NationStates

As repeatedly promised earlier, this is the topic to discuss a long awaited "Next WA Update" feature. There is another topic which, once opened, will be the official space for comments. This topic, however, is for players to discuss the feature with one another and ask clarifying questions. The description of the feature may be revised prior to opening the other topic for comments for better clarity, or to correct any mistakes.

Be advised that, as hinted earlier, I am also bringing up for discussion two further changes I believe would improve the update, in other topics.

Ground Rules
This topic is intended to be a way for site administration and the community to discuss the feature and improve understanding. This topic will not be permitted to become a flaming free-for-all of an argument, useless to everyone and infuriating to many.

1. A player may post using one (and only one) nation in this topic. Do not use puppet nations.
2. Please address other players only in a completely respectful and cordial manner.
3. Please keep in mind site rules in general and the bad faith policy specifically (which applies to this topic).

Eluvatar wrote:Next WA Update
This is a topic that has seen some previous discussion, with considerable confusion all around.

There was a previous plan called "Estimated Update Times Displayed". This is not that plan.

Instead of the current "Last WA Update: 12 hours ago" (with hover text "11/1/2017, 12:21:05 PM EDT") we would have something like "Next WA Update: in seconds" (with hover text/clickable timestamp "11/2/2017, 12:28:07 AM EDT ± 5 seconds"). Clicking "Next WA Update" would switch back to "Last WA Update". Clicking on the relative time would turn it into a timestamp (which is currently only rendered as title text, displayed on hover). The time shown for Next WA Update would be dynamically computed based on the number of nations before the region in the update (and, during update, how many have already updated). As a result, the time shown will change over the course of the day and over the course of the update.

For the sake of full transparency, I should note that when implementing this feature I may also adjust the variance system. The amount of random variance will be reflected in the Next WA Update prediction. Unless I am persuaded otherwise, I intend to aim for the following: if Next WA Update indicates a time 10 seconds from now, the region will update within 5 seconds of the indicated time. I believe such a 10 second window is appropriate for update Gameplay. Finally, all updates will in the future be approximately one hour long.

The goal of this change is to allow players to better understand gameplay as it happens to them or they make it, and participate without a great deal of training. Furthermore, update time prediction tools should no longer be necessary to participate.

Please provide input regarding the pros and cons of showing the next update time, of a 10 second window, or related matters. If you have suggestions for improving this plan, I would welcome them.

Please discuss the pros and cons of showing the next update time, of a 10 second window, or related matters. If you have questions about the feature as described, please feel free to ask. Please keep in mind, however, that I do not guarantee I will read every post in this topic with full care and consideration: that is the purpose of the request for comments topic.

Eluvatar wrote:AFurthermore, update time prediction tools should no longer be necessary to participate.

This sounds different from what we've heard before. Before it was that the displayed time would be the most accurate. Is this suggesting that, like now, it is still possible to get more accurate?

I think that would be a good balance between allowing for skill in triggering, and making it easy for new players to get into.

I'm all for it, and I like that we're not calling it elections. As for specifics I much prefer the version where the potential variance is indicated, it's just one less thing people need to learn about before jumping in.

Raionitu wrote:

Eluvatar wrote:AFurthermore, update time prediction tools should no longer be necessary to participate.

This sounds different from what we've heard before. Before it was that the displayed time would be the most accurate. Is this suggesting that, like now, it is still possible to get more accurate?

I think that would be a good balance between allowing for skill in triggering, and making it easy for new players to get into.

I don't intend to try to make the posted time absolutely definitely the most precise possible prediction. I intend for it to be accurate.

I do not like estimated update times with a window of accuracy. Depending on the window frame, you heavily skew things one of two ways. I will exaggerate a bit here to convey the point -

If the window is quite narrow, say, +/- .5 seconds, you have a total window of 1s in which the region will update. This makes all offensive actions effectively pinpoint accurate. Raids are exactly on time. Chasing and intercepting is near impossible. Liberations are also precisely on time, no need to try and judge how early or late to show up to balance the risk of being late. No skill involved in timing.

If the window is quote wide, say, +/1 10 seconds, you have a total window of 20s in which the region will update. This makes defensive actions very easy, on average. If the offensive forces aim for the front of the window, which they generally will because being late is a total non-starter, there is now likely a good amount of time to respond. Could be 0 seconds, could be 10 seconds, could be 20 seconds. That's luck of the draw, which I dislike to start with (skill should play in more than luck). It'll also, most of the time, mean plenty of response time. Tagging becomes incredibly difficult AND slower (not only will defenders generally have plenty of time to chase and intercept, but with any window, tags will need to be planned further apart to ensure time to switch between latest possible on one and earliest possible on another). Defeating liberations becomes far easier, with likely plenty of time for the occupiers to eject liberators. Starting operations becomes much harder, with easier chasing. Again, no skill involved in timing.

