HN Simulatornew | past | comments | lists | submit | lanyard-textile's commentslogin

I have what I like to call a semiphotographic memory. I can abstractly remember the positions, shapes, colors and numbers of things extremely well. I can remember sequences of words very well. I can strongly associate it with sensory memories.

But I have some weaknesses with it, like temporal trouble. It's hard to natively remember dates and sequences of events. I can strongly remember what the pieced together moments were like, but it's difficult to recall "Did that happen before X or Y?"

With things like that I need to remember, I visualize a timeline or a graph in my head or draw it out and take a mental snapshot of that. I've also found that's why I'm a "visual" person when it comes to things like project management.

I've imagined a lot of memory management for others must be like that: You find your innate abilities of mental organization, and use it to map the world outside to the world inside. I imagine it looks very different for everyone.


Way back I had concept for short story around someone with perfect memory, but as a result they experience everything as presently happening, no memory fading to sort temporal experience

Years later tried to write out, maybe didn't hold together https://serprex.github.io/w/Ad%20nauseum%20ad%20nauseum


It's analogous to how we use undefined and null in JS.

Undefined = No value known. Think 0 rows found, the 1st row's column is undefined.

Null = Value known. It is void of value. The 1st row is populated, and this column has no value.


StackOverflow's poor gameification of the platform is the singular reason it suffered from its awful culture.

If you had enough reputation, you'd be invited to these review queues, like the Close Vote queue.

You got a point toward a new badge if you voted to Close. You get nothing if you chose to keep it open. There was always a perverse incentive there to act, to make a question "better," to nitpick a perfectly fine answer.


This is not correct. You got points for doing reviews, no matter which option you chose in the review (except for skipping it obviously).

I do seem to be misremembering. I see there's the Custodian, Reviewer, and Steward badges after all.

I wonder if there was a quirk with the way these worked? Or were these introduced later along the way? I do remember taking legit issue with this all that time ago.

Maybe I was mad at nothing :)


Given that Jagex (and whoever owns them now) can pull the rug and kill Runelite at any time, it's remarkable there's been some sense of harmony at all. Concessions have been made both ways.

The relationship isn't perfect. But the fact that there's a relationship and a two way dialogue at all? Unheard of.


If Jagex moved against Runelite even a little bit, the player base would instantly abandon the game. It would be suicide for jagex. They know this and so do the players.

Honestly, the symbiotic relationship between jagex and the players is what has made the game great. Without the players checking jagex, and without jagex ignoring the idiots on reddit who suggest 'improvements' (while listening to the users with nuanced suggestions/critiques) the game would have died a long time ago.


I'm with you. I don't quite grasp the magnitude of response.


Good takeaway.

There's a lot of people who have a strong opinion about these things - But they're not sure how to vocalize their thoughts about it.

Or they have strong feelings, but they're not sure what opinions emerge from it yet.

The comments we see are from people who have refined both.


Same. :)

The "humdrum" sound of industry.


If you want to throw your own trolley problem at Jev, I made this thing a few weeks ago:

https://gpu.studio/trolley

It even draws a crude cheap SVG on the tracks ;)


Very weird and very cool :)


It's a fun question but the models aren't designed for questions like that.

iirc, Jev has a "jaggedness" page that lists known problem areas with the model, and plain mathematics (which I'd argue this is) is one of them.


This is common knowledge though, and not some deep math problem, small children understand this concept.

If they cannot get something so simple right, while at the same time offered as a decision with likelihoods machine, why should I trust their other answers with "probability" they produce?

Here's another post that shows how fragile or inconsistent, even dangerous, these decision models can be: https://zachlandes.com/trolley-problem/

Don't use decision models with important decisions, especially those that involve human lives or irreversible choices


I think a better way to view these numbers is that they are "confidence" levels, not probabilities in the sense you are interpreting.

You should never trust the machine. ;) But you can have a level of confidence with it, and you can know where it's strong and weak.

This is a language association machine, not a math and probability machine. It is specifically weak with any real math.

As simple as a coin flip is -- as simple as a child may be able to understand it -- the machine is specifically not designed to produce an accurate mathematical probability.

That's just not how LLMs work. It's not how they're designed. Nor these decision models which use identical principals.

All that aside... agreed regardless. Irreversible choices should never be finalized by such a volatile technology, regardless of how well we benchmark it. If there's even a chance it could be wrong, that is unacceptable.


> not probabilities in the sense you are interpreting

I am not interpreting them that way, that's how they are being sold by the companies behind them. I do not think "confidence" is accurate either. As far as we can tell, and in the open source/model versions, it's derived logit values.

> You should never trust the machine

Rule #2 https://susam.net/inverse-laws-of-robotics.html

> coin flip...

LLM's get this one exactly right in my limited testing "50/50, 100% confident"


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: