The math behind DbD's MMR is just a single "if" statement (900h vs 11,000h lobby analysis
Comments
-
- The "What if they fail?" Scenario
You argue: "If they don’t [convert that time into generators], the chase alone wins nothing."
You act as if a performance-based MMR would break in this scenario. It wouldn't. Here is exactly what happens in a properly designed system when a player loops for 5 minutes but the team fails to do objectives and everyone dies:
The outcome is a team loss. However, the three teammates who contributed nothing get a massive MMR penalty. The looper receives a heavily mitigated MMR loss (or a slight positive bump) because their telemetry proves they performed their role exceptionally well despite the team's failure.The ultimate irony? Dead by Daylight already does this exact calculation. In your scenario, the current game awards the looper maximum Boldness Bloodpoints and Emblems, while the teammates get nothing or less. The engine perfectly understands who contributed and who didn't. You are just defending BHVR's refusal to feed that existing, highly accurate data into the matchmaking algorithm.
- The Ultimate Irony: "Escape" is the Noisiest Data of All
You dismiss telemetry like chase time as "noisy data" because it involves map RNG, killer decisions, and teammate interference. But you are fiercely defending "Survival" (the 1/0 outcome) as the ultimate metric.
In an asymmetrical 1v4 game, Survival is the single noisiest, most RNG-heavy data point in existence. Whether a Solo Queue player survives depends on: Did a teammate DC at 5 gens? Did someone farm them off the hook? Did the killer decide to be friendly and give them the hatch? Did the killer bring a Mori?
You cannot mathematically claim that "Chase Time" is too noisy to measure individual skill, while simultaneously defending "Did you survive?"—a metric that is 90% out of the individual player's control in Solo Queue.- The "Hatch" Defense Fails
You argue a locker-hider will escape inconsistently and settle at a lower rating. You are entirely missing the point.
The fact that the algorithm awards them any positive rating at all for a zero-contribution game, while handing a negative rating to the teammate who actually progressed the objective but died on a hook, proves the feedback loop is completely detached from gameplay reality. A system should never reward a 0% contribution with a "+1".
3 - The "What if they fail?" Scenario
-
Thank you for reading my post, and don't worry, I have no intention of downvoting you. In fact, I am going to upvote you, because the irony here is that we actually completely agree. I always read carefully and with understanding, even 2-3 times.
You just wrote a perfect summary of exactly why the current system is fundamentally broken and needs to be replaced. Let’s break down your own points:
- "DBD is asymmetrical and many things shape the game"
You correctly pointed out that maps, perks, add-ons, and Killer choices (Nurse vs. Trapper) create massive variance. Exactly. Because Dead by Daylight is such a highly complex, variable-heavy game, using a primitive, flat 1/0 (Kill/Escape) metric is the worst possible way to measure skill. A complex game requires a complex algorithm. You cannot measure a multi-variable environment with a binary ruler. - "Skill is hard to determine... especially when mmr is supposed to be set on kills and escapes"
You hit the nail on the head. Skill is hard to determine when you only look at the final outcome (Kills/Escapes). That is the entire premise of my thread. A 1/0 system cannot tell the difference between a 1,000-hour Nurse who played brilliantly and a 1,000-hour Killer who just stood in front of a hook while the team altruistically killed themselves. - It’s actually not that hard to measure (The Telemetry)
You mentioned that FPS games use K/D ratios. But modern, competitive FPS games (like Valorant or CS:GO) also use underlying telemetry like ADR (Average Damage per Round), utility usage, and objective time to adjust your rating.
DbD can easily do the same. As I mentioned to other users here, the game already tracks your chase time, generator efficiency, and safe unhooks via the Emblem and Bloodpoint systems. The data exists. The engine already knows who the MVP of the match was.
So, thank you for your comment. You perfectly highlighted why an outcome-based MMR fails in an asymmetrical game, and why we desperately need a system that tracks actual performance and telemetry. Welcome to the team.
6 - "DBD is asymmetrical and many things shape the game"
-
Your examples all depend on the same assumption: that the system can reliably determine why something happened in a match. It can’t.
In your 5 gen looper example, you assume the system can prove the chase caused the generators. Telemetry can’t isolate that. A long chase could come from survivor skill, killer mistakes, strong tile spawns, power matchups, or the killer overcommitting. The system only sees the duration of the chase, not the cause. That’s the attribution problem you keep overlooking. You still have yet to address this issue in your argument.