So, either way you're throwing things way out of the relative balance we have currently, in favor of either the aggressor or the defender, as well as reducing skill, either by just removing it entirely (narrow window) or by replacing it with the GP equivalent of slot machine luck (wide window). Finding a balance between the two windows would be incredibly difficult at best, require extensive data collection and argument, and still ultimately reduce the factor that skill plays in the game in favor of one thing or another.

Ever-Wandering Souls wrote:I do not like estimated update times with a window of accuracy. Depending on the window frame, you heavily skew things one of two ways. I will exaggerate a bit here to convey the point -

What would an estimated update time without a window of accuracy look like?

Ever-Wandering Souls wrote:If the window is quite narrow, say, +/- .5 seconds, you have a total window of 1s in which the region will update. This makes all offensive actions effectively pinpoint accurate. Raids are exactly on time. Chasing and intercepting is near impossible. Liberations are also precisely on time, no need to try and judge how early or late to show up to balance the risk of being late. No skill involved in timing.

If the window is quote wide, say, +/1 10 seconds, you have a total window of 20s in which the region will update. This makes defensive actions very easy, on average. If the offensive forces aim for the front of the window, which they generally will because being late is a total non-starter, there is now likely a good amount of time to respond. Could be 0 seconds, could be 10 seconds, could be 20 seconds. That's luck of the draw, which I dislike to start with (skill should play in more than luck). It'll also, most of the time, mean plenty of response time. Tagging becomes incredibly difficult AND slower (not only will defenders generally have plenty of time to chase and intercept, but with any window, tags will need to be planned further apart to ensure time to switch between latest possible on one and earliest possible on another). Defeating liberations becomes far easier, with likely plenty of time for the occupiers to eject liberators. Starting operations becomes much harder, with easier chasing. Again, no skill involved in timing.

So, either way you're throwing things way out of the relative balance we have currently, in favor of either the aggressor or the defender, as well as reducing skill, either by just removing it entirely (narrow window) or by replacing it with the GP equivalent of slot machine luck (wide window). Finding a balance between the two windows would be incredibly difficult at best, require extensive data collection and argument, and still ultimately reduce the factor that skill plays in the game in favor of one thing or another.

I imagine that skill in understanding what sort of timing is advantageous would remain?

I believe in the design I stipulate a 10 second (± 5 seconds) window (once you're close enough in time that the game can stop trying to estimate 'natural' differences due to nations moving around or future server load). I think a last minute edit of the example may have gone poorly, I'll go back and adjust that.

What would that look like to you?

I'm not a fan of precise estimated times, period, because they're either exact or have a window, and either way reduce the skill needed heavily. Last Update has done well at serving the need of "give natives an idea of about when the region will update again." Changing that to last major+last minor, or having equal length updates, would improve the utility of that without adding in all the effects of an estimated future time.

Understanding what is advantageous? If it's precise, what's advantageous is jumping on that precise time. If it's wide, your choice is either to jump at the very start and 100% not be late, or take a total gamble of luck and jump sometime in the middle, hoping to narrow the gap and also not be late.

+/- 5 seconds, i.e. a ten sec window, would be something I'd consider fairly wide. Anything that has the potential to add more than a second or two basically ensures that all defenders chasing will get in. As is, we juggle a balance between our slower updaters getting in, and the mass of defenders chasing. The ideal time probably cuts off some slower raiders, but only allows the fastest defenders in. Going wider for the sake of two raiders is not worth it if it allows in 6 defenders. A lot of skill goes into working to set the trigger, watch update trends, and call the trigger as close to that decided and desired window as possible. If I want the trigger to be 6 seconds, I want to see it be no greater than 7 or less than 5, so +/- one. A good trigger on a good day can absolutely do that. A lesser trigger or a good trigger on a bad day should still be able to get with +/- 2, or a window of 4 seconds. You're talking about a window of 0-10, added on top of that natural skill window. A trigger or jump directly off the clock with estimated times could be *perfect*, but then we draw a short straw, update doesn't happen for 5 more seconds, all the chasing defenders make it in. You're both killing the skill of trigger by reducing it to "watch this internet clock, and jump at this time exactly" and adding in a slot-draw of added time after the fact.

So, IMHO, a window of 2s is pretty comparable to the existing scene. We still have to balance the move speed of various updaters, so we'd probably still set a 6 seconds "trigger" before the start of the window, and then that +/- 1s (2s window) added onto that is comperable to potential "off-ness" of a good trigger's call to actual update in the present world. As said, a trigger that I'd consider acceptable but needs practice would be closer to +/- 2s (4s window), and +/- 3 (6s window) would be on par with a trigger just starting their training.

