Advertisement
by Lumio Magika » Mon Mar 29, 2021 3:54 pm
Lumio Magika wrote:Burn them with holy fire until nothing but ashes remain!
by Trotterdam » Mon Mar 29, 2021 10:20 pm
by Eshialand » Wed Apr 07, 2021 11:10 am
by Racoda » Thu Apr 08, 2021 1:10 pm
by Trotterdam » Fri Apr 09, 2021 1:01 pm
Racoda wrote:Would you be willing to share/publish your code?
Trotterdam wrote:What do you want this for?
In principle, I'm willing to share my code if it would be helpful. However, I think this would be of limited use because (A) it's written for a fairly specific programming environment that won't simply run out of the box for most people, and I also have a bad habit of not commenting my code that much, and (B) I don't actually encourage other people to run a program equivalent to mine, since it'd just be redundantly gathering the same data (and if we were to combine both versions' data into a single better repository, that could be seen as violating the rule against sneakily exceeding ratelimits).
by Paffnia » Sat Apr 10, 2021 8:27 am
by Trotterdam » Sat Apr 10, 2021 1:21 pm
Umm, I've already had this one up for days? Is something wrong?Paffnia wrote:#1444, Option 1 (internally numbered 0) effect line: "Returning vacationers often find that their property is now a construction site."
by Paffnia » Sat Apr 10, 2021 7:16 pm
Nope, I just hadn't refreshed the page!Trotterdam wrote:Umm, I've already had this one up for days? Is something wrong?
by Rocain Founder » Tue Apr 20, 2021 6:14 am
by Vicken » Tue Apr 20, 2021 1:49 pm
Trotterdam wrote:Racoda wrote:Would you be willing to share/publish your code?Trotterdam wrote:What do you want this for?
In principle, I'm willing to share my code if it would be helpful. However, I think this would be of limited use because (A) it's written for a fairly specific programming environment that won't simply run out of the box for most people, and I also have a bad habit of not commenting my code that much, and (B) I don't actually encourage other people to run a program equivalent to mine, since it'd just be redundantly gathering the same data (and if we were to combine both versions' data into a single better repository, that could be seen as violating the rule against sneakily exceeding ratelimits).
by Trotterdam » Tue Apr 20, 2021 5:17 pm
Hmm. I suppose there is some merit to this concern. I don't intend to ever abandon this project (certainly not without giving fair warning first), and I am currently in good health, but there's always some risk of a freak accident.Vicken wrote:I suppose that in my case the main reason I'd desire access to the code is as a sort of safety measure in the case you stop supporting it or disappear from nationstates. I use this data to programmatically run a bunch of puppets, so if suddenly the site went down or isn't updated any more I'd be in a bit of a pickle (and have to rewrite a similar thing).
The basic program logic is really pretty simple. It can be summarized as:Vicken wrote:I've also been wanting to look more in-depth on the distribution of issue outcomes or see how they change as a function of the nation's visible stats. It'd be much more practical to be able to see how it's done here than completely reinventing the wheel. Not that this is a thing I plan on doing any time soon, but it's definitely on my mental list of possible projects.
Thanks!Vicken wrote:(As this is my first post here, I have to end this comment with a thank-you. This is a great resource!)
by Vicken » Tue Apr 20, 2021 11:06 pm
Trotterdam wrote:The basic program logic is really pretty simple. It can be summarized as:
1. Whenever the program doesn't have something better to do, constantly check the world happenings feed for nations answering issues. API query: q=happenings&filter=law&limit=200&sinceid=<last_world_happening_seen>
2. For any interesting-looking happenings (some filtering is done to get rid of happenings that obviously can't be used), check the nation's new stats. Store these provisionally, but don't do anything with them yet. API query: q=dbid+wa+policies+census&scale=all&mode=score&nation=<nation>
3. Once at least 30 seconds have passed since a nation's data has been provisionally retrieved (which, at least in theory, should always be longer than the 28-second delay that the API manual claims there is before happenings become visible, though I sometimes wonder if that's actually reliable given some suspicious data I'm getting), check the nation's happenings again to make sure it hasn't done anything new that would invalidate the data. API query: q=happenings&filter=law+member&limit=2&sinceid=<previous_happening_processed_from_this_nation>
4. If both happenings retrieved in step 3 are what was expected, then the retrieved stats from step 2 can be compared against the previously-stored stats of that nation to determine the changes. If the last happening retrieved in step 3 is correct (i.e., the same as the one gotten in step 1) but the second-last happening is incorrect or missing (or if no second-last happening was expected because the nation wasn't previously in the database), then the data cannot be used to determine issue effects, but it can still be saved as the nation's current stats in preparation for being used to calculate the effects of the nation's next issue. If even the last happening has already changed from what it's expected to be, then the data can't be used at all and the nation is dropped from the database.
