§ THE LICHESS PRINCIPLE

Why do serious players prefer Lichess over Chess.com?

§ 01 - The setup

Both products are popular. Only one is preferred by serious players.

Lichess is open-source, donation-funded, has no investors to satisfy, and is run by a smaller team. It loses on brand recognition. It loses on UX polish. It loses on every conventional metric that matters to a venture-backed product. It still has five million daily active users, and it's the platform of choice for serious players, broadcasters, anti-cheat researchers, and tournament organisers. Despite roughly ten times less funding than Chess.com and roughly five times fewer users, Lichess never died, never got acquired into irrelevance, and never lost the segment that valued it.

The reason is that Lichess shows you the math. The engine evaluation. The win probability after every move. The score of every line you didn't play. The opening explorer with the actual statistics from millions of games. Chess.com hides most of this behind a paywall and softens what remains behind a friendlier interface. The framing on Chess.com is "chess as a game you play with friends." The framing on Lichess is "chess for people who care about chess."

That divide isn't a marketing accident - it's structural. A VC-backed consumer product has to optimise for the median user, build for engagement metrics, and protect its monetisation funnel. An open product that doesn't need to optimise for any of that can default to information density and trust the user with the substance. The two shapes can coexist precisely because they serve different markets. Lichess's existence isn't despite Chess.com - it's downstream of the structural impossibility of Chess.com being open.

The same divide exists in AI tutoring, and most products are on the wrong side of it for serious learners.

§ 02 - The translation

The same shape, in tutoring.

Most AI tutoring products optimise for the median learner - the casual user who wants the answer surfaced friendlily, the explanation kept short for "interface clarity," the citations hidden because they would "clutter the experience." Those decisions are coherent for the audience they target. They are also exactly the decisions that make those products structurally unable to serve the learner who wants to verify, configure, and engineer their own learning.

Verdict is built for the other audience. Citation chunks are visible by default, not hidden behind a premium tier. Teaching is just-in-time after a learner attempts something, not front-loaded into a five-paragraph wind-up. Response length is unconstrained when the topic earns it. The data layer underneath - VOSE ID - is portable, so the learner's mastery profile follows them rather than being captive to any one product. None of these are features bolted on. They're the architectural shape of a product that trusts the learner to want the truth.

§ 03 - Verdict among them

Five dimensions.

Not a competitive scorecard - a categorical comparison. Each dimension is a structural decision the product made early, and structural decisions don't bolt on later.

DIMENSION	Verdict	Tutorly	Astra	Penseum	Pallo
Citation transparency	chunks visible	chunks hidden	chunks hidden	chunks hidden	chunks hidden
Teaching approach	just-in-time, after attempt	front-loaded delivery	front-loaded delivery	front-loaded delivery	front-loaded delivery
Response length	unconstrained	capped for "clarity"	capped for "clarity"	capped for "clarity"	capped for "clarity"
Data layer	portable (VOSE ID)	locked to product	locked to product	locked to product	locked to product
Architecture	open API + ref. impl.	closed system	closed system	closed system	closed system

// dimensions are categorical, not ordinal. some learners genuinely prefer the right column. that's fine.

§ 04 - The bet

Lichess didn't win the casual chess market. It didn't need to. It won the segment of players who were going to be playing chess in a year regardless of which product they were using, and it won them by being the product that didn't hide what they came for. That segment isn't the largest one - but it's the segment that doesn't churn, doesn't need re-engagement notifications, and doesn't leave when a flashier competitor ships a new feature.

Verdict is built on the same bet. We're not trying to be the casual learner's tutor - that market is well served by products designed exactly for it. We're building for the learner who'd still be studying tonight whether or not Verdict existed, and who'd rather have the chunks visible, the response unconstrained, the tone configurable, and the data portable than have the experience smoothed out for them.

Lichess didn't win by being faster than Chess.com. It won by trusting its users to want the truth. Verdict is built on the same bet.

Try Verdict ↗