What is a good contract?

In the traditional English-law view, a good contract gives effect to “the reasonable expectations of honest [people].” Let's put aside the obvious caveats here of contracts that give effect to unreasonable expectations or those drafted on behalf of dishonest people and see if we can find some way to objectively quantify what a good contract looks like.

In machine learning we are always looking for an objective function. A singular value to maximise in order to guide an automated process. The dream is that when a lawyer is presented with two versions of a contract with similar effect and a similar balance of interest, we can have a quantitative way of arguing which one of the contracts is better. At the moment such a singular metric remains elusive.

Below are some metrics from business we're already using, or considering, at Kyra Law.

Business indicators

Boring from an AI perspective, but the bedrock any fancier metric should support: slow, sparse, and hard to optimise against, yet they matter most.

Happy customers

So far ours have been happy, bordering on ecstatic, partly our pricing, partly our blend of lawyers and AI. If that ever changes, it's vital feedback.

Returning customers

The best feedback comes from our clients' wallets. A client returning for more means we're doing something right; one who holds off tells us just as much.

Request for clarification

A good contract can be read by any reasonable person. Every time a client comes back for clarification, either we or the legal system failed that bar.

No contest signing

There's always more than one party. Clients often prefer a fast resolution to haggling, so a no-contest signing by the counterpart is a win.

Analytical measures

Now the fun stuff: metrics we can optimise AI for without input from client or lawyer.

Soundness

No internal contradictions, which breed unpredictability, conflict, and even nullity, severability clauses notwithstanding.

Completeness

It covers all reasonable eventualities while staying legible.

Balance

Not every contract is balanced, nor has to be; parties often accept imbalance to smooth a wider arrangement. But severe imbalance can signal dishonest intent or mis-drafting.

Brevity

Shorter is better. A kind of legal keeps drafting and comprehension costs low while keeping outcomes reasonable.

Commonness

Common phrasings are understood by all, so well-worn clauses should beat esoteric ones, though common usage is no proof a clause isn't absurd.

The weighting problem

Even if we could reliably produce all these metrics (we can't, yet!), combining them into a single number to prefer one outcome over another is a different ballgame. Building a business that puts them together so everyone wins is the hard part, and that's what we do at Kyra Law: lower costs for clients, better conditions for lawyers and, occasionally, an easier sign for a counterparty.