That's it. Those three API queries make up the bulk of what I'm doing. (There's also a daily check to retrieve a list of all WA nations to save time on having to disqualify them each individually, but that's not essential.) I think I actually spent more time on writing the user interface to make stuff like adding new issues and identifying macros easy than I did on the core logic.
by Trotterdam » Wed Apr 21, 2021 4:41 am
Oh yeah, definitely. It doesn't take up very much of my time, but I do need to regularly check in, even if just briefly. And of course, any corrections reported in this thread need to be entered manually.Vicken wrote:Sounds like there's still a lot of manual work to it all though :blobsweat:
by Marxist Germany » Sun Apr 25, 2021 3:54 am
Companies have begun running campaigns to stop people from remembering their brand name.
at least 4 unique companies in @@NAME@@ claim to sell Eckie-Cola
by Trotterdam » Sun Apr 25, 2021 10:40 am
No, I most likely just made a mistake. Fixed now, and do tell me if you see any other cases like this.Marxist Germany wrote:Are there several possible effect lines for some options?
by Trotterdam » Wed May 05, 2021 12:59 am
by Trotterdam » Sun May 16, 2021 9:57 am
by SherpDaWerp » Mon May 31, 2021 6:21 pm
by Trotterdam » Mon May 31, 2021 11:25 pm
by Rocain Founder » Tue Jun 01, 2021 5:09 am
by SherpDaWerp » Thu Jun 03, 2021 2:52 am
Trotterdam wrote:Urgh. This calls for some testing.
What I need is for someone to take a (non-WA!) nation, record its stats (https://www.nationstates.net/cgi-bin/api.cgi?q=census&scale=all&mode=score&nation=<nation>), wait a few days or weeks without answering any issues (dismissing them is fine), retrieve the stats again, and check which ones changed. I distinctly remember that, back in the day, someone reported Averageness rising even when nothing is being done. I never checked Ideological Radicality, and it's also possible that the behavior of one or both stat has changed since this was last tested.
Rocain Founder wrote:Most Extreme probably would not need any such behavior. I would think it would be very hard to create a nation that would begin its existence at the top of the Most Extreme ranking.
by Valentine Z » Thu Jun 03, 2021 3:33 am
♪ If you are reading my sig, I want you to have the best day ever ! You are worth it, do not let anyone get you down ! ♪
Glory to De Geweldige Sierlijke Katachtige Utopia en Zijne Autonome Machten ov Valentine Z !
(✿◠‿◠) ☆ \(^_^)/ ☆
♡ Issues Thread ♡ Photography Stuff ♡ Project: Save F7. ♡ Stats Analysis ♡
♡ The Sixty! ♡ Valentian Stories! ♡ Gwen's Adventures! ♡
• Never trouble trouble until trouble troubles you.
• World Map is a cat playing with Australia.
by Trotterdam » Thu Jun 03, 2021 3:52 am
Oh, I'm pretty sure that census scores don't actually depend on comparisons to other player-run nations. However, I definitely recall someone, way back, finding that Averageness steadily increases over time even when dismissing/ignoring issues (I think it was one point per day). However, that may no longer be the case.SherpDaWerp wrote:Maybe the whole "comparison to world average" thing was a misconception?
I also recall finding that Ideological Radicality corresponded to roughly "abs(civil_rights - 50) + abs(economic_freedom - 50) + abs(political freedom - 50)". However, you need to take care over the difference between smooshed and unsmooshed freedom scores, and I believe there was a small additional term that doesn't make a big difference, but that I never found an adequate explanation for.SherpDaWerp wrote:Come to think of it, that should have been an indicator that maybe the two stats aren't related to world averages - there's a couple thousand default-anarchy Womble puppets that would drag the "average" up towards a default-anarchy puppet, so you'd think it wouldn't start with a high ranking in Ideological Radicality...
No. Mind you, a one-child policy isn't even API-checkable.SherpDaWerp wrote:Side note, rather than double-posting: have you noticed any reasonable validities for #1353, beyond what's mentioned in the draft thread?
by SherpDaWerp » Thu Jun 03, 2021 5:09 am
Trotterdam wrote:SherpDaWerp wrote:Come to think of it, that should have been an indicator that maybe the two stats aren't related to world averages - there's a couple thousand default-anarchy Womble puppets that would drag the "average" up towards a default-anarchy puppet, so you'd think it wouldn't start with a high ranking in Ideological Radicality...
I also recall finding that Ideological Radicality corresponded to roughly "abs(civil_rights - 50) + abs(economic_freedom - 50) + abs(political freedom - 50)". However, you need to take care over the difference between smooshed and unsmooshed freedom scores, and I believe there was a small additional term that doesn't make a big difference, but that I never found an adequate explanation for.
Trotterdam wrote:No. Mind you, a one-child policy isn't even API-checkable.SherpDaWerp wrote:Side note, rather than double-posting: have you noticed any reasonable validities for #1353, beyond what's mentioned in the draft thread?
I do note that the draft says "should come as an immediate issue", which I don't think is how chain issues work (they're not supposed to be able to have additional validities, unless the original issue was edited to have variant options where one gives the followup and one not). I reactivated tracking to check whether this is actually the case, but that may take a while, since this seems to be a really rare issue. My tracker currently lists only 20 data points (1 for the first option, 12 for the second option, 7 for the third option). Though oddly, the one data point for the first option is reported as being from just four days ago (which means that before then, it must have been in my database for a long time without successfully collecting any data points, since the effect line itself has been known for a while), while the other options are around a month ago.
Advertisement
Users browsing this forum: Totally not an Arlandian spy
Advertisement