Three replicas each keep a slot of a shared counter, and the counter shows the sum of its slots. Replica A holds 4, replica B holds 1 and replica C holds 6. What does the merged counter show?

Replicas receive the same updates in different orders. What makes them converge?

• A merge that is commutative, associative and idempotent
• A faster network that delivers every update in the same order
• A leader that collects every write and reorders them first

With a concurrent add and remove of one element, the merge rule decides the outcome.

A set uses add-wins. An add and a remove of one photo race. What does every replica show?

• The photo is absent everywhere
• Each replica keeps its own answer
• The photo is present everywhere

Replica C sends its slot value of 6 to a replica. The same message arrives twice. What does the duplicate do?

• Nothing, the slot still holds 6
• It raises the slot to 12 and doubles the count
• It breaks the replica until a leader repairs it

Why can these replicas accept writes with no coordination?

• Writes are held in memory and never stored durably
• Conflicts are settled later by the merge
• Networks never reorder or drop any of their messages

An app keeps a set of allowed contacts. Blocking someone removes them, and unblocking adds them back. A block and an unblock race. The person must stay blocked. Which merge rule fits?

• Add-wins, because the unblock keeps the person on the list
• Remove-wins, because the block wins when the two race
• Either rule, because both give every replica one answer

Kim says her merged set always matches what the users intended. What is wrong?

• Merges are too slow for users to ever notice a result
• Replicas that merge can never actually reach one state
• The rule picks mechanically and may surprise users