I've had enough depressive periods in my life to have a bit of a sense of the pattern. I have chronic stress for some sustained period, whether it's because of something happening at work, in my relationships, or personal projects. After a few months of Not Having Fun but otherwise functioning like normal, something starts to give way internally. To me it seems that if the stress was work-related, it's 'burnout' (slightly more sympathetic?), otherwise it's just garden-variety depression. I started noticing this stress pattern a few years ago, even seen the crash coming and acknowledging that I'm pushing myself too hard, but when you have a lot of long and short term worries to weigh up, it's hard to know how much control you have to stop the train.
I first heard the term 'allostasis' in two contexts around the same time:
- 1.
The devastating anonymous essay 'Hot Allostatic Load'
- 2.
A fairly obscure cognitive science research presentation I stumbled across called 'Allostasis Machines as Continuous Cognitive Modelling' by Bradly Alicea
Wikipedia first gives us some general context:
Allostasis is a physiological mechanism of regulation in which an organism anticipates and adjusts its energy use according to environmental demands.
The first essay quotes from somewhere, considering in particular the chronic stress caused by (C)PTSD symptoms:
“The allostatic load is ‘the wear and tear on the body’ which grows over time when the individual is exposed to repeated or chronic stress.”
Anyone familiar with PTSD will know intimately how non-traumatic events can be psychologically experienced as 'triggers', producing anticipatory responses in the body and mind that would be much more 'normal' and proportionate to the historically traumatising event than they are to the current triggering event.
Alicea's presentation is way more abstract and technical. Allostasis is comparable to homeostasis - both are models for systems involving feedback loops, and thus in the cybernetics wheelhouse. But an allostasis machine has a primitive notion of 'memory' - the regulated state it wants to return to. Alicea defines it as:
a system where the output characterises a cognitive architecture under allostatic regulation, subject to environmental perturbations. [...] Trajectory either recovers from the [environmental] perturbation or towards a new stable state. Failure to do so is an ill-matched response, and produces instabilities.
I have only an untrained enthusiasm for cognitive science, but this concept has been like a fun chew toy for me for a while, and I can imagine components of a brain neuron working a little bit like this in the process of learning. I've been reading Donella Meadow's Thinking in Systems in preparation for my ATmosphere Conf presentation (see you Sunday morning?) and one thing I've been especially thinking about is how to model system phase transitions. Meadows says feedback dynamics can cause systems to be resilient to disruption and resist phase transitions, in proportion to the depth of the lever being pushed. It's interesting how in the allostatic model, the perturbations can be cumulative and cause a transition to a new regulated state within some time period.
System phase transitions feel relevant to our broader questions in the decentralised web about the transition from a monopoly-based societal communications infrastructure to one that is plural and open. It also feels relevant at the smaller scale of communities using one or another platform to communicate, which is closer to what we're concerned with working on Roomy.
It also resonates for me with the phase transition between 'functioning stress' to 'burnout'. Stress is typically modelled as an environmental condition, so stress is causing the repeated perturbations that my well-functioning system has for the last few months been effortfully, repeatedly recovering from to keep the system stable, until, just recently I noticed it's subtly entered a new phase.
Urgency and pressure is not active in our collective work culture on Roomy - we care about the work and about each other as well, we value emotional aspects of the work and reject toxic productivity culture. At the same time, we do have some big unknowns about the future for Roomy which have been on my mind a lot, and getting it ready for the conference as well as stuff in my personal life has all been a lot. But just acknowledging and processing this shift to low level burnout has already helped. Like I said, it's not my first rodeo, so I'm not worried - it's just a bummer. I think having some kind of model of what's going on internally helps me avoid downstream things like shame making it worse.
The other thing about allostasis machines I found interesting has been a bit of a reframing around stress. I was tempted to say above that users of Instagram, for example, are subject to some low level 'stress' every time they encounter an advertisement. This might be true, but looking at allostasis machines as more abstractly responding to environmental perturbations makes me think how 'environmental stress' can be something we adapt to and can routinely recover from without causing system instability. When I'm feeling good and starting work, I feel a level of 'stress' that I actually like, just as coffee-induced cortisol starts spiking. If I'm lifting weights, my muscles are stressed, and then they recover and get stronger. Bounded, predictable routine stress and routine recovery feels like a healthy life.
I don't like ads or think they are healthy, but I can adapt to them, and this calibration for stability is the level Meta plays at. In this sense it makes sense how migration to non-monopoly social media platforms seems to often happen in response to some big disruptive event crossing a lot of peoples' thresholds at once. It would be interesting to study these transition events at a more local community level. Shifts can be painful and scary in the unstable period - you can imagine the unfortunate prolonging of this instability as users trying to escape Big Social scramble to figure out which Mastodon homeserver to sign up for. Given how 'high-leverage' (in Meadows' terms) the change of platform is personally for a user, modelling the negative feedback dynamics resisting that shift helps us see how critical those onboarding moments are for the entire system changing.
For me, maybe having ways to model what's happening can help me quickly arrive at a new stable state - including shifting what I'm expecting of myself, how much rest I need, and how hard I choose to push myself.