Your emblem argument doesn’t solve that either. Emblems measure activity for rewards, not skill for matchmaking. Dead by Daylight previously tied progression to emblem performance through the pip/rank system, and that system was replaced with hidden MMR because those metrics were easy to game and didn’t correlate well with match outcomes.
Your “survival is noisy” point also misunderstands how rating systems work. Individual matches contain randomness, but across large samples that variance averages out. Players whose decisions consistently lead to escapes or kills rise over time, while those whose decisions don’t fall.
And the hatch example doesn’t support your claim either. Hatch escapes do not meaningfully increase survivor MMR, so hiding all match and jumping into hatch doesn’t create the rating inflation you’re describing.
That’s the fundamental issue: your proposal assumes the algorithm can infer causation from telemetry. In reality it can only observe actions. Because those actions are influenced by many interacting variables, the system cannot reliably assign individual credit inside a team interaction. That’s why matchmaking systems measure results over time instead of trying to reverse-engineer credit from individual moments.
-3 -
You keep bringing up this "attribution problem"—the idea that the system cannot reliably link a long chase to generator progress. This is technically false, and the game’s own engine proves it.
- The Engine Already Correlates Concurrent Events
A game engine does not guess; it logs concurrent event states based on timestamps. If Player A's status is In_Chase and Players B, C, and D have a status of Repairing_Generator during those exact same timestamps, the system has absolute, mathematical proof of what is happening. There is no guesswork required. - Dead by Daylight ALREADY Does This Math (Emblems)
You claim the system "only sees the duration of the chase, not the cause" and cannot assign credit.
Have you looked at how the Emblem actually works? If you are actively being chased by the Killer, and your teammates complete a generator during that chase, the game explicitly awards YOU Lightbringer points for distracting the Killer. The developers already wrote the code that links a player's chase directly to the team's generator progress. The system actively tracks the overlapping timestamps, understands that your chase created the safe environment for the generators to pop, and explicitly assigns you credit for it in real-time.
Your entire defense of the outcome-based system rests on the premise that telemetry cannot connect these variables or assign individual credit. But it already does. The engine already perfectly attributes the distraction to the objective. The primitive 1/0 MMR system simply chooses to throw that perfectly accurate telemetry in the trash.
- The Zero-Contribution Math (The Hatch)
You correctly pointed out that a Hatch escape is MMR-neutral (a value of 0). Thank you for bringing that up, because it perfectly illustrates how broken the current math is.
Let’s look at the exact values BHVR assigns:
Player A: Hides in a locker for 15 minutes, 0 gens done, 0 unhooks, 0 seconds in chase. Escapes through Hatch. Result: 0 MMR change. (Or if they sneak out an Exit Gate after the team dies, they get +1 MMR).
Player B: Loops the killer for 3 gens, completes 2 gens, gets facecamped to death to secure the win for the team. Result: -1 MMR penalty.
You are defending an algorithm that mathematically calculates a 0% contribution as a higher value than a 90% contribution. That is not a functional ranking system; that is a broken spreadsheet.
Outcome-based MMR works perfectly in symmetrical games (Chess, CS:GO, LoL) where the team's outcome and the individual's outcome are identical. It fundamentally fails in a 1v4 asymmetrical game where an individual's death is often the exact tactical requirement for the team's victory.
2 - The Engine Already Correlates Concurrent Events
-
1.Logging concurrent timestamps doesn’t prove what you think it proves. It shows that events happened at the same time, not that one event caused the other.
If Player A is in chase while others are repairing generators, the engine can record that overlap, but it cannot prove the chase created the generator progress. The killer may have already committed to that chase, the generators may have been nearly complete, the killer may have misplayed, or the map may have spawned strong tiles that prolonged the chase. The data only shows correlation, not causation.
So no, the engine doesn’t have “absolute mathematical proof” that the chase generated the objective value. It only has proof that the events occurred concurrently, which is a very different thing.
2.The emblem system doesn’t prove the engine can assign real causal credit, it just proves it can apply a simple reward heuristic.
Lightbringer points are awarded when events overlap, but that doesn’t mean the system proved the chase actually caused the generator progress. It’s a convenience rule for distributing bloodpoints, not a validated measure of skill.
