Reviewing code can be hard but it's not as hard as writing the code. Even with the best autocomplete, and ergonomic editors like vim, it still takes quite a bit of time to write code for some features compared to the actual concepts being implemented. There are also lots of decisions like variable names that can be automated with a LLM. If you don't like what it came up with, you can tell it to change them. I recommend that you keep them fairly unique like you would for your own handwritten code, because ambiguity creates problems for people and machines alike.
as a comparison, the gemini cli agent with gemini 2 half the time writes its own tool call parameters incorrectly. it didnt quite know when to make a tool call, which tool result was the most recent(it always assumed the first one was the one to use, rather than the last one, when multiple reads of the same file were in context) etc.
gemini 3 has pretty clearly been trained for this workflow of text output, since it can actually get the right calls in the first shot most of the time, and pays attention to the end of the context and not just the start.
gemini 3 is sitting within a format of text that it has been trained to be in, where for gemini 2, it only had the prompt to tell it how to work within the tool
i think these already have you screwed. that anonymity is going to be superficial at best. you will be recorded making these purchases, and tracked to your identity
Then you're going to take it home for >8 hours per day, and to your job several hours per day, and likely call at least one or two of your important contacts. At which point that's the ball game - the pool of people that live in the immediate vicinity of your building, and work in the immediate vicinity of your job site, and call your partner / parents / kid, is made up of pretty well exclusively you
we just saw the biden admin not do that, and the polarization only grew. They very specifically slow walked their investigation into trump's treason, so they wouldnt have to Nuremberg him.
trump mind you, is nuremburging non-voters. they dont exactly have side beyond trying to work and eat
its not some pattern of abuse by shady actors manipulating opinions youre noticing, its voting algorithm and attention economy itself.
new ideas are constantly being published, and popular ones gain momentum by being shown to more people. as the idea gets saturated, the popularity gets overshadowed by the time based downranking.
if the idea is still popular though, in this case that ice murdered some woman as part of their shock and awe campaign, variations are going to show up such as "legal observer" and "mother of a three year old"
id just bring my phone. if the secret police are going to arrest or kill me, theyre going to do it either way, probably while im there.
if i bring a phone, i can at least document the secret police's actions, and make friends and get contact info for other people that are there.
security by obfuscation isnt particularly good, and its a state level threat.
there's a different absurdity with your child torture example - that youd be ok with children being tortured over phone usage. the bulletin and the people doing the kidnapping at torture are the problem, not the phone. there's a third option of stopping said torturers, and youll likely want your phone as one of the tools in doing so
what i really want to see from this article is a curve showing a tradeoff between speed and embedded text quality. there's the preamble that just going by character has quality problems, but i dont think delimiters are necessarily the best either, vs being able to find paragraph or even chapter boundaries.
how much of a problem is it that ~1 sentence per chunk gets corrupted in the by-character solution? what level of sentence corruption is left in by switching to these delimiters? what level of paragraph/idea corruption is left in with each? chapter/argument level?
reply