3
points
StevenWaterman
Joined 1,349 karma
https://github.com/stevenwaterman
HackerNews@StevenWaterman.uk
- StevenWaterman parentThat is true for gpt-image-1 but not nano-banana. They can do masked image changes
- Yep. Insufficiently stimulated by normal life, a crisis brings your dopamine levels back up to normal and you hyperfocus. Get tested and medicated, for you and your family
- If you want to run it overnight, or while you're at work, so it finishes as you arrive and doesn't leave the clean clothes in a clump for hours (or so it runs during cheaper power hours)
- and your wireless modem has wires
- I'm currently working on adding a bot to our support chat at TalkJS. And it's great, it has probably a 90% success rate at handling complex queries. But that's because we're throwing money at it. That chat is normally staffed by senior devs, meaning it's not unusual for a single response to cost $10 of labour.
If you approach it as a cost cutting exercise, you end up with crap. If you approach it as a way to make a better experience while you sleep, it's achievable.
- I'm glad I'm not the only one thinking that. These are such minutiae. Where's the discussion about humans? They're probably the most important part of your system, and the most chaotic, and the part that needs the most careful design.
It's hinted at a little bit in the OP, with:
> What does good system design look like? I’ve written before that it looks underwhelming
This is because there are humans in your system! Other developers! You in the future! You have to resort to heuristics like "simple == good" because you're only looking at a small part of the whole system.
And zoom out even more, you get to the actual users. How do they interact with the system? If you implement a rate limiter, how do the users respond when they hit it? Do they just spam-refresh the page? Open more tabs? Use their phone? Do they develop weird superstitions about it? Do they spam-call your phone support lines? Does your response to a thundering herd anticipate the second-order impact of your phone support lines being DDOSed?
- Essentially the idea is that there is an "optimal" amount of alertness (inverted U curve). People with ADHD start below the optimal point, and stimulants move them up towards the optimal point. People without ADHD are typically closer to the optimal point, and stimulants move them past it.
Someone with ADHD taking a large dose will therefore feel the same as someone without ADHD taking a small(er) dose.
Methylphenidate improves sleep in people with ADHD: https://pmc.ncbi.nlm.nih.gov/articles/PMC2276739/
> Compared to [non-adhd] controls untreated [adhd] patients showed increased nocturnal activity, reduced sleep efficiency, more nocturnal awakenings and reduced percentage of REM sleep. Treatment [of those with adhd] with methylphenidate resulted in increased sleep efficiency as well as a subjective feeling of improved restorative value of sleep.
I can't find a corresponding paper studying the effect of stimulants on sleep in healthy adults. I would assume it hasn't been studied because it's common knowledge and it's not worth the risk of making healthy people take stimulants. I also don't think that's the part you were disputing.
- We already have a phrase for "if you start a fight you get punished"
The phrase is "normal, non-zero-tolerance policy"
- https://en.wikipedia.org/wiki/Zero_tolerance
> A zero-tolerance policy is one which imposes a punishment for every infraction of a stated rule. Zero-tolerance policies forbid people in positions of authority from exercising discretion or changing punishments to fit the circumstances subjectively; they are required to impose a predetermined punishment regardless of individual culpability, extenuating circumstances, or history.
If you use "zero tolerance" to mean "zero tolerance for starting a fight" you need to make that very clear, because that's not how it's used in schools currently.
- You see this in ADHD groups where someone will start stimulant medication for the first time and say "This is incredible. I can't believe this is what everyone else feels like all the time"
And the crowd emerges to reinforce that, no, you're euphoric, this isn't normal, after about a week it'll go away and you'll just feel normal but more productive and have better executive function.
And that's on a starter dose, the parent commenter probably took 2-3x that
- > It would be weird if stimulants only had an effect of ADHD patients
One example of this actually happening is the concept of a "stimulant nap" in people with ADHD, where stimulants actually make them sleepier. Also manifesting as "I tried coke once, it didn't do anything, I just felt sleepy"
Terrible source but it's a pretty common thing: https://www.reddit.com/r/ADHD/comments/hkkyjl/you_know_your_...
- What do you call system design, when it's referring to the design of systems in general, and not just computer services?
As in:
- writing a constitution
- designing API for good DX
- improving corporate culture
I intuitively want to call all of those system design, because they're all systems in the literal sense. But it seems like everyone else uses "system design" to mean distributed computer service design.
Any ideas what word or phrase I could use to mean "applying systems thinking to systems that include humans"
- And likewise, a single neuron is clearly not conscious.