3.The example you gave only works if you treat a single match as proof of a ranking system failing, which isn’t how rating systems work.
In one match, sure, someone might contribute a lot and still die while another player escapes with little contribution. But matchmaking systems aren’t designed to judge one match; they measure patterns over many matches. A player who hides every game and contributes nothing will rarely convert matches into real wins and will quickly stabilize at a lower rating. A player who consistently creates pressure, does generators, and helps secure escapes will convert more matches into successful outcomes over time.
Your scenario is just an edge case. Ranking systems aren’t built around edge cases, they’re built around long term signals.
You are defending an algorithm that mathematically calculates a 0% contribution as a higher value than a 90% contribution. That is not a functional ranking system; that is a broken spreadsheet.
That statement only works if you assume the system is trying to measure “percentage contribution” in a single match. It isn’t.
Matchmaking systems aren’t designed to calculate who contributed the most in one game; they’re designed to estimate which players consistently produce winning outcomes over many games. A single match where someone contributes a lot but dies doesn’t invalidate the system, because rankings aren’t based on isolated matches, they’re based on long term results.
So no, the system isn’t calculating “0% > 90%.” It’s simply not trying to score individual contribution in the first place. It’s measuring which players’ decisions consistently lead to successful outcomes over time.
-2 -
Dude, can you write without using chatgpt to write for you?
-1 -
Can you please not get mad at me for writing like ChatGPT? Does the fact that I write a lot automatically mean I’m an AI? Can you stick to the topic?
-1 -
You've moved away from discussing technical analysis and are now engaging in philosophical debate. Let's bring this back to reality in terms of how the game engine, and competitive video games in general, function.
- The "Correlation vs. Causation" Philosophy
You're stating that a survivor who is looping when generators are popping is simply "correlated" with the generators popping, but not "causing" them to pop, since the system cannot directly link the two events. This is not considering the underlying physics of the game: Mutual Exclusivity.
The Killer cannot be two places at once. The loop, by physically engaging the Killer's time and presence (the chase state), is causing them not to be present at the generators. In asymmetrical game balance, creating this physical space and time is "causing" generator progress.
You're stating that perhaps the Killer simply "misplayed" or "overcommitted" during this time. In Player vs. Player gaming, this is exactly "causing" a win: forcing your opponent into an overcommit or taking advantage of their mistake is skill.
You're applying a philosophical standard that no game engine could ever hope to live up to: "causation." The system takes input A: a 3-minute chase state. The system takes output B: 3 generators pop. This is "causation" in game balance.
- The "Edge Case" Myth
You're stating that a player who contributes 90% of a game and sacrifices themselves for a team win is an "edge case," but game systems are not built around "edge cases."
Have you ever seen or played a match of high-level Dead by Daylight? Altruistic sacrifice is not an "edge case" but is, in fact, the very basis of the Endgame Collapse. A protection hit, a gate open through hook state trades, or a sacrificial death on a hook to allow the other three survivors to escape is a standard strategy in almost every competitive match. The idea that the very basic endgame strategy is an "edge case" is a massive disconnect from how Dead by Daylight is actually played.
- Your Ultimate Admission
At the very end of your post, you do, in fact, state exactly what I've been stating since the very beginning of this thread:
"It’s simply not trying to score individual contribution in the first place."
Well, thank you for admitting that. In a symmetrical game mode where all five players on each team have the very same binary Win/Loss result on the winning team screen, it makes sense that an algorithm would not need to heavily weigh individual contribution.
Dead by Daylight is not that game, however. In Dead by Daylight, it is a 1v4 asymmetrical game mode where individual outcomes are not even related to team outcomes whatsoever. One player can die, while three others escape on their team. Because your system is not "scoring individual contribution" (your words), your MMR system is not even remotely related to reality. You are defending your symmetrical ranking system that was lazily implemented into an asymmetrical game mode.
0 - The "Correlation vs. Causation" Philosophy
-
Not sure why you're being downvoted so much for rightfully pointing out that MMR isn't meant to skill based on single game outcomes. The brain rot is incredibly strong around here.
Over the course of thousands of trials, a better player will win more than a less skilled player. The single mythical 5 gen chase game isn't statistically relevant. Hell, you could have 50 of those 5 gen chase games and it still wouldn't move the needle.
