing anything of substance, important, or critical, my advice would be to write it yourself. In a world of AI generated information, this is the easiest thing you can do to stand out from the crowd.