Forgot two other notes

1 - Wide windows also heavily counteract the goal of making things more approachable to newcomers.

Why?

Numbers.

As already discussed, wide windows add free time, potentially a lot of it, in which defenders can chase. This has a disproportionately large effect on newcomers, who are far more likely to be a small group jumping with 2-3 people. We have a hard enough time running tags with more than 3-4 people, and many startups operate with just two people jumping. Increasing the ability of a reactionary force to chase them/removing their ability to gain an edge despite their size by being good triggers is a negative factor towards their involvement.

2 - Rendering triggers effectively useless by providing times also reduces overall participation. At any given time, about half of our tags or operations are being triggered by people immobile for reasons (both R/D like sleepers, and in some cases due to being involved legitimately in other region). Removing triggers reduces a valuable way in which they can participate and contribute without a WA. They could still plan to the effect of finding a list of targets, but an excel sheet can also do that automatically in 30 seconds with no effective difference ;P

As I understood it, this change would not eliminate traditional triggering methods.

Generally I think anything that lowers the skill floor while keeping the ceiling the same height (or even better raising it) is usually better. As long as traditional triggering can allow for more precise timing, I'd be fine with this change.

Disseminating this kind of information to the public actually helps several groups. Notably newbie military groups and regional governments. It sure makes it easier for regional governments to hand off power by showing exactly when the changeover will occur and when they need to be aware of any attempt on any outside operation.

Roavin wrote:As I understood it, this change would not eliminate traditional triggering methods.

It would not make them impossible. It would make them useless. What's the point? If you have a window set to a clock, you aim for the front of that window, minus however long you need to move. A triggers adds nothing but complexity and variability.

Even if it's possible to read trends (i.e. all of these are updating towards the end of the window, update is running slow right now), there's no need to have a trigger to adjust. You just tell people to move 4s before the window starts, rather than 6, etc.

So, eliminate? I mean, they'd be possible. Have any reasonable benefit? I struggle to see how they would.

As a raider - no. It's supposed to be hard to trigger - you have to know what programs to use, have to know update length, etc. Raiding is supposed to be hard, and this would make it too easy.

I agree with Souls on this. By doing this you will be removing a lot of skill away from the game. It would cause the use of triggers to become much less valued with less for them to be able to actively be involved in gameplay for. As it currently exists users are able to watch at update and see when update occurs by it saying it happened seconds ago or however many minutes ago. It then allows them to have a basic understanding of when the region updated. They are also able to go into regional happenings currently and be able to see the exact time that nations influence changed in the region. That in itself would make it possible for others to participate seeing around when the region updated. As Souls has said a window will either go to far in one way or another and take the skill out of the game which having the skill involved makes it enjoyable. Also adjusting the variance system is not something that we would like done. It has been adjusted in the past making variance worse for updates months ago and having it adjusted it even more to cause a higher degree of variance would tip the level of balance of gamplay even further away for those on the raiding side.

Regarding "window" vs triggering: I should repeat that my intention is not to change how update works fundamentally. I will, at most, tweak the Variance code. Therefore, triggering via nations will work about as well as now. I'll certainly try to give as accurate a measure of the precision of the estimate presented as I can, but it's certainly possible that triggering will still be a more precise method.

Benjabobaria wrote:As a raider - no. It's supposed to be hard to trigger - you have to know what programs to use, have to know update length, etc. Raiding is supposed to be hard, and this would make it too easy.

The goals of this change run counter to the values you've articulated. If you wish to submit a comment along these lines, I'd encourage you to establish what values exactly you think are more important than inclusion of more players, and why you believe they are more important in terms of improving the game.

I am going to raise the point that defenders right now can move into a region within the exact same second as raiders. I think the reason we needed update calculation tools has completely flown over everyone's heads. It is because unless you make a _really_ good manual trigger, your trigger is not going to be tight enough.

And now you're going to make them even less accurate? I have absolutely no issue with provided the best possible time on the site. I actually would love that. However a +/-5 second window of accuracy? Tag raids aside, unless you're really lucky, you're basically saying "Depending on how RNGesus and our lord and savior Variance feels, you may as well have a 50% chance of missing, a 10% chance of hitting, and a 40% chance of defenders pressing F to pay respects as NSBreeze sticks them in the region instantly."

You can say all day that "Defenders suffer from it too" because they have to do liberations, and whine about how tag raids are bad and should die off (this will kill tag raids). But all this is going to do is kill what little activity most raider regions have left (tag raids have made up most raider activity recently), because they can't afford to do anything but play the numbers and politics game. There won't be a choice because the chances of them actually hitting anything are so astronomically low that if they don't have more people than the opposition, they aren't even guaranteed to hit unless they jump 10s in advance.

