Return to “Dev Logs”

Post

Re: [Josh] Friday, April 20, 2018

#46
Flatfingers wrote:
Fri Apr 20, 2018 8:58 pm
In the meantime, I'm interested in learning how the whole-game project is progressing. What's Adam working on? How's the whole game feel with respect to the promises in the Kickstarter? Is there a big-picture plan for getting all the individual features to converge on a Limit Theory 1.0 release?
I'm planning to post a log at the end of the current sprint. It should answer a couple of your questions.
Post

Re: [Josh] Friday, April 20, 2018

#47
RedDwarfMining wrote:
Mon Apr 23, 2018 8:35 am
That's why you simulate the universe.......SETA(egosoft)does this....You take your chance that the universe will change on a 100 ly voyage. (simulated in a couple of hours)

And when you return home...200 years have passed! And pirates have destroyed and looted all your stuff.
Egosoft doesnt. SETA straight up removes/reduces the synchronisation timers between simulation ticks.

The simulation is always the same.
(Besides OOS simulation, but that doesnt remove nearly as much as you suggest)

also: how does that make the data blob you cant load any more useful?

How does that remove the need to run the stuff in "real time"? Your in game time just runs at 100years/[hour/minute] instead of some other rate.
You change the label on "real time", not the actual computational needs.

And if everything can change anyway by the time you next see it you dont need the terabytes of data stored you want.
Because it has no value anyway.
Post

Re: [Josh] Friday, April 20, 2018

#49
Cornflakes_91 wrote:
Mon Apr 23, 2018 8:56 am
RedDwarfMining wrote:
Mon Apr 23, 2018 8:35 am
That's why you simulate the universe.......SETA(egosoft)does this....You take your chance that the universe will change on a 100 ly voyage. (simulated in a couple of hours)

And when you return home...200 years have passed! And pirates have destroyed and looted all your stuff.
Egosoft doesnt. SETA straight up removes/reduces the synchronisation timers between simulation ticks.

The simulation is always the same.
(Besides OOS simulation, but that doesnt remove nearly as much as you suggest)

also: how does that make the data blob you cant load any more useful?

How does that remove the need to run the stuff in "real time"? Your in game time just runs at 100years/[hour/minute] instead of some other rate.
You change the label on "real time", not the actual computational needs.

And if everything can change anyway by the time you next see it you dont need the terabytes of data stored you want.
Because it has no value anyway.
No value?!? Data BLOB?!

Just how many star systems do you expect to visit in LT's infinite universe? 100? 1000? 10,000 out of infinity?

What about the 16 trillion systems in NMS? 99.9999999% will NEVER be visited. Pretty much all worthless. visited one system...you visited them all!

Or Elite Dangerous...99.9999999% of their 400 billion systems that I won't visit(i've visited 4,000 systems in 4 years).

Talk about bloat and useless data!

In a simulator(say: LTSIM), you'll be able to visit/explorer/colonize/conquer trillions of star systems....visit them all.(though I only plan on colonizing the Earth types......maybe have a billion folders with system data)

Moving the sim from volatile memory to non-volatile memory....a fine persistent granular universe.

http://helium-rain.com/ has started down this path.
Post

Re: [Josh] Friday, April 20, 2018

#50
RedDwarfMining wrote:
Mon Apr 23, 2018 4:16 pm
No value?!? Data BLOB?!

Just how many star systems do you expect to visit in LT's infinite universe? 100? 1000? 10,000 out of infinity?

What about the 16 trillion systems in NMS? 99.9999999% will NEVER be visited. Pretty much all worthless. visited one system...you visited them all!

Or Elite Dangerous...99.9999999% of their 400 billion systems that I won't visit(i've visited 4,000 systems in 4 years).

Talk about bloat and useless data!
you mean all those systems that are procedurally generated and need basically no different amounts of storage space if they were 100 or 100000000?

RedDwarfMining wrote:
Mon Apr 23, 2018 4:16 pm
In a simulator(say: LTSIM), you'll be able to visit/explorer/colonize/conquer trillions of star systems....visit them all.(though I only plan on colonizing the Earth types......maybe have a billion folders with system data)
so what?
99% of the data that a system produces is trivial to reproduce and doesnt need to be stored at all beyond working memory.
theres nothing to be gained to store a planet that has been generated from a seed and an algorithm beyond skipping the generation time when revisiting.

