@kemon. A neat thing beyong higher quality textures and more detailed map optimization would becto give ETL a new set of voice announcements. Remastering the old voice announcements with a new voice would be very cool imo. Anything here planned?
Since they are still protected by the license, we technically have to recreate them, if/when we want to become a stand-alone game and maybe move to Steam.
The problem with sounds and especially voice announcements is, that they are incredibly characteristic for ET. If you look at most custom maps with own voice announcements, then it nearly always sounds weird. Except for Transmitter, shout-out to those guys.
The Transmitter mappers actually recorded those announcements in a professional studio, phisherman contacted them once to inquire about that.
Textures and models can be done fairly easily at home on 'any' kind of set-up. However, for sounds we would actually have to do that in a studio, or find someone with a studio to do that for us. Recording something with a standard microphone at home is just horrible. That's why we moved that waaaay back on the roadmap
But for getting ET: Legacy to Steam, we need to recreate them. We'll see about that in the future. For now they are not planned, sorry.
I am sure a half decend audio neard with a neat computer tool can do magic. New voices would add excitement for sure.
If the quality is at least comparable and it's not immediately recognizable that it is amateur work, then yes. I absolutely agree with you. Own voice announcements would open up a lot of possibilities. Even unconventional map objectives could easily be covered and properly announced. Depending on how we get along with the plans and implementation of own playermodels, we could implement different teams. We would not be limited to map specific skins, but could actually create different nations with their respective quick chat voices. British for siwa oasis for example. The possibilities and the motivation are certainly there.
For starters, to stay realistic, it would be nice to have own custom announcements that cover the basics, yes. Let's first bring the new textures and models into the game and then work on top of that. Until then we'll definitely keep our eyes open for voice actors/volunteers.
If someone knows anyone, maybe even contacts to a Twitch streamer with a lovely voice, please get in touch with us. You can reply here, contact me directly or join us on Discord.
An intresting interview with Kemon from ET:Legacy has been published.
Follow the link and read more what Kemon had to say
No more XP save!
Let's measure your skill instead
Wolfenstein: Enemy Territory is a free and open source multiplayer first-person shooter set during World War II and heavily focused on teamwork. More than a decade after its release, after spending countless hours of escorting tanks and trucks, stealing gold and radar parts and transmitting confidential documents, there comes a project which aims to relaunch and dust off a game which consistently refused to go down.
What are XP?
Enemy Territory implements XPs, which measure player scores based on specific actions, such as kills, construction, or revive. They are also used to control the various skill levels, similarly to a RPG. While not part of the original game, the ability to accumulate XPs indefinitely, as known as "XP save", has been a popular addition in many mods. We all know it: players crave for their XPs. Players accumulate XPs like coins map after map, and they love exhibiting XPs on servers to show how hard and how long they worked to earn them.
The downside of XP save
However, XPs don't give any indication of the real global performance of a player. At best, they merely give a hint about the time the player has played on a specific server. Saving XPs also create some imbalance by enabling special abilities given by skill levels permanently, such as adrenaline for medics which directly leads to the infamous "Rambo medics" that everyone hates. This has never been accounted for in the original game balance, where XPs were reset at the end of matches or the end of campaigns.
With Legacy mod aiming to stay close to the original game design, it makes sense to not encourage the use of XP Save as so many mods do.
Starting from the next 2.76 release, only the XPs earned in the current map or campaign will be displayed.
Instead of the total accumulated XPs, we're introducing a new metric, known as "Skill Rating".
The Skill Rating has been gradually implemented in the past Legacy release, with the final version being currently tested on our official test server.
Connect now to etlegacy.com to have a preview of the system!
The idea is simple:
Skill Rating is a new, proven metric that can be used to compare skill of players over time. As FPS require skills in multiple areas like reflex, planning, tactical analysis or teamwork, with the relative importance of these skills depending on the map, game mode, player roles or team compositions, a skill rating can be defined as a measure of "all parameters of a player that help his team to win" and calculated using a statistical approach.
The idea of using statistics to estimate the skill of players isn't new. ETPub implements such a metric with its "Player Rating" that gives a normalized skill score. However, "Skill Rating" of Legacy mod is modeled after TrueSkill, a more modern algorithm that has been developed by MicroSoft for its XBox matching service.
• It has many advantages over the ETPub's PR, and is very well documented.
• It has been extended to take map bias into account.
• Ranks will also be adjusted to reflect the rating of the player, instead of reflecting the skill levels and XPs.
The better you perform the faster you'll rank up. However beware, several 'bad' matches and you might lose a rank again.
