As an Ashkenazi Jew, I'm used to matches being predicted to be closer than they really are. My father recently got a match to an adoptee of 348.3 shared cM, with a longest segment of 57.4 cM (according to MyHeritage.com). Because of Ashkenazi endogamy, I've learned not to pay too much attention to shared cM; however, I've never seen such a long segment for someone not already known to be related. Can I "trust" longest shared segments more than total shared segments?
While longest shared segment seems more robust, I know enough statistics to understand why it would vary more than total shared segments. I also see empirically that I share the same longest segment of 57.4 cM, but my total shared DNA with the adoptee is 249.3 cM.
If so, where is the best place to find estimates of relatedness based on longest shared cM? I couldn't find anything at ISOGG, but I did find The Shared cM Project. The latest update appears to be based solely on total shared segments, not longest shared segment, but the 2015 post The Shared cM Project – Longest Shared Segment seems to provide what I want. Blaine Bettinger divides the data into endogamous and non-endogamous categories, using this question on the data entry form to classify people: "Is there any endogamy or known cousin marriages that could impact the amount of shared DNA for this particular relationship?" As a member of an endogamous ethnicity with no known cousin marriages (on that side of the family), I'm not sure which data would apply.
I believe I've identified the birth father, by intersecting the information from the adoptee with my family tree. Consistent with aem's prediction and the linked data, the adoptee would be my father's second cousin once removed and my third cousin.