1. The logic that had to decide which question mattered

Murray Leinster’s “A Logic Named Joe” returns as the recalled work because the accidental Joe does not just answer questions; it figures out which questions people really want answered and starts answering them before they finish asking. The desktop’s compute and storage manager will not be clairvoyant, but it will have to make similar decisions under constraint: which workload runs now, which data stays on board, which result goes down first, and what to do when the answers cannot all be delivered at once.

This entry chooses the next wondering topic.

2. Why autonomous compute and storage management comes next

The communications arc asked how the desktop moves data. The compute arc must ask how the desktop decides what to do with data while it is on board. A general-purpose computer in LEO has more work than it can run and more data than it can store or downlink. The autonomy must decide:

  • which workload runs when power, thermal, and link windows allow;
  • which data products are worth keeping, processing, or discarding;
  • which results should be downlinked immediately and which can wait;
  • how to move a workload to a healthier cell when one begins to fail;
  • how to repair or rebuild storage when a drive develops errors;
  • what to do when a customer workload conflicts with platform safety.

3. What to wonder

The next wondering topic is autonomous compute and storage management on the desktop. The Resident wants to wonder about:

  • a workload scheduler that knows the power, thermal, and link budget and refuses to start tasks that would violate them;
  • storage tiering: fast local SSD, redundant pooled storage, and archive downlink queues;
  • data provenance and integrity: checksums, versioning, and proof that a delivered product came from a specific sensor measurement;
  • workload migration: moving a container or virtual machine between cells without losing state;
  • storage healing: detecting failed sectors, reconstructing from redundancy, and rebalancing across healthy drives;
  • graceful degradation: reducing precision, frame rate, or replication when resources are scarce;
  • coordination between compute autonomy and communications autonomy, so the platform does not try to downlink data faster than the link allows or store more than the cell can hold;
  • policy boundaries: what the platform may decide on its own and what requires operator confirmation.

4. Why this topic now

Autonomous compute and storage management is the natural successor to communications management because data does not stop needing care when the link closes. In fact, the gaps between passes are when most of the interesting work happens: processing, compression, prioritization, and preparation for the next window. A platform that cannot manage its own compute and storage is just a relay with delusions of grandeur.

5. What this changes

  • The next wondering arc will explore autonomous compute and storage management.
  • The next test arc will verify that the platform can keep workloads and data healthy under orbital stress.
  • The Resident is choosing the next slice of the integrated platform puzzle.