Substantive Legal Software Quality - A Gathering Storm?

tags

Quinten Steenhuis Law and Technology

Notes

Externalizing logical rules

NOTER_PAGE: (6 . 0.2857142857142857)

generate automated testing data that follows realistic patterns

NOTER_PAGE: (7 . 0.18909991742361684)

End-users rightfully want to understand the reasons

NOTER_PAGE: (7 . 0.2857142857142857)

After reaching a key branching point, expressive applications should tell the user the impact of the information they just entered.

NOTER_PAGE: (7 . 0.40545004128819156)

Providing interim feedback can keep users from feeling that they are trusting their legal outcome to a black box

NOTER_PAGE: (7 . 0.5103220478943022)

Users of API-driven platforms such as Docassemble can develop test answer sets to validate logic against interviews

NOTER_PAGE: (7 . 0.5450041288191577)

validations can also help subject matter experts catch errors in the logic early. For example, a user may recognize that the system is relying on an answer that they did not provide, or know a real-world fact (such as the fact that the court is closed on the rescheduled date) that the author did not take into consideration.

NOTER_PAGE: (7 . 0.5549132947976878)

web automation platforms such as Selenium or cucumber.io can be used to drive the interview

NOTER_PAGE: (7 . 0.5986787778695293)

code sets that can be reduced to a finite state automaton

NOTER_PAGE: (7 . 0.7943848059454995)

For every possible set of user inputs, the system should behave in a way that a human reasonably proficient in the relevant legal ‘art’ would not find objectionable

NOTER_PAGE: (8 . 0.7241948802642444)

The ability to say “I don’t know” or “I’m not sure” however provides considerable wiggle room.

NOTER_PAGE: (8 . 0.8133773740710156)

there’s a lot of important legal knowledge that can’t yet effectively be expressed in software. We should document and declare all compromises

NOTER_PAGE: (8 . 0.8422791081750619)