Universal Paperclips • Hacker News Board

Re: Universal Paperclips

#31

Post by l0b0 » Thu, Nov 03, 2022, 8:50 AM UTC

NKosmatos wrote: post #15

A warning to fellow HNers that are not familiar with universal paperclips. Please don’t click the link as it will create a black time hole that will start absorbing your time and reduce your productivity. Be warned that this isn’t a naive lazy clicker game, it’s an instrument of destruction from the future, created by our paperclip overlords :-)

I've played through UP twice. When it comes to time sinks, it's nowhere near the ultimate death spiral that is Kittens Game. To date it has taken me about 5 times as long as a UP playthrough (and 500+ lines of JS automation which I seem to have to tweak every day to deal with the changing economy) and I don't seem to be anywhere near the end.

Re: Universal Paperclips

#32

Post by arc-in-space » Thu, Nov 03, 2022, 8:50 AM UTC

Oh, god, not again, I was intending to do work today.

Re: Universal Paperclips

#33

Post by yurishimo » Thu, Nov 03, 2022, 8:51 AM UTC

l0b0 wrote: post #31

NKosmatos wrote: post #15

A warning to fellow HNers that are not familiar with universal paperclips. Please don’t click the link as it will create a black time hole that will start absorbing your time and reduce your productivity. Be warned that this isn’t a naive lazy clicker game, it’s an instrument of destruction from the future, created by our paperclip overlords :-)

I've played through UP twice. When it comes to time sinks, it's nowhere near the ultimate death spiral that is Kittens Game. To date it has taken me about 5 times as long as a UP playthrough (and 500+ lines of JS automation which I seem to have to tweak every day to deal with the changing economy) and I don't seem to be anywhere near the end.

Have you ever played Trimps? Started a week ago and I'm hooked... I wish it had better mobile support though.

Re: Universal Paperclips

#34

Post by pietroppeter » Thu, Nov 03, 2022, 9:01 AM UTC

a recent podcast on the game: https://caneandrinse.com/universal-paperclips/

endorsed by game author: https://twitter.com/flantz/status/1581736112925511680?s=20&t...

and an older article with the story behind the game: https://www.wired.com/story/the-way-the-world-ends-not-with-...

Re: Universal Paperclips

#35

Post by redox99 » Thu, Nov 03, 2022, 9:04 AM UTC

iNic wrote: post #23

somat wrote: post #22

I can't help but cheat in this game, the source is there, in well written javascript, after a while, I go, "well I will just get over this next hump and give myself a few dozen dingleberrys(or whatever resource is next in the techtree)", after all the only person I am hurting is myself... And it is a clicker game... The lowest and most insidious form of gaming. Honestly at this point I see it as a feature.

What I love is that this exact behavior is one of the minor problems of advanced AI. How can we make sure that a sufficiently advanced AI doesn't just edit it's reward center? Usually called wire-heading or reward hacking. Humans do it too :)

That's basically just a matter of the inputs and outputs of the AI.

Re: Universal Paperclips

#36

Post by simonh » Thu, Nov 03, 2022, 9:07 AM UTC

arc-in-space wrote: post #32

Oh, god, not again, I was intending to do work today.

It could be worse, you could have just lost The Game.

...oops!

Re: Universal Paperclips

#37

Post by willemlabu » Thu, Nov 03, 2022, 9:10 AM UTC

This reminds me of another excellent time sink game: http://clickingbad.nullism.com/

Re: Universal Paperclips

#38

Post by rpigab » Thu, Nov 03, 2022, 9:48 AM UTC

Looks fun, but I'm too lazy to not open the js console and type: clipClick(100)

Re: Universal Paperclips

#39

Post by mr_ploppy » Thu, Nov 03, 2022, 9:59 AM UTC

After 30 minutes I managed to close the browser. I feel like I danced with the devil, but then managed to run out and get in an Uber while they were in the bathroom.

Re: Universal Paperclips

#40

Post by iNic » Thu, Nov 03, 2022, 10:00 AM UTC

redox99 wrote: post #35

iNic wrote: post #23

Earlier quoted context omitted.

What I love is that this exact behavior is one of the minor problems of advanced AI. How can we make sure that a sufficiently advanced AI doesn't just edit it's reward center? Usually called wire-heading or reward hacking. Humans do it too :)

That's basically just a matter of the inputs and outputs of the AI.

If the AI has sufficient "real life" access, think a robot body, then it can do wire-heading. Assuming it is smarter than us, and we have thought of this idea - it will also think of this idea. Now it doesn't have a reason anymore to do anything else (except maybe kill all humans so that we don't stop it from wire-heading).