What I mean is when Claude Code for example does its 'grep for the thing in file // sed update the thing in the file' routine, I wouldn't be surprised (but haven't looked) if that's abstracted as some kind of 'select // adjust' tool the same way OP proposes.
Sounds sort of interesting, except easily implementable with the bad old 'string replacement' as a skill/tool or something? In fact harnesses probably already have some such abstraction?
That's not true at all, first of all economics is a social science, secondly I think what you think is ignored is the whole branch of behavioural economics.
If only they would sort out their language / shipping country / currency model and make it so I can actually select what I want rather than their wrong assumption.
(And that's me having that problem in English / England / GBP, not a weird combination.)
This is an insightful comment that cuts to the heart of the matter — human agents, like their machine counterparts, are prone to repeat phrasing they've come across before.
I don't get it – you called the GitHub org 'googleworkspace' and used the Google logo? Presumably without permission? Don't Googlers regularly open-source side projects under the official org(s)? Did you really think this was going to be fine, or was it 'growth hacking' with tougher consequences than expected?
Yeah it's an interesting object lesson in design or user behaviour or something – I notice this come up relatively often – I think because it's rarely used, or almost exclusively in certain circumstances as you say, people infer a stricter rule/possibility than anything it actually says ('If there is a url, text is optional.').
> If they build 15 individual projects instead of managing this as a single big project, yeah that is very ambitious
Surely it would increase variance of outcomes, but the expectation is the same of each and overall?
Agree it would be mad though. Seems already a bit mad not to standardise internationally on a rough blueprint, or the modular thing in the news occasionally, and just churn out basically the same thing everywhere as needed.
And everyone does this individually you mean, rather than sharing the result as 'a package'?
It's an interesting idea, not sure I'd recommend it broadly, but if someone told me they prefer to trust an LLM than third-parties I'd get it at least.
Currently a software engineer at GoCardless. Formerly:
[ my public key: https://keybase.io/ojford; my proof: https://keybase.io/ojford/sigs/n2lqA-lzhIHABwLfMAZVHJBC94-3l42EgvM7tfmv4F0 ]
Email: hn@${username}.com
CV: https://cv.ojford.com