I'm increasingly convinced that intelligence (and maybe some form of consciousness?) is an emergent property of sufficiently-large systems. But that's a can of worms. Is an ant colony (as a system) conscious? Does the colony as a whole deserve more rights than the individual ants?
- Yeah, I think this is exactly it. If something sounds extremely stupid, or if everyone around you says it's extremely stupid, it probably is. If you can't justify it, it's probably because you have failed to find the reason it's stupid, not because it's actually genius.
And the crazy thing is, none of that is fundamentally opposed to rationalism. You can be a rationalist who ascribes value to gut instinct and societal norms. Those are the product of millions of years of pre-training.
I have spent a fair bit of time thinking about the meaning of life. And my conclusions have been pretty crazy. But they sound insane, so until I figure out why they sound insane, I'm not acting on those conclusions. And I'm definitely not surrounding myself with people who take those conclusions seriously.
- That paper was that thinking more makes models worse at easy problems, not hard ones
- I might have to adopt this way of differeniating OR and XOR in English https://ygdp.yale.edu/phenomena/alternative-one
- I definitely agree with the difference in formality.
> You should try and recalibrate your photon detector
I can totally imagine this, in a lab where all the equipment is old, and out of calibration, and the person saying it knows there are 10 other things that are more important, but this thing is still pretty bad and they feel obligated to point out the issue.
Whereas "try to calibrate" sounds to me like the process of calibration is quite hard and it's likely to end up no better calibrated than you started with.
- I'd describe it as:
- "try and" implies that the reason for failure is slightly more likely to be from laziness / not actually attempting it
- "try to" implies that the reason for failure is slightly more likely to be from incapability
As in:
- I'll try and kill the mosquito... that has been annoying me all day
- I'll try to kill the mosquito... but it's quite hard to hit with this gun
But nobody would notice if you used the wrong one.
- I disagree that increased employment and increased labour always makes the pie bigger. If minimum wage was low enough, we would decommission our cement mixers and use a human with a shovel instead. But that's not an improvement. Automation is happening, jobs can be replaced right now. The problem is that humans are too cheap to bother automating, and that the profits of the automation are not being distributed to the displaced workers.
- The total salary would go down if you did that
- If the total salary has gone up, for less work done, it is a positive change. You can solve the inequal distribution via taxes and benefits.
Start: 100 people paid $100
After minimum wage change: 90 people paid $125, 10 people paid $0
After tax increase: 90 people paid $113 + $12 taxes, 10 people paid $108 from taxes
Now everyone is paid at least as much as they were before, and fewer people are forced to perform labour
In practice it was only 3% unemployment not 10%, which means the tax increase is less and there is more of an incentive to continue working. You can also pay the displaced workers less than their original wage, to reach an equilibrium where everyone is happy with either work+more money, or leisure+less money. Or have it be age-based with an earlier retirement. Or have people work part-time.
We need to stop seeing having a job as being inherently good. Being able to live is good. Humanity should strive for 100% unemployment.
- Pineapple has a +22 approval rating https://d3nkl3psvxxpe9.cloudfront.net/documents/Pizza_poll_r...
- The AI's idea of developing a startup is eerily reminiscent of a hacking scene in CSI
- > Norwegians are taking much more sick leave than a decade ago, driving up costs for health services. Student test scores have worsened more than in other Scandinavian countries
...COVID?
Norwegians are taking more sick leave after the largest mass-disabling event in generations?
And test scores are worse compared to other Scandinavian countries, like Sweden, which famously had very few COVID restrictions?
Must be the oil money
- Your question reminds me of the image showing how SpaceX raptor motors evolved https://imgur.com/a/4w3q3lS
- Yep, every AI call is essentially just asking it to predict what the next word is after:
And we keep repeating that until the next word is `</assistant>`, then extract the bit in between the last assistant tags, and return it. The AI has been trained to look at `<user>` differently to `<system>`, but they're not physically different.<system> You are a helpful assistant. </system> <user> Why is the sky blue? </user> <assistant> Because of Rayleigh scattering. The blue light refracts more. </assistant> <user> Why is it red at sunset then? </user> <assistant>It's all prompt, it can all be engineered. Hell, you can even get a long way by pre-filling the start of the Assistant response. Usually works better than a system message. That's prompt engineering too.
- "Weights available" perhaps
- Are you getting confused between tariffs and VAT? VAT is the equivalent of sales tax.
From what I can tell, the tariff on "Potatoes, Thin slices, fried or baked, whether or not salted or flavoured, in airtight packings, suitable for immediate consumption" from Ireland is 14%, reduced to 0% if they originate in Ireland https://www.trade-tariff.service.gov.uk/commodities/20052020...