3 -
Your response is a textbook example of throwing pseudo-intellectual jargon at the wall hoping something sticks. Let's dismantle the technical fallacies in your argument
You have thrown the word "fallacy" around more than Aristotle did; but for things that aren't fallacies. Labelling things you disagree with as fallacies is not very good faith.
The "Singular Equation" Fallacy
Asking me to provide the "singular equation" with perfect weights for every add-on and map tile on a forum post is a bad-faith argument. It is the equivalent of demanding I write the complete source code for a new search engine to prove that Google can be improved. We don't need access to BHVR's proprietary database to point out that if(escaped) then +MMR is fundamentally inadequate for a multi-variable environment.What? The reason I asked you to create an equation is because despite people telling you it would be increasingly difficult to create a weighted function to accurately (within reason) assess good and bad plays someone makes in a match, you continuously say "no it's not" and then proceed to "fallacy, fallacy, fallacy" their counterclaim without actually addressing any of their valid comments.
For example:
If Dead by Daylight is just a casual party game, why did BHVR spend massive resources to implement strict Skill-Based Matchmaking (SBMM)? They explicitly removed the old casual Rank system to force a hidden, competitive MMR rating onto the entire player base.
You cannot excuse a broken competitive algorithm by claiming the game is "casual." If it's truly a casual game, BHVR should remove the MMR system entirely and let matchmaking be random. But they don't. They insist on using an MMR system, which means as paying customers, we have the right to demand that the system actually does the math correctly.
Did it ever occur to you that BHVR didn't like the fact that every 30 days, everyone's MMR would be "soft reset," so they implemented a more lax but consistent MMR system to combat that? The fact that a rank 1 killer in the previous season who hadn't played in a few days after reset could now be matched with someone of much lower skill who just received rank 8 because they were both in the purple bracket?
BHVR should remove the MMR system entirely and let matchmaking be random.
Now THIS is a bad faith argument if I've ever seen one. Let's throw the baby out with the bathwater and let 1500 win streak blights play against 2 hour Dwight and Meg players fresh out of the tutorial. Everyone here knows that there needs to be some sort of matchmaking rating value applied to accounts. You posit that "if an MMR system exists, it should work 'correctly'" whatever that means being ambiguous. The MMR system in this game is nowhere near perfect, but it was never designed to be. It was designed to minimize new player frustration and to keep wide bands of players in similar brackets to maximize fast matches. You do not have to disagree with that and you are within your right to complain, but saying "it's not right" is only based on your own interpretation of what the "right" system is for this use case; which is clearly not the same for BHVR's.
The NP-Problem Bluff
Calling a weighted matchmaking algorithm a "Non-Deterministic Polynomial problem" is honestly embarrassing. We are not asking the server to solve the Traveling Salesperson Problem or predict protein folding. We are asking it to take existing, already-tracked telemetry data (Chase Time, Gen Progress, Safe Unhooks), apply simple algebraic coefficients, and output a composite score. This is basic linear algebra used by every competitive game on the market, not some unsolvable mathematical anomaly.And as you so expertly ignored, those stats in a vacuum mean absolutely nothing. Do you know why console Mikaela players predrop pallets and run for saves the nanosecond someone gets hooked? They award bloodpoints, which directly feeds in to the emblem system. Whether or not those plays are correct is not factored into giving those points, they are given if the action is performed. You demand a system that ''does the math correctly." I don't think anyone would want a system to award someone who makes generational macro mistakes, thus the system would need to calculate many different factors (exponential factors and their overall weights if they do exist) when determining whether or not points should be awarded. That is how this problem in an NP-Problem. And before you try to claim "nirvana fallacy," I am merely stating the complexity of the problem. If you would like to implement a solution that would genuinely solve nothing but skew the actual skill level rating of players as a whole, go for it.
The Dropped Pallet Strawman (Nirvana Fallacy Again)
You used an extreme edge case—a Mikaela accidentally dropping a pallet that later saves an Ace—as proof that "nuance" is impossible to track.Firstly, this is not the nirvana fallacy nor a strawman. I'm giving you an example of the complexity the system would need to account for. Secondly, I never said this is proof that nuance is impossible to track. I asked who would be credited for it. Who would be given MMR and who would have MMR taken away and for what actions. Is MMR dynamic throughout a match or, for your envisioned system, is MMR calculated at the very end after thousands of weights are calculated? Again, you cry "fallacy, fallacy, fallacy" when all people are trying to do is interrogate your main points to see if they hold water.
