I don’t think approaching boiling temps is quite enough unless it’s kept there for a long time (see pasteurisation times at various temps). I would agree with the author that if the contamination levels are high I wouldn’t risk it.
boiling is more than enough to kill bacteria in general (Ex: if one had the ability to bring chicken instantly up to a temperature of 160F or so, it be considered perfectly safe. In the sous vide world one might only want to bring it up to 140-150F to not overcook it, and therefore would hold it at that lower temperature for some longer time).
With that said, this is only for bacteria. Bacteria also can produce spores (Ex: the botulism toxin produced by a bacteria, which is produced in an anaerobic environment, and exactly why sous vide cooking at low temperatures can be dangerous, as the anaerobic environment that it can thrive in and if kept at temperaturese in the danger zone where it can thrive can result in the production of botulism that won't be later killed/destroyed by higher temps that kill its producing bacteria)
I feel that agent coding is actually giving a second wind of life to solid principles, “proper” software architecture. Now you can nag the llm to follow them and A- it will actually apply them if well directed and does not mind the (small?) extra complexity upfront B- you pretty much immediately see the effects
Thanks for reading! I find myself thinking along these lines pretty often, old school, "proper" software engineering (referring to solid, architecture design before coding, TDD ... ) that in my experience would get you sideways looks in some companies with the argument of practicality, are now feeling suddenly very practical.
reply