61. Data Model Personhood 🔗
December 13, 2020
In which I propose 'data model personhood' — that sufficiently complex data streams become virtual persons who should own themselves, like biological offspring — and wonder whether I myself might be a data-ghost of multiple beings who merely thinks it's one person, a la Thomas Nagel's bat.
🔗
AI personhood would be nice, but is a sideshow. What we need is data model personhood.
Which is kinda the same as instrumentation personhood, since data doesn’t exist until it is measured, and measurements don’t exist without data models derived from instrument models
Which is kinda the same as instrumentation personhood, since data doesn’t exist until it is measured, and measurements don’t exist without data models derived from instrument models
🔗
Example: say a digital home thermometer samples temperature at 100hz via a thermistor. That’s ~80 MB of data a day that rests on the meanings of terms like “Celsius” and “byte”, a sensor curve, etc. Collectively that conceptual model of data is embodied by a long list of numbers.
🔗
It’s useful to think of that column of 86,400,000 numbers plus some meta-information is a virtual person attached to the physical device.
The richer and more varied the data, the more it needs to be digested and summarized in reified models to be legible at all.
The richer and more varied the data, the more it needs to be digested and summarized in reified models to be legible at all.
🔗
Past a point of complexity, the data has to speak for itself via bundled analytics etc. It has to make choices about how to present itself at high level. A stream of data turns into a system’s narrative memory of itself and then data turns into Data.
🔗
This may seem silly for a thermometer producing a one-dimensional stream of numbers whose meaning is based on stable physics models, but just peek into say climate data debates around temperature records and you’ll realize how fraught this is... personhood is a way to model that.
🔗
Or take something obviously more complex like a Dexa scan. It produces a data ghost of your body. I got one a few years ago. The tech hands you a piece of paper with a drawing of a human silhouette and color coded visualization of your fat distribution. Dual-energy X-ray absorptiometry - Wikipedia
🔗
In Culture novels, people routinely save entire states of themselves, including memories, that can be used to resurrect them in cloned bodies or as AIs. Im skeptical of “uploading” as a sentience/consciousness preserving process, but the data-agency take on that is a weaker claim
🔗
The question then is: who owns data-persons created from your data? You the source of the data? The owners of the instrumentation IP? The owners of the copyright on the data model used to cast physical signals like a current valuation into data?
🔗
I think the only solution is: the data owns itself. Just like biological offspring do. Anything else leads to an unholy mess of conflicts of interest, bad faith, etc.
🔗
Like children, data persons need parenting until they get “grow up.” But the principle of fostering ontogenic development into full personhood should be the true north. Ideas like “researchers should open-source data and code used to write peer-reviewed papers” are soft versions.
🔗
This builds on my idea of boundary intelligence from a while back.
1/ I'd like to make up a theory of intelligence based on a 2-element ontology: boundary and interior intelligence
🔗
See also B. F. Skinner’s famous “On having a poem” lecture where he makes a close analogy between birthing a human and a poem. youtube.com/watch?v=PauL2K…
🔗
I’m not a data-rights absolutist, and am willing to trade “my”
data for services, but the farther data is along ontogenic path to personhood, the more wary I am of it being locked up. Not “information wants to be free” but “data shouldn’t be in school/jail without good reason”
data for services, but the farther data is along ontogenic path to personhood, the more wary I am of it being locked up. Not “information wants to be free” but “data shouldn’t be in school/jail without good reason”
Hey @whoop I’m enjoying your product a lot, but how do I get my data downloaded in some useful format like csv or json or via api/zapier/ifttt? There’s no documentation that I can find or plans/announcements.
It’s almost 2021! Weird to see a consumer health data walled garden.
🔗
Galaxy brain version: you’re already a data-person yourself. A particular idiosyncratic theory of personhood populated by data streams from about 100-250 lbs of biomass. That you are you is just one opinion about the nature of that mass. What Is an Individual? Biology Seeks Clues in Information Theory. | Quanta Magazine
🔗
Thomas Nagle, 1974: What is it like to be a bat?
Me, 2020: How fo you know you’re not a lichen? A data-ghost of 3 or more beings that just thinks it’s one person 🤣
Me, 2020: How fo you know you’re not a lichen? A data-ghost of 3 or more beings that just thinks it’s one person 🤣
🔗
See also... this idea is becoming more popular by the decade. How to understand cells, tissues and organisms as agents with agendas | Aeon Essays