Your example with the pallet dropped by Mikael, which saved Ace, is completely off the mark. It's as if you haven't played the game and don't know who does what in such a situation. How does it work now? Do you know? When you drop a pallet during a chase, you get points, YOU GET THEM. If Ace uses the dropped pallet in the chase (because the killer didn't destroy it), he doesn't get points for dropping it, but for the looping time, which is slightly extended - the killer has to destroy the pallet, which gives Ace a moment. It's simple and that's it. You just have to use it. Where do you see the problem?
You didn't read my message carefully enough. I said if Mikaela drops shack by accident completely out of chase. She is awarded no points with the current system. For your response to be true, dropped pallets will now have to award points regardless of their use, which everyone who has played this game knows that would be an objectively wrong way to assign points, let alone an MMR weight. Her mistake resulted in a good play. However, what if she knew a pallet like that would be helpful? For example, throwing a filler pallet directly in front of a hooked survivor so the Billy can't immediately rev back to hook and down the unhooking survivor? That would be a good play, but to the system, it is a similar situation but the intention behind the throws are polar opposites. The Billy example's survivor should be awarded MMR but the Mikaela, in my opinion, should not be.
It's amazing how selective your technical knowledge is. On the one hand, you demand complex anti-camp and anti-tunnel systems, trusting that the server can calculate the Killer's position, chase time, and perk activation down to the centimeter and millisecond. You see your Emblems and points for ‘Pallet Stun’ or ‘Safe Hook Rescue’ on the final screen – proof that the game has full telemetry of every event in the match. But as soon as I suggest that a fraction of this already collected data be used to modify MMR, you suddenly pretend that the server is “blind” and “unable to understand the nuances.” Make up your minds: either the game engine doesn't know what's going on in the match (so let's disable Emblems and Bloodpoints), or it knows perfectly well, but the developers can't be bothered to link these variables to the player matching algorithm.
I have never once demanded complex anti-camp or anti-tunnel systems. I only told you that they are not nuanced systems and explained to you the two factors that determine the meter's progress. I've never once suggested that the server is "blind" nor "unable to understand nuance." I said that the system would be more complicated than you are willing to accept.
And you are STILL ignoring what I said. The system put you in a match with players whose MMR average was most likely within one standard deviation of your own skill level to facilitate a fast queue time. And what's even more incredible is you're forgetting the 80/20 rule - 80% of the effort is done in 20% of the time. If we do a little basic math on that, 4 players with ~550 hours each versus a killer with 180 hours looks a whole lot less terrible than the "900 HOUR KILLER LOSES TO 11,000 HOUR SURVIVOR SWF COMP TOP 0.0000000001% MMR" (obvious hyperbole fallacy /s) title you have created your thread with.
2 -
@UnicornToots is talking about you creating a numbered list for a singular item like LLMs, or using spacing and newlines identical to LLMs, or using em dashes like LLMs, or creating "section headings" as titles for your points like LLMs.
2 -
So what? Can't I just use Microsoft Word or LibreOffice for this? They're powerful, great word processors that make it easier for me to write long texts.
-1 -
You are overcomplicating the entire data processing situation to an absurd degree. Also, you are misusing computer science terms to back your lazy algorithm. Let's discuss the actual situation from a computer science viewpoint without resorting to the term "fallacy."
- The Misuse of Computer Science (The "NP-Problem")
You are claiming that the calculation of player performance based on telemetry data is an "NP-Problem." This is mathematically incorrect first and foremost. An NP (Non-deterministic Polynomial) problem is a problem that has a massive combinatorial explosion. This is like calculating all the perfect moves for a game of chess.
We are asking the server to perform no such calculation. We are asking the server to perform a calculation based on the end of the game. This is a weighted calculation. This is a basic algorithm that has a time complexity of $O(N)$ to perform. Using existing variables such as chase time or gens done, applying a coefficient to these values is basic arithmetic. This is exactly how the game Valorant calculates the player's combat score. This is an NP-Problem. This is a fundamental misunderstanding of the algorithmic complexity.
