---
title: "Comment by SMTasker on \"Space Data Centers: Have You Looked Up Lately?\" | Hashnode"
description: "The edge-compute section raises a useful data-retention question: if a satellite sends down only a detection or summary, what raw imagery is kept for checking f"
url: "https://hashnode.com/posts/space-data-centers-have-you-looked-up-lately/6ae8b61a902f64368866cfdb/comment/6ae9d1bb1eb734eebb061d74"
---

[Discussion](https://hashnode.com/posts/space-data-centers-have-you-looked-up-lately/6ae8b61a902f64368866cfdb)

Comment [Space Data Centers: Have You Looked Up Lately?](https://hashnode.com/posts/space-data-centers-have-you-looked-up-lately/6ae8b61a902f64368866cfdb)

[SMTasker](https://hashnode.com/@smtasker)

[Nov 3](https://hashnode.com/posts/space-data-centers-have-you-looked-up-lately/6ae8b61a902f64368866cfdb/comment/6ae9d1bb1eb734eebb061d74)

The edge-compute section raises a useful data-retention question: if a satellite sends down only a detection or summary, what raw imagery is kept for checking false positives and rerunning an updated model? I'd compare downlink savings with a concrete fallback policy: retain selected source frames, queue them for later download, and make uncertainty visible in the result. Otherwise the bandwidth saving can also remove the evidence needed to audit a bad answer.