And now fire up the game and /connect etlegacy.com to give the system a test-run on our official test server.
Feel free to join our Discord voice chat for proper teamplay.
What is Skill Rating?
It is a value between 0 and 50 assigned to each player. It represents its ability to help his team to win the map. Rating will be updated after each map played at intermission. New players will have a rating of 0, and it will increase over time. The rating value determined by using the strength of the team, time played in each team and the map. The performance of disconnected players is taken into account.
What do I need to do to have a good Skill Rating?
Do whatever is necessary to win the map.
What are the percentage numbers displayed on scoreboard (using double Tab)?
The first value represent the real time estimate of each team winning the map, while the second is the map bias.
The numbers show that the current game is unbalanced!
The rating system takes balance into account when updating rating. It is thus "fair" to lose an unbalanced map. Winning a map against a much weaker team will not increase your rating much, but losing against a weaker team will make your rating drop sharply. Conversely, to increase your rating quickly, join the weaker team and win as an underdog!
Where do I learn more about this Skill Rating system?
For interested people that want additional details, see the related wiki document and implementation ticket.
Imo XP per minute is still the best single stat metric for measuring skill (assuming XP rewards are balanced).
Problems with trueskill:
- Doesn't promote switching to the loosing team. Why risk loosing skill even if winning won't give you much increase?
- Promotes trying to switch to winning team at the last minute.
- Players that do their best for the loosing team are punished. For example the only engineer on your team and your team finally looses when your respawn time is too long so you can't defuse. You loose skill even if you prolonged the match by 10 minutes single-handedly.
* Winning on a much stronger team gives very minimal reward only, sometime by an unnoticeable amount. No risks, no reward. Similarly, losing on a mush weaker team decreases the rating by a minimal amount only. but winning as an underdog can tremendously increase the rating. Also, keep in mind the algorithm computes an estimated win probability, it doesn't mean the predicted team will necessary win. Players that want to increase their rating quickly should actually play in the underdog team.
* Time played in both team is taken into account into the rating update of each player, switching team at the last minute doesn't work.
* Maybe the team should have thought of a better strategy than having a single engi? Keep in mind a statistical approach removes outliers and reflect long term tendencies. As a good player, you can lose sometimes but you'll still be rated highly if you are good overall. Your rating will be low only if you tend to lose often, which might indicate that you are... a bad player.
I am pretty familiar with trueskill.
* Winning a small amount is still better than risking loosing anything. Trying to join the loosing team when they are much weaker might give a bigger increase, but it's just like gambling imo. The safe bet is to just stick to your stacked team.
* Time played - this is good, I didn't know that.
* Unfortunately in a pub server you can't influence other players and scenarios like that happen all the time. I was hoping there might be a way to support these cases, but I guess not. When I used to have my own servers I often switched to the loosing side, otherwise people quit after the game if it was too unbalanced. So then my rating would be **** since I always loose. Same thing happens in modern games though (looking at you BF1) get put into the loosing team and even if you're on the front lines and get more kills than the rest of your team combined you get no rewards.
If I'm thinking about gaming the system what would happen if I join an unbalanced game and convince everybody to switch sides? Does the win probability get recalculated or does the winning team get a big increase since the original chance of winning was low?
You do raise valid points and I understand the concern. The name "Skill Rating" kind of promotes the feeling that players will do anything to keep their rating up, however the effects are different. We have the system running actively on our testserver (/connect etlegacy.com) and people still balance out the teams. This skill rating actually has a major advantage in terms of fairness compared to XP save.
As I understood it, XP save was introduced so people can compare their performance. The more XP you have the better you are as a player. Which is wrong, because it just shows that you have been playing on a server for a longer time than others. XP save also brings a nasty side effect and that is, that the skill levels are kept (battlesense, light weapons and classes). This provides an impressively huge disadvantage for new players, because everyone else is running around with 156 HP, adrenaline and enough ammo in stock that a Field Ops is rendered completely useless, while a new player comes in with 100HP, doesn't see landmines, has a chargebar that doesn't allow for anything really and has only one spare magazine.
With the new Skill Rating in ET: Legacy, players can now properly compare their skill, which was intended by XP save but didn't work. Everyone also starts a map with a clean slate. For example, nobody sees landmines except the cvops, so it might be useful to have one in your team and go with them. People also only have one spare magazine, a normal health pool, etc. so stay close to your Field Ops and Medics. This feature makes team-play truly necessary again.
