Thread by @emilymbender

This is a very thoughtful reflection by @zephoria --- and it's striking to be offered so much inside info about the CTL/Loris debacle --- but it also doesn't fully connect the dots. A few thoughts/questions:

zephoria.org/thoughts/archi…

@zephoria boyd steps the reader through how the organization went from handling data for continuity/quality of service to texters (allowing someone to come back to a conversation, handoffs between counselors, connection to local services) to using data for training counselors to >>

@zephoria using data for internal research for using data for vetted research by external partners as one thread. That last step feels rickety, but still motivated by the organization's mission and the fact that the org didn't have enough internal resources to all the beneficial research.

@zephoria Another thread seems to be: data for training counselors to counselors benefit from that training in their lives more broadly to can we offer this training to others (for a fee) to make money to sustain the core mission? >>

@zephoria One thing that's not clear in the essay is why any data would need to be shared outside the org for that purpose, even if the training is done by a separate, for-profit entity. How is the data used in that (external) training?

@zephoria But more importantly, the essay doesn't get to the question of how Loris came to be licensed to use the data to train "grammarly for emotion" in customer service contexts:

https://twitter.com/_jack_poulson/status/1487636982507397120?s=20&t=xYYLnns88441gzNDZBqYGg

@zephoria I read this essay with curiosity to see how the inside view would complicate that moment and to see if I could spot where it felt like things went off the rails (with full benefit of hindsight & boyd's own reflection). The former wasn't provided (essay didn't get there), but:

@zephoria Reading along, this is where it felt, to me, like things went off the rails:

This seems to be predicated on a notion of scale (and implicitly, on the idea that masses of data can let us do things on a scale that individual people working with data individually can't) that echoes all of the things we hear out of Silicon Valley (disruption etc) >>

and in opposition to boyd's own mantra about "responsible scaling". There's a jump in there from "how do we make this thing we're doing more sustainable and accessible to more people?" to "how do we (use this data to) change society?"

Would I be able to tell if I were making that same kind of jump in real time? I have no idea. I think it should have been obvious to anyone who heard "grammarly for emotion" that something had gone terribly wrong and needed fixing. But boyd's essay doesn't cover that part.

It is very valuable to have @zephoria 's insider perspective here, and I think there is a lot to learn from this. My initial primary takeaway is that this underscores the risks of data aggregation:

Aggregated data is more attractive to business interests and is (maybe?) harder for us as individuals to understand on a visceral level as requiring the same fierce protection as its individual components do.

Maybe this is analogous to the saying that one death is a tragedy but a million deaths are a statistic? Anyway, there is definitely a lot to learn here about how to cultures of care around data that are robust to the siren call of "scale", in all its guises.

How to build* cultures of care. (sigh)