This is a document for ethical practices in creative bot making. It is written as a companion resource to go along with a planned set of Coding Train video tutorials related to making Twitter Bots. At the moment it is a list of guiding principles and questions to ask as part of the practice of creative bot making.
-
Develop a practice of filtering.
- A simple solution is to use a word filtering module or database, for example wordfilter by @tinysubversions
- More sophisticated methodologies should be considered, for example, Transphobic joke detection also by @tinysubversions.
-
Don't say anything you personally wouldn't say to a stranger!
My bot is not me, and should not be read as me. But it's something that I'm responsible for. It's sort of like a child in that way—you don't want to see your child misbehave. —@xor
-
Limit interactions!
- One option is to block your bot from any replies or @ mentions of any kind.
- Another option is to restrict your bot to replying to or @ mentioning only people who follow the bot ("opt-in")
-
Are you able to actively moderate the bot and remove postings if necessary? Would manual curation of generated content be most apporpriate for the context of your bot?
- What is the full range of possible outputs?
- Before you deploy, run your bot for a period of time logging the output without publishing and evaluate.
- How might others misuse the bot?
- Whose voice is the bot speaking with?
- Are you able to speak with that voice from an ethical and legal standpoint?
- Can you trace the authorship of what the bot says?
- Are you able to properly credit the source data and text and the associated labor with its creation?
- reference thread from @aparrish.
- How to Make a Bot That Isn't Racist by @sarahjeong
- Bots should punch up by @leonardr
- Queer your bots: Bot Builder Roundtable by @carolinesinders
- "whatever your method of preventing your bot from committing violence, it must be relational, interpersonal, tactical and contextual as well from @aparrish