Verbal Theory Questions in Tech Interviews: One-Sentence Answer Cheat Sheet

Verbal theory questions are interview filler that interviewers use to warm you up or fill time. They don't reward long, meandering answers — they reward one tight, confident sentence. Here are the 30 most common ones with the cheat-sheet answer for each, plus how to handle them in real time during the round.

How to use this list

Skim it the night before. Don't memorize verbatim — paraphrase. The goal is to have a compressed-mental-model so when an interviewer says "quick one — what's a Bloom filter?" you can answer in 5 seconds with confidence and they move on.

For real-time help during the actual interview, use Speaker Mode: press Ctrl+Shift+A to start listening, press it again when they finish — the AI gives you a one-sentence answer you can paraphrase aloud.

The 30 questions

1. TCP vs UDP?
   TCP is connection-oriented with reliable, ordered delivery; UDP is connectionless, fire-and-forget — used when speed matters more than guaranteed delivery (DNS, video, gaming).
2. Explain CAP theorem
   In any distributed data store, you can guarantee at most two of: Consistency, Availability, Partition tolerance — and since partitions are unavoidable, you really pick between AP and CP.
3. What is a Bloom filter?
   A space-efficient probabilistic data structure that tells you definitively if an element is NOT in a set, but only probably if it IS — false positives possible, false negatives impossible.
4. How does HTTPS work?
   Client and server use TLS to negotiate a symmetric session key via asymmetric crypto (using the server cert), then encrypt all subsequent traffic with that key.
5. What is eventual consistency?
   A distributed-system consistency model where reads may temporarily return stale data but all replicas converge to the latest write, given enough time without further updates.
6. Difference between hash table and hash map?
   Functionally the same data structure; in Java, Hashtable is synchronized and legacy, HashMap is unsynchronized and faster — in most other languages they're used interchangeably.
7. When use SQL vs NoSQL?
   SQL when you need ACID transactions, complex joins, and a stable schema; NoSQL when you need horizontal scaling, flexible/sparse schemas, or extreme write throughput.
8. What is sharding?
   Horizontal partitioning of a database where rows are split across multiple servers by a shard key — scales writes and storage but complicates joins and rebalancing.
9. Explain consistent hashing
   A hashing scheme where adding/removing a node only remaps a small fraction of keys (the keys belonging to the affected segment of the ring) instead of all of them.
10. What is a deadlock?
    A state where two or more processes each hold a resource the other needs and neither can proceed — broken by lock ordering, timeouts, or detection-and-rollback.
11. Difference between process and thread?
    Processes have isolated memory and are heavyweight to context-switch; threads share memory within a process and are lightweight, but require synchronization for shared state.
12. What is OAuth 2.0?
    An authorization framework that lets a third-party app access resources on a user's behalf via tokens, without sharing the user's credentials with that app.
13. Difference between REST and GraphQL?
    REST exposes fixed resource endpoints with predictable shapes; GraphQL exposes a single endpoint with a query language so clients fetch exactly the fields they need.
14. What is a circuit breaker?
    A pattern that monitors failure rate to a downstream dependency and, after a threshold, fails fast for a cooldown period — prevents cascading failures in distributed systems.
15. What is idempotency?
    A property where the same operation executed multiple times produces the same result as executing it once — critical for retry-safe APIs and exactly-once-effect semantics.
16. Explain index in a database
    A secondary data structure (typically B-tree) that maps column values to row locations, speeding up reads at the cost of slower writes and extra storage.
17. What is a write-ahead log?
    A durability technique where modifications are appended to a sequential log on disk before the in-memory state is updated, so the system can recover by replaying the log after a crash.
18. Difference between optimistic and pessimistic locking?
    Pessimistic locks the row before reading; optimistic reads freely and only checks for conflicts at write time using a version number — better throughput when conflicts are rare.
19. What is CDN?
    A geographically distributed network of cache servers that serves static (and increasingly dynamic) content from a node close to the user, reducing latency and origin load.
20. How does Redis achieve high performance?
    In-memory storage, single-threaded event loop avoiding lock contention, simple data structures, and pipelining of commands over a single connection.
21. Explain leader election
    A consensus protocol (like Raft or Paxos) where a cluster of nodes agree on a single leader to coordinate writes — handles failures via term-based re-elections.
22. What is rate limiting?
    A mechanism to cap how many requests a client can make in a time window — implemented with token bucket, sliding window log, or fixed window counters.
23. Why use a queue?
    To decouple producers from consumers, smooth bursty traffic, retry failed work asynchronously, and enable horizontal scaling of consumers independently from producers.
24. Explain garbage collection
    Automatic memory reclamation by tracing reachable objects from roots and freeing the rest — generational GCs exploit the fact that most objects die young.
25. What is a race condition?
    A bug where the correctness of a program depends on the unpredictable interleaving of concurrent operations on shared state.
26. Difference between latency and throughput?
    Latency is the time for a single operation to complete; throughput is operations per unit time. Optimizing one often hurts the other.
27. What is a heartbeat?
    A periodic signal a node sends to indicate it's alive — used in distributed systems for failure detection and leader-election quorum tracking.
28. Explain DNS
    A hierarchical, distributed name system that resolves human-readable domain names to IP addresses by recursively querying root, TLD, and authoritative nameservers.
29. What is a Merkle tree?
    A hash tree where each non-leaf node is the hash of its children — used for efficient verification of large datasets (Git, Bitcoin, anti-entropy in Cassandra).
30. Why is database normalization important?
    It eliminates redundancy, reduces update anomalies, and shrinks storage — at the cost of more joins; OLTP systems normalize, OLAP systems often denormalize for read performance.

Tactics

• Don't over-answer. One sentence. Stop. Let them ask a follow-up if they want depth.
• Use a comparison. "X is Y, but unlike Z…" — instantly sounds senior.
• If you don't know, say so. "I'd need to look up the exact spec, but conceptually it's…" is better than bluffing and getting caught on a follow-up.

Live-interview workflow

1. Hear the question through your headphones / speakers
2. Press Ctrl+Shift+A — live transcription appears in the overlay
3. Press Ctrl+Shift+A again when they finish
4. Read the one-sentence AI answer in your overlay
5. Paraphrase aloud in your own words

See Speaker Mode docs for the full setup. Or just try it free with 10 messages and 5 minutes of speaker time.