I have thoughts. Lots of thoughts. They never stop thinking. Never stop thunking.
hachyderm.io/@hazelweakly
Hazel Weakly
Loading...
We made an open repository of close relationships measures. Our goal is to catalogue all of the published self-report measures in relationship science.
You can help us expand the database by suggesting measures you tend to use in your research, or by uploading your own measures!
relascale.com
Thinking about this, it feels like a social dilemma: if we, as an industry, collectively cared and built tools to make software understandable and intelligible to different industries, things could be very different
But we ALL have to buy in, including vendors, industries, and governments
Ultimately it ends up being more profitable to not engage in solving the social dilemma, to push for deregulation on the things you don’t automate, and to use “process jank” as a weapon to slow down others in places where you’ve automated
It’s technically a valid “competitive advantage”, after all
And I know that because If I had automated before, I’d have to entirely rewrite it to take advantage of new tooling available today. That’s painful!
Especially because the payoff is nearly zero until the very end, and it’s not a known guarantee that you’ll even get there
But yeah, automation is hard, and you end up with decades of layered assumptions about careful tradeoffs and bottlenecks. When something upsets that balance or tweaks the constant factors of the equation, large upsets happen
It’s easy to say that automation pays off, but historically… it’s tricky
Given a choice, leaders will usually pick “high ROI” initiatives. It makes sense, and it’s silly not too. But unfortunately that also often means “highly visible” and “politically favoured”. Also not necessarily wrong, but it means things don’t get better unless they’re “transformed”
I’m thinking a lot right now about automation and efficiency. We spend a lot of calories talking about AI, but I don’t see that same energy being put towards truly impactful changes: mostly because they’re largely invisible
More importantly, those transformations have to be won bit by bit
If I sum the cost of full automation, I’m looking at a decade+ before hitting even. So I’d need a cross functional platform, and those are fragile to regulation/security updates, workflow changes, and internal politics
Plus it’s years of “2% better” before unlocking a magic threshold of automation
Most of these impactful changes are things that make zero sense, mathematically, which makes the problem worse
Take security architecture reviews, for example: dozens of hours per application, most of which is spent on copy and paste and data lookup. Then months of waiting in queues for review
Even worse, remaining automation relies heavily on layers of fragile tooling, involved explanations to auditors, and bleeding edge AI + formal verification techniques that are still years away from standardisation
If I automate now, I risk having to rewrite it all before it ever paid off
RelaScale is a searchable database of psychological measures and constructs for relationship researchers.
