-
Notifications
You must be signed in to change notification settings - Fork 3
Roadmap
Ozzie Gooen edited this page Apr 10, 2018
·
14 revisions
Note: This is still very much in flux
Epics:
- Budgets V1
- Create 1-2 trees
- Admin View V1
- Budgets*
- Caching**
- Blog post that displays trees.
- Pointer Lists
Epics:
- Refactor (*)
- Episodes V1
- Versions**
- Pointer Unlocking**
Epics:
- Deployment
- Workspace Diffs (show users what imports are stale and can't be relied on)**
- User Accounts
- Smart Episode Scheduling V1
Epics:
- Smart Episode Scheduling V2
- Long texts can be converted into pointers in top-level workspaces
Tests:
- Do 10k+ tasks in small study, for machine learning tests.
- Do 100k+ tasks in medium size study, for machine learning tests.
- Improving cache hits by showing similar questions
- Using ML to automate abstraction (pointer introduction) for questions
- Tasks with built-in search
- Dialogue Trees
- Users can send messages to specific people involved in the main questions. The messages and answers get treated like pointers.
- Consider switching to an event-sourced database: https://blog.risingstack.com/event-sourcing-with-examples-node-js-at-scale/
- Clean up & test various slate.js /pointer logic
- Other frontend testing (e.g. testing of various React components)
- Backend testing (should probably wait until after the refactor is completed)
- adding real types throughout the codebase (should perhaps wait until after the refactor is completed)
- set up compilation of the backend code (it's currently run via ts-node: https://github.com/TypeStrong/ts-node)
Robustness
For the first year, we focus on having 3-20 quality people, approximated as one person. This means quality control is not much of an issue. In the future, the system could include hundreds of people, so quality control will be a thing to work on.
Comparative Advantages
Very long term. Use machine learning to figure out who is good at what and assign them to those tasks.