^what aurum said, with the added reminder that new R/D entrants also lack numbers, and thus will suffer from it even more than established orgs which at least have a shot at having more people on at update than defenders.

From what I understood of the suggestion Elu plans to add a semi-estimated semi-accurate timeframe for the region to update in. He won''t be making it impossible for say, Miso to work, or for Money to use his triggers. As far as that goes, I really see absolutely no problem with this feature. If you don't think 10sec windows are accurate enough, then can't you just use a manual trigger?

Chingis wrote:From what I understood of the suggestion Elu plans to add a semi-estimated semi-accurate timeframe for the region to update in. He won''t be making it impossible for say, Miso to work, or for Money to use his triggers. As far as that goes, I really see absolutely no problem with this feature. If you don't think 10sec windows are accurate enough, then can't you just use a manual trigger?

He mentioned adjusting variance as well, though.

He was kinda vague about whether he will or won't. Though yeah, unless he plans to adjust in in favor of accuracy I don't think it would be a great idea. Variance right now is a killer.

Variance right now is fine. Honestly, the game as it is right now is fine.
What's going to happen is that the absolute best time possible will be displayed on the main site (and hopefully provided as an API shard)

That means that your jumps are at the mercy of what the site provides. And there is a pretty significant non-zero chance that the time the site provides will tell you to jump after the region has already updated.

If there's no way to at least try to predict exactly when the region will UD within the variance window, triggering does indeed go away as a useful option and forces all raiders to either risk missing entirely, or go in right at the start of the window - leaving them 100% wide open to defenders and awake WADs or BC ROs.

And the same would hold true for Libs.

As I've been all along, absolutely for the change. The TBH posters I'm guessing are either misunderstanding the topic, or trying to prevent other groups from being able to participate to increase their own regions recruitment. So will Next WA update appear in the API/dumps? I'm guessing it won't appear in the daily dumps as it likely would be too far removed.

With the one hour updates, can it be aimed for 59 minutes etc. - just so it stays under an hour? I would prefer 59:59 exactly, but given population fluctuations etc. slightly less works so there is room for expansion. It's doable saying to people that I'll be doing something for an hour, but more than hour is more difficult to sort (although doable, and it's more a preference for it to be less than an hour).

Nothing prevents other groups from raiding. All you need is an update order, and as I understand it, that's pretty freely available. After that, it takes a little effort to find a target, figure out the rough time and then find a trigger.

Flanderlion wrote:I'm guessing are either misunderstanding the topic, or trying to prevent other groups from being able to participate to increase their own regions recruitment.

Eluvatar wrote:

Eluvatar wrote:Instead of the current "Last WA Update: 12 hours ago" (with hover text "11/1/2017, 12:21:05 PM EDT") we would have something like "Next WA Update: in seconds"

For the sake of full transparency, I should note that when implementing this feature I may also adjust the variance system. The amount of random variance will be reflected in the Next WA Update prediction. Unless I am persuaded otherwise, I intend to aim for the following: if Next WA Update indicates a time 10 seconds from now, the region will update within 5 seconds of the indicated time. I believe such a 10 second window is appropriate for update Gameplay. Finally, all updates will in the future be approximately one hour long.

Please provide input regarding the pros and cons of showing the next update time, of a 10 second window, or related matters. If you have suggestions for improving this plan, I would welcome them.

Please discuss the pros and cons of showing the next update time, of a 10 second window, or related matters. If you have questions about the feature as described, please feel free to ask. Please keep in mind, however, that I do not guarantee I will read every post in this topic with full care and consideration: that is the purpose of the request for comments topic.

Me on Discord wrote:5s either side still means you have a pretty significant non-zero chance to be given a time after the region updates.
It means there's a 10s window, and the time it gives you can be anywhere inside that 10s window

Assuming you only care about those 10 seconds, he's a timeline
0 1 2 3 4 5 6 7 8 9 10
If the time it updates is 5, that means it can say that it updates at 10, 5 seconds after it actually updated.

No, nobody has misunderstood it.

Ideally what I would rather see than a "On average you have a 50% chance of missing, a 40% chance of getting defended, and a 10% chance of hitting" is "On average you have a 66% chance of getting defended, and a 33% chance of hitting" because I'd rather see a defender delegate than no delegate change at all.

To do this, the time provided will NEVER be later than the region's actual update time, and the variance is somewhere in the range of 3 seconds.

This means that people trying to pick up the game won't be frustrated because they can't ever hit anything, defenders trying to lib and raiders trying to raid will actually be able to get into regions before they update, overall reducing barriers to entry for R/D

Which is less bad, though I still dislike rendering triggers practically useless for multiple mentioned reasons.