RedDwarfMining wrote:
Mon Apr 23, 2018 4:16 pm
Moving the sim from volatile memory to non-volatile memory....a fine persistent granular universe.
you know what? egosoft games have a whole lot of dynamic, persistent data as well.
they dont fill my harddisk with save states either.

RedDwarfMining wrote:
Mon Apr 23, 2018 4:16 pm
http://helium-rain.com/ has started down this path.
3 978 181 847 bytes in size on my harddisk.
3 510 947 840 bytes of which are the assets (models, textures, music)
0 004 970 000 bytes (plus minus something) are claimed by steam to be the size of the savestate

0 462 264 007 bytes remainder is the binary and a chunk of steam dll and directx dll collections

so now where is it going down that route?
Post

Re: [Josh] Friday, April 20, 2018

#51
AdamByrd wrote:
Mon Apr 23, 2018 8:43 am
Flatfingers wrote:
Fri Apr 20, 2018 8:58 pm
In the meantime, I'm interested in learning how the whole-game project is progressing. What's Adam working on? How's the whole game feel with respect to the promises in the Kickstarter? Is there a big-picture plan for getting all the individual features to converge on a Limit Theory 1.0 release?

I'm planning to post a log at the end of the current sprint. It should answer a couple of your questions.

Very much appreciated -- thanks!

Although I suppose I should ask: how long are your sprints? :D
Post

Re: [Josh] Friday, April 20, 2018

#52
RedDwarfMining - I think what Cornflakes is trying to get across is that you'll have a hard time playing LT to the point that the hard drive starts to fill up that far. :) The stuff LT stores is likely to be fairly small in size, and all procedural, so it's not like we'll have to store textures and the like - I expect reaching 100GB of stored data will be quite difficult. Even with a large world and lengthy history genned in Dwarf Fortress, for instance, it's hard to reach a gigabyte. On a different note, E:D has a static economy, while LT does not; the more systems you discover, the more stuff will be happening in the game world. I suspect eventually it would hit the limit of what your CPU could handle. Probably after a fairly long time, of course - abstracting away most of it with LoD could mean you only need an update every X minutes or so for the farthest systems. Eventually, though, I expect you'd see some kind of performance impact.

I'm impressed by your playstyle, though! 4000 systems in E:D is a lot more than I could've expected someone would get to. :P
bkdevil wrote:
Mon Apr 23, 2018 10:19 am
Am I the only one who looked at those screenshots and thought that a pixelated LT could be kind of awesome? Like if you kept the graphics of Star Control 2 but strapped a new brain underneath.
That does sound pretty neat to me, but I don't think you could ever convince Josh to do it! He likes shinies too much. :lol: And for once, I don't actually think that could be done via a mod.
Have a question? Send me a PM! || I have a Patreon page up for REKT now! || People talking in IRC over the past two hours: Image
Image
Image
Post

Re: [Josh] Friday, April 20, 2018

#54
Talvieno wrote:
Tue Apr 24, 2018 9:09 am
RedDwarfMining - I think what Cornflakes is trying to get across is that you'll have a hard time playing LT to the point that the hard drive starts to fill up that far. :) The stuff LT stores is likely to be fairly small in size, and all procedural, so it's not like we'll have to store textures and the like - I expect reaching 100GB of stored data will be quite difficult. Even with a large world and lengthy history genned in Dwarf Fortress, for instance, it's hard to reach a gigabyte. On a different note, E:D has a static economy, while LT does not; the more systems you discover, the more stuff will be happening in the game world. I suspect eventually it would hit the limit of what your CPU could handle. Probably after a fairly long time, of course - abstracting away most of it with LoD could mean you only need an update every X minutes or so for the farthest systems. Eventually, though, I expect you'd see some kind of performance impact.

I'm impressed by your playstyle, though! 4000 systems in E:D is a lot more than I could've expected someone would get to. :P
bkdevil wrote:
Mon Apr 23, 2018 10:19 am
Am I the only one who looked at those screenshots and thought that a pixelated LT could be kind of awesome? Like if you kept the graphics of Star Control 2 but strapped a new brain underneath.
That does sound pretty neat to me, but I don't think you could ever convince Josh to do it! He likes shinies too much. :lol: And for once, I don't actually think that could be done via a mod.
Thanks! Elite is fun and 4,000 is probably on the low end of some of the hard core players!