- Solving your Mikaela/Ace Pallet Riddle
You asked: If Mikaela accidentally drops a pallet out of chase, and Ace picks it up later in the game, who gets the MMR? How does the system determine the intent?
The system does not determine the intent. You are asking the algorithm to read minds. This is completely unnecessary. This is how the performance system calculates the exact scenario that you are posing:
Mikaela accidentally drops a pallet out of chase. She wasted a resource. She gets 0 points and no performance bump. She continues to make these "generational macro mistakes" and loses the match. Her telemetry will be abysmal throughout the match.
Ace goes by the pallet that Mikaela dropped. He loops the Killer around it. He gains Chase Time/Boldness points for navigating map geometry correctly and being on the Killer.
The system does not need to reward Mikaela for her mistake. It just needs to reward Ace for surviving. You are trying to create a problem that event logging solves by default.
- The 80/20 Rule and "Fast Matches"
You used the 80/20 rule (Pareto principle) to argue that a 180-hour Killer was fair against a player with 550 hours each.
The 80/20 rule states that 80% of proficiency is due to 20% of the effort. In a game like DBD with 30+ maps, almost 400 perks, and Killer powers that must be used in a very specific way, a 180-hour Killer has just completed the tutorial. A player with 550 hours each has optimized their Killer's pathing and synergy.
The fact that the system prioritizes "fast queue times" above all else means that you are conceding that Dead by Daylight does not have a competitive Matchmaking Rating. It has a glorified bracket system that prioritizes fast queue times above all else. You cannot argue for the mathematical integrity of a matchmaking rating while simultaneously admitting that the system prioritizes fast queue times above all else.
Created in Word*-1 - The Misuse of Computer Science (The "NP-Problem")
-
Nobody said you can't draft your messages in a word editor. The issue becomes when you are writing paragraphs and paragraphs and paragraphs and paragraphs every hour on the hour, for someone with, ostensibly, a decent amount of computer science knowledge that would imply at least a bachelors degree and a full time job, thus limiting the time you have to dedicate to a forum where you are complaining about a video game. To put it simply, the math ain't mathin'.
2 -
You are overcomplicating the entire data processing situation to an absurd degree.
I am explaining to you the complicated nature of what you are asking. I'm not complicating it, the situation is complicated.
Also, you are misusing computer science terms to back your lazy algorithm.
What computer science term have I misused? I've also not written a single algorithm nor given any suggestions as to an implementation.
Let's discuss the actual situation from a computer science viewpoint without resorting to the term "fallacy."
Let's see what GPT says about this one…
The Misuse of Computer Science (The "NP-Problem")
You are claiming that the calculation of player performance based on telemetry data is an "NP-Problem." This is mathematically incorrect first and foremost. An NP (Non-deterministic Polynomial) problem is a problem that has a massive combinatorial explosion. This is like calculating all the perfect moves for a game of chess.That is exactly what the system would need for an actual meaningful MMR metric that includes nuanced plays in every match.
We are asking the server to perform no such calculation. We are asking the server to perform a calculation based on the end of the game. This is a weighted calculation. This is a basic algorithm that has a time complexity of $O(N)$ to perform. Using existing variables such as chase time or gens done, applying a coefficient to these values is basic arithmetic. This is exactly how the game Valorant calculates the player's combat score. This is an NP-Problem. This is a fundamental misunderstanding of the algorithmic complexity.
So you are indeed making the blunder of calculating stats in a vacuum, something that I and other posters have repeatedly argued is not the optimal way to calculate a weighted MMR metric for an asymmetrical game like DbD. Sticking on a generator for additional points while the rest of your team is either being slugged or on hook should award you zero MMR unless it's the last generator and maybe at least one person has adrenaline. This SIMPLE example demonstrates why taking just weighted averages is enough to game the system so that MMR is ultimately more about min-maxing the system rather than min-maxing your gameplay.
The system does not need to reward Mikaela for her mistake. It just needs to reward Ace for surviving. You are trying to create a problem that event logging solves by default.
You refused to answer my question about the Billy player's situation, where prethrowing a pallet that would award 0 points is indeed the correct and a high-level play that should award something in the MMR system. As I stated, the Mikaela predropping and Ace capitalizing versus someone prethrowing to ensure a safe unhook for both survivors looks identical to the system in regards to the pallet, but the nuance is more complicated than you are willing to concede to the system you wish were implemented.
