Courtbot at Brigade Congress 2018




This is a transcript of some notes taken during the courtbot session during Brigade Congress, 2018. Some modification in an attempt to increase clarity. Sometimes the names were not known. Anyone can feel free to add them in and/or give feedback. If we want something more accurate, it may be useful to watch the vids.


  • Pins in slack channel #courtbot to discourse
  • Place to get code source is unclear, but being worked on

To Overcome

  1. Support in community [how?]. State data not there.
  2. Code.
  3. Climate of adoption


1. Community

1a. Buy-in

Carlos: PD (public defender’s) office came to them. Court is at county and state level. Oklahoma is unique - court cases are through one website. Data: Through PDO - County court clerk knows data. Marketing: Churches and non-profits is where the audience is.

Tulsa: University law program.

Vermont?: Non-profit legal service provider (connected to the state).

1b. Partner discomfort

Alaska: Problem: Long term sustainability/cost. Solutions: Not resolved yet

Tulsa: Not a big problem there - Report from NY said formal courtbot needed. Brigade offered support when needed, like helping to train people. State payed for service (formal paperwork).

1c. Pre-existing motivation for tool

?: Failure to appear data. When person fails to come, person doesn’t pay fines and county doesn’t (was this supposed to be ‘does’) pay for more court time.

Alaska: Hangout - FAQ wording right - tool(?) problem is not excuse for missing court. (Not sure what this was about)

Tulsa: Behavior insights: tried different types of message - kind or harsh. Those were the only wording questions.

Vermont: The first priority is to get the right partners, otherwise will miss the mark

2. Code

2a. Where’s the data?

?: Data not available - costs money

Tulsa: Lots are different state-wide. Just have one county now.

Oklahoma: Same problem

Tulsa: [Regarding using a call-in system with the case number as the id] Can’t have just one phone number because court case numbers are repeated and can’t have non-readers type in letters.

NC Durham: Has web portal

Tulsa: ACLU, etc. doing web scrapers for this. (ACLU is very helpful)

Vermont: Boise can get good, clean data. Sometimes need to make scraper.

2b. Why not reuse code others have written?

Sacremento: Way of grabbing data is different. After grabbing data, processing should be the same, but existing repos entangle those parts.

Tulsa: Built on early version of node. No application workflow. Working on describing that. Maybe docker container.

[Note from later] Tulsa: Atlanta version tangled with paying with phone.

Conclusion: A lot of exploration work left to do

3. Deployment/adoption/publicity

Tulsa: Bail bonds, defence attourneys, non-profits - not successful. Lost our champion. Rebuilding relationship. Advise finding internal person (county clerk is useful). Promotion [publicity] mostly through partner.

Durham: Put app on county website. SEO. Multilingual.

Alaska: Looking at instructions on page. Case # is long to type in.

4. Extra Questions

  1. Which people choose to work on it?
    1. Vermont: Lawyers and lawyer programmers
    2. Tulsa: Lawyer came to talk and light fire. Were able to ask Alaska for help. Atlanta version tangled with paying with phone.
  2. What does the app look like?
    1. Vermont: Propose demo on the web
    2. ? (Matt): Dial phone number, put in name and case #, get back date, time, location.
    3. Alaska: That and offers to remind user. If can’t find #, option to watch for # for 10 days.
    4. Durham: Want questions for transport, child care, whether told employer or not [I believe they want to cue the user to think about and plan for these obstacles]
  3. Success?
    1. Tusa: 18% less failure to appear or pay fines. 10 - 15 people per day [call in, or is that the number of additional people that manage to appear?]
  4. Want hard data/statistics
    1. Tulsa: Will share papers
    2. Durham: Will share… application…?
  5. There are more folks out there