Unconcerned with validating the idea with Cornflakes ....not looking for his/her approval. The conversation isn't directed at him/her....it's directed @LTdev team and future space programmers.

And I do mean S P A C E programmers!
Post

Re: [Josh] Friday, April 20, 2018

#55
JoshParnell wrote:
Fri Apr 20, 2018 12:48 pm
In the end I'm sure my model will end up being simple (like everything I love)
First off, really fun stuff Josh, thanks for sharing.

I think the issue is the simulation is working too well. It's possible that just increasing "friction to change" could smooth things out the way you want, without changing anything about how you calculate it.

Below are some real world observations:
  • Cyclic Rebounding - the cyclic rebounding you show is exactly how the economy behaves, perhaps it's just the timescales and speed of change that is the issue.. the ridges and valleys perhaps just need to be over the course of 10 years rather than 10 days.
  • Resistance to change - using the 80/20 rule, I'd say most humans actively resist change (changing jobs, houses, cities etc...) you could add in a friction coefficient to the actors that predisposes them to NOT change their activities... people are fickle and "knowing a better action exists" != "doing it" for large segments of the population.
  • Imperfect information - People do not always "have" the best information, and economic theories often run up against this baffling truth... thus factoring in some actors choosing a "random" action rather than the logical one, in addition to a higher chance of inaction may simulate this.
  • People are stupid - Me included, I "knew in my bones" the housing bubble was going to burst, I did nothing. I knew I should buy stocks after it burst. I did nothing. So, people are dumb, even if they have perfect information. I'd suggest running a simulation that overestimates this type of stupidity, plus this may help keep the universe interesting... some poor sod will venture out into the outer belt, even though he knows pirates are lurking there.
  • People are slow - it takes time to learn enough about pork bellies to actively trade them, or to sell your house, find a school for the kids, change your job and move to a new quadrant of the asteroid belt to mine iridium, especially if you are an accountant. The timescales that these changes happen should be more protracted, so as they cannot necessarily be timed to perfectly coincide with changes in the price of water this month, since it takes you 2-3 months to move.

I hope these thoughts help and aren't redundant! Keep up the great work!
Post

Re: [Josh] Friday, April 20, 2018

#58
RedDwarfMining wrote:
Tue Apr 24, 2018 1:37 pm
Unconcerned with validating the idea with Cornflakes ....not looking for his/her approval. The conversation isn't directed at him/her....it's directed @LTdev team and future space programmers.

And I do mean S P A C E programmers!
him.

and you addressing josh doesnt remove you from scrutiny by basic logic.
meaning that theres absolutely no point with current computers and gaming trying to fill hundreds of gigabytes of data.
you either cant use the data, because it'd need that much time to process (loading and doing actual computation) or you are really crappy at pruning the savestate, aka wasting space.

neither of which is in your interests (beyond your weird attraction to full harddisks)
Post

Re: [Josh] Friday, April 20, 2018

#59
Baile nam Fonn wrote:
Tue Apr 24, 2018 3:30 pm
jwmickelson wrote:
Tue Apr 24, 2018 2:45 pm
[..]
Just want to say that I thoroughly enjoyed this post. Great stuff and a delight to read.
This should be in the simulation, all of it, imho. :thumbup:
Thanks! It's easy to see why Josh enjoys thinking about this part, I don't envy him having to translate it into code :D

Within procedural generation there seems to be an eternal battle between creating the lightest weight algorithm that produces a believable result and a simulation that is highly realistic but computationally expensive. The former is really what Josh needs here, so where a full economics simulation may have many discrete variables, Josh just needs the right macro variables, plus some random noise, that might create a similar result.

It's interesting to me, that when using Procedural Generation for visuals, such as landscapes and planets, there is visual feedback the coder can look at to see if it's producing realistic results. For PG behavior, you are balancing the macro results, like the economic rebounding Josh was showing (which is why he needed a tool to let him see those things... his "economic landscape picture over time"), but he also has to evaluate effects on behavior of a single ship. For instance, if the you are fighting a pirate who, just noticed the price of water jumped and he's now decided to become a commodities trader instead ;)

Suffice it to say, I can't wait to play LT! :thumbup:

Online Now

Users browsing this forum: No registered users and 16 guests

cron