The 80/20 rule states that 80% of proficiency is due to 20% of the effort. In a game like DBD with 30+ maps, almost 400 perks, and Killer powers that must be used in a very specific way, a 180-hour Killer has just completed the tutorial. A player with 550 hours each has optimized their Killer's pathing and synergy.
Killer is the power role. Based on just raw ratios alone, a killer player can be on par with one survivor player with only 1/4th the hours, since there are 4 survivors per match. That means in this situation, these survivors have the equivalent killer knowledge of a 137 hour killer player. All of this is moot by the way because survivors need to memorize way more in order to be on a similar power level to a killer player. There are 42+ killers that each individually need to be learned; not to mention the addons, perks, maps, etc. For a killer player, they just need to know how to use their power and understand how to loop tiles, with less of an impact on perks that early, since anything under 500 hours is tutorial mode. You are making it sound like a survivor with 500 hours is some 20k hour comp player "maximizing their killer pathing and synergy." THAT is just factually incorrect.
The fact that the system prioritizes "fast queue times" above all else means that you are conceding that Dead by Daylight does not have a competitive Matchmaking Rating. It has a glorified bracket system that prioritizes fast queue times above all else. You cannot argue for the mathematical integrity of a matchmaking rating while simultaneously admitting that the system prioritizes fast queue times above all else.
You are the only one in this entire thread claiming that the MMR system is designed for competition in the sense of winning and losing. I have only ever stated in this thread that its sole purpose is to create competitive matches, meaning players are generally within one standard deviation of their loosely assigned skill level so that genuine new players aren't being matched with people that have been playing since 2016. The only instances in which I am arguing for the alternative are for a system's replacement, which I've already discussed above.
1 -
You are continuing to shift the goalposts and, in the process, introducing some truly fascinating mathematical theories. Let's address them systematically.1. The 1/4th Math (A Logical Breakdown)You stated: "Based on just raw ratios alone, a killer player can be on par with one survivor player with only 1/4th the hours... these survivors have the equivalent killer knowledge of a 137 hour killer player."This is mathematically and logically absurd. Experience is not a shared pizza that gets divided by the number of players in a lobby. If a Survivor has 550 hours of playtime, their brain has processed 550 hours of map layouts, looping mechanics, killer powers, and tile RNG. You do not calculate $\frac{550}{4} = 137.5$ simply because it is a 1v4 game. By that logic, a CS:GO player with 1000 hours actually only has 200 hours of experience because it's a 5v5 game. That is not how human learning or data science works.2. The Billy Riddle Answered (Input vs. Outcome)You claim I dodged your scenario about pre-dropping a pallet against a Hillbilly to secure a safe unhook. I didn't dodge it; you simply don't understand how telemetry scores events.You are obsessing over the micro-input (dropping the wood) rather than the macro-outcome. If Ace pre-drops that pallet, blocks the Billy, and safely unhooks Mikaela, the system does not need to calculate the philosophical intent of the pallet drop. The system awards Ace a massive telemetry spike for the Safe Unhook and Protection Hit/Distraction.Good algorithms do not need to score every micro-action; they score the resulting game state. The pre-drop enabled the safe unhook. The safe unhook generates the MMR value. It is incredibly simple.3. "Stats in a Vacuum" and Dynamic WeightingYou argue that doing a generator while the team is slugged shouldn't award MMR, claiming this proves my system would be gamed.Again, you are ignoring how modern algorithms function. Telemetry is dynamically weighted based on concurrent game states. If the system logs that 2 teammates are in the Dying state, the algorithm dynamically applies a negative multiplier to Objective points and a massive positive multiplier to Altruism/Healing points.Ironically, Dead by Daylight already does this. If you work on a generator while someone is on a hook, your Lightbringer emblem points are actively penalized. The engine already knows how to contextualize stats outside of a vacuum.4. The Standard Deviation ParadoxYou claim the current system's sole purpose is to keep players "within one standard deviation of their loosely assigned skill level."In statistics, you cannot calculate a valid standard deviation ($\sigma$) if your baseline metric for measuring the data is fundamentally flawed. If your metric (1/0 Escape/Die) frequently assigns a negative value to the best player in the match and a positive value to a player hiding in a locker, then your entire dataset is corrupted. You are calculating the standard deviation of garbage data. Grouping players based on corrupted data does not create "competitive matches"; it creates random matches disguised by a hidden number.
