Nested false awakenings are the worst. Now I have a phrase to use for when that happened to me.
I dreamt that I was attacked and killed by a monster. Getting killed in a dream makes one wake up immediately, lol. I was hanging out with a friend of mine and telling him about the dream and as soon as I told him what happened he turned into a monster and killed me.
I was driving down the road with my friend Lloyd who was a passenger beside me in the car and telling him about this crazy dream. As soon as I mentioned the monster killing me, he turned into a monster and killed me - while I was driving my stinking car! Then I woke up and I was like "am I fucking really awake now?" I was.
ugh, I would get those as a kid. I've never heard the term 'nested false awakening' applied to them, but I like that. I'm glad I don't get them anymore.. the feeling of wondering if it is a dream or not and feeling like you are locked onto a roller coaster you can't get off of. Luckily in my experience once I actually woke up I would know for sure that it was real. I think it would mess with your psyche if you couldn't tell or convince yourself you weren't in another nested dream.
I thought this was universal. It's pretty easy to trigger for me, even quite abstract geometric shapes that look a bit like something falling from a height can trigger it.
The logical answer is to send what you would have put into AI.
However, terse lists of bullets without any padding or phatic expressions are considered rude. Perhaps previously a long message indicated you'd put time and thought into it.
Maybe AI will invert the social signalling, and make conciseness polite. As it should be.
That is another way to do. Or we can automatically figure out which models the subagents should be using for you. And update them as new models come out and the work your subagents do changes. More than one way to skin a cat.
This does make sense. I generally only switch between models in pi when creating a new session. And it is apparent from the promt if this just a "how to see open ports on linux" or "make a concrete plan for feature X"
What if the drugs just make us... better? As in happier, less stressed, and more suited to what modern life requires from us.
Happiness isn't necessary to pass on your genes. Most animals, especially prey animals, spend much of their lives scared, because death is inevitable otherwise. We've now created a safe and pleasant environment our ancestors could only dream of, yet many of us are still unhappy. There's no real evidence we were broadly happier in the distant past either, we're just a different kind of unhappy now.
Seems to me we've developed a tool that makes people feel better and function better, perhaps by blunting some of the vestigial leftover emotions we needed when every tree might have had a sabre-tooth tiger hiding behind it.
"more suited to what modern life requires from us." This part scares me because I would prefer that we rearrange our economic and social world so that people can have more free time, more vacation, more relaxation, and way more social time too.
I have known many people in my personal life in the united states who have an extremely stressful and depressing life due to external stressors (job, debt) and in response are on powerful psichiatric meds with the sometimes-severe withdrawal problems mentioned in this thread.
It shouldn't be an excuse for not making people's lives better long term, but you also have to treat people where they are in life.
In some ways, certain problems seem to emerge from modern lives being too comfortable for the environment our brains are adapted for, but it's not as if unhappiness is a new thing, we just have better tools to treat it now. My life is about as good as any life can reasonably be expected to be economically, and I have as much social interaction as I want... but I'm still anxious (without sertraline).
It really seems like humans have been fairly unhappy throughout history, are fairly unhappy now in the best societies, and are fairly unhappy now in the worst societies too. We're a pretty unhappy species prone to quite a lot of mental health problems. Only difference being that in the past, eventually hunger would override your unhappiness and you'd have to get up and do something about it, which would take your mind off worrying about dying from that worsening foot infection and mourning over the twelve kids you've had die in their childhood.
I would also suggest that being overly negative around psychiatric meds itself can be harmful. I know many people who lived deeply unhappy lives and avoided taking them for years over fears of side effects and things they'd read about, only for them to eventually give in and wonder why they hadn't done it years ago. They do have side effects, but man, as a N=1, all I can say is sertraline improved every single aspect of my life. It's indescribable how much I owe to that compound.
I could function. I would be a lot more anxious. I'd probably function better in certain scenarios, just ones that I'm unlikely to encounter.
About 80% of people are prescribed psychiatric medications at some point in their life. Add in how many people use non-prescribed ones (smoking, alcohol, illicit drugs) for psychological support and you're going to have a hard time finding a control.
Maybe others have found otherwise, but I find the benchmarks drastically different to real world "feel" of a model, even within the same harness. I'm not sure if this just reflects personal interaction styles, or if it is indicative of benchmaxxing or unrealistic automated benchmarking methodology.
Opus 5 consistently comes at or near the top, but outputs constant unreadable jibberish. Meanwhile GPT 5.6 Luna medium tends to be rated pretty poor on agentic tasks compared to the Chinese lab open models, but I find the latter much more likely to lose track of their own behaviour during a long-horizon task or get stuck in a doom loop.
(This isn't a comment on GLM-5.3 Flash as I've not used it!)
We shouldn't have to resort to this, but it can be mitigated enough that it stays as my daily worker agent. Although, I mostly use Fable to farm out to Opus agents so I don't have as much exposure to what kind of blathering is going on in there.
It's crazy that sometimes I ask Opus 5 to explain what it just wrote to me, and it declares "that was word salad" (its words, not mine, without any hint from me other than "explain it").
For me it's been this: Opus (at least in my experience) is unbeatable at "planning the work". That includes a lot of things, including getting arch. sorted, a chassis/skeleton done. Filling that up and doing the actual "coding," though, I've noticed no real difference between the Claude model and GLM. So it'll be interesting to see how this flash model compares cost-wise to what I'm currently using, which is GLM 5.3, for coding. Looks like it will be reduced even further and might be great if it's faster (and better?) than 5.3 in my real world/personal experience.
(I'm someone who doesn't really care about delays of a few seconds, or even more than few seconds. But if I am trying to notice then sure Claude is definitely faster as well).
I've been using 5.3 since they initially announced it and my gut feel is that it's not as good as 5.2 for agentic tasks. I'm still using it - I don't think it's bad. I'm just not convinced it's better.
I have had the opposite where I felt 5.3 as a stronger model than 5.2. Its feels way more in tune with my code, and does more nuanced edits. Though I do handhold my models a lot, so might fall outside the agentic term.
Yeah, I just ignore the benchmarks at this point. For open-weight models the provider's setup impacts performance so you can have different experience's with the same model at the same quantization from different provider's. Just have to use them on real tasks with your actual harness to really know how they will perform and hope the provider doesn't do something to degrade performance (e.g. update the middleware to a new version with a defect that impairs performance).
reply