The Skill Rating ranges from 0 to 50, which means the 10 ranks are spread over the full range. For example the ratings 21-25 have the same rank, so there is not really a need to worry about losing some matches. Especially, since the system takes team balance into account. Also your rank doesn't give you any benefit in regards to gameplay. The people that will worry about their precious rating are the same people that are currently annoying everyone as rambo-medics with the only focus on K/D and scoring "Highest Fragger". These people will still be there and the people who focus on the objective will also still be there. That is not going to change.
The only real thing that changes, is that skill comparison between players now works properly and that team-play is a major aspect in the game again. The system analyzies a map-bias factor with which maps can be balanced in the gameplay in the future. Theoretically, this feature brings the possibility of having some sort of match-making to revive the competitive scene. There are numerous advantages coming with this feature.
Just because it is called "Skill Rating" doesn't mean that everyone will suddenly switch to lone-wolf gameplay.
I hope that helped clear it up a little.
in addition to kemon more general answer, here is a detailed response to your raised concerns:
This is an excellent resource to grasp the concept behind the original TrueSkill algorithm. You'll also find a link to other relevant resources in the post above if you're interested, but I highly recommend this first link.
While TrueSkill uses the end result to determine rating, the rating update is *not* determined as a simple win/loss ratio, but takes the estimated win probability into account.
Here's a simple analogy to illustrate how this work:
Let's say players A and B make a bet by tossing an unbalanced coin. The result is estimated to be heads 80% of the time, and tails 20% of the time.
Player A wins $5 when the result is head, but looses $100 when the result is tail. Conversely, player B looses $5 when the result is head, but earns $100 when the result is tail. Would you prefer to be player A or B?
The mathematical answer can be found by using the expected value of the variable.
For Player A: E[x]=0.8 * 5 - 0.2 * 100 = -16
For Player B: E[x]=-0.8 * 5 + 0.2 * 100 = 16
In average, player B will earn money, while player A will be the looser. TrueSkill always favors the underdog when winning by a big margin, because when this happens the algorithm actually learns more information about players rating. If A is estimated to be much stronger than B and A wins, the algorithm knows nothing new and will only slightly increase the rating of A. If A is expected to be stronger but loses, then this means B isn't that weak and that the rating of A is overrated - the updated rating will reflect this new information accordingly.
Players A and B are like Team Axis and Team Allies. The individual rating update for each player will then be determined from the Team performance and adjusted according to the initial rating of each players and their respective contribution to the team performance.
The algo has been enabled on our test server for a couple of days, and in practice we've observed the following:
* much stronger team beating a weak team as expected. The rating update is positive, but so small that it is virtually zero (<0.005 on a 0-50 scale). The losing team also virtually looses 0.
* much weaker team beating a much stronger team against all odds, resulting in a very big update. This happened for example the first day of testing with bots, with the Axis team having won only 1 map while the Allies team won the other 13 others first maps. The Axis team was rated very poorly, but won the 15th map, resulting in the Axis players having a much better rating than the Allies players (about +15 points on a 0-50 scale), and despite having won only 2 maps on 15! The Allies players update was similar in size, but indeed negative. Ouch!
In a nutshell, "playing it safe" in the strongest team results most in the time in minimal gains, but with a smaller probability of losing quite a lot. Unbalanced games aren't an issue since the rating system takes them into account.
I haven't played BF1, but from what I researched skill level upgrade is based on a XP systems too. Nobody seems to know how the score system works either, and I haven't found a single reliable source that shows that this system being based on Bayesian statistics.
Last edited by Spyhawk; 17th Mar 2017 at 21:08. Reason: typo
Well even if you don't think I know what I'm talking about I'd just like to clarify a few points.
* With measuing by XP I mean XP per minute not total XP since total XP gained is indeed meaningless.
* I don't think there is any game out there atm that has a perfect rating system. As soon as you take players out of a 1v1 scenario it's very difficult to determine individual skill (on a team).
* What ever system is in place people are going to try and figure out how it works to game it.
* XP per minute is heavily skewed on one side, depending on the map.
* TrueSkill is specifically designed to estimate individual rating from team performance. A rating system isn't meant to be perfect, otherwise it would predict the future: the goal is to approach the reality as much as possible. We have a
pretty good idea on how much reliable it is (TS was tested on the public Halo data set). Also, as a Glicko generalization, uncertainty of the measure is compounded in the rating itself, which isn't done by any measure using XP.
* Players are welcome to try to "game" this implementation. There is only a single variable that players can "game" and that is helping their team to win the map - and this doesn't include switching side at the last minute.
If you have any further question, feel free to ask. Again, I'd suggest to first read the link above, it's really interesting and not too difficult to grasp if you have some knowledge of basic statistics.