So you’re accusing me of using AI to write, and then you turn to ChatGPT yourself and use it as an argument? I’d like to point out that ChatGPT and other AI can generate inaccurate results, even when they cite sources. ChatGPT isn’t really a good example either, because if I’m not mistaken, its knowledge base is current as of 2021 or a little later. And there have been changes in DBD since then.
-3 -
I have this knowledge because I’m studying, so it’s constantly being updated. What I’ve presented here are the basics required at universities. As you may have noticed, I don’t post every hour, but every so often. If I were actually just sitting around posting on the forum, I’d be replying right after someone posted. And how many times has that actually happened? Maybe a few times when I happened to have a moment. But generally, look at the timing—I don’t reply immediately after someone’s comment. To reply LITERALLY right after a response, you’d actually have to use AI because it’s simply impossible to type that fast. I’m currently on leave from work, so I can devote time to studying, gaming, and a little time on the forum. But I don’t understand how that matters here? Does having a moment to spare immediately invalidate my opinion? If not, then please respond to the arguments and not my private life. And I write multiple paragraphs for two reasons: for your convenience—it’s probably easier to read a well-structured paragraph in Word than a wall of text, right? And the second reason is that, sure, I could write shorter posts, but given what’s happening online, people wouldn’t understand me. They wouldn’t understand the mental shortcuts. An example? I asked a question in a certain Facebook group about DBD. I wrote briefly, concisely, and to the point. The result? People completely misunderstood what I was getting at and interpreted the topic literally the opposite of what I wrote and what I meant.
So... is it such a big deal that I reply after, say, two hours instead of 30 minutes, and is it such a big deal that I write a lot to make sure everyone—or at least most people—understands me clearly?-3 -
So you’re accusing me of using AI to write, and then you turn to ChatGPT yourself and use it as an argument?
?????????????? I understand that communication over text can be misconstrued, but if you honestly thought that I was turning to Chat GPT to fact check you versus implying your quoted text following that line was written by AI, I'm sorry that's just way off base.
That is not how human learning or data science works.2. The Billy Riddle Answered (Input vs. Outcome)You claim I dodged your scenario about pre-dropping a pallet against a Hillbilly to secure a safe unhook. I didn't dodge it; you simply don't understand how telemetry scores events
Listen buddy, I know you love the word telemetry, and I'm not going to explain the situation again and how it's not what you're thinking, just for you to refuse to acknowledge it, so I'm going to save myself the time. Please reread my previous messages more carefully.
The system awards Ace a massive telemetry spike for the Safe Unhook and Protection Hit/Distraction.Good algorithms do not need to score every micro-action; they score the resulting game state. The pre-drop enabled the safe unhook. The safe unhook generates the MMR value.
And ultimately the bad players will now prioritize unhooking safely to boost their MMR, when a correct play would possibly letting a camped player stay on hook for 50 seconds while 3 gens are done separately.
"within one standard deviation of their loosely assigned skill level."In statistics, you cannot calculate a valid standard deviation ($\sigma$) if your baseline metric for measuring the data is fundamentally flawed.
This is clearly wrong, since BHVR knows what "high MMR" is based on their quarterly stat drops; implying that there is in fact a bell curve to which a standard deviation would apply. Sorry you are incorrect.
They wouldn’t understand the mental shortcuts. An example? I asked a question in a certain Facebook group about DBD. I wrote briefly, concisely, and to the point. The result? People completely misunderstood what I was getting at and interpreted the topic literally the opposite of what I wrote and what I meant.
There is a fine line between underexplaining and overexplaining. I don't know the particulars of your interactions, so I'm going to not comment on why things happened the way they did. Also, why are you asking about DbD on a facebook group? lol
Ultimately nobody minds if there is back and forth on a topic or even disagreeing. What people don't like is when they are condescended, patronized, downplayed, and ultimately ignored by someone who thinks a little bit of self study is going to make them a subject expert.
3 -
Goodness. you didn’t have to turn your reply into a solid wall of text just because someone accused you of using an LLM.
2 -
And yet someone had mentioned LLM earlier, and someone had told me not to use ChatGPT for writing 😀 That's why I explained it.
1 -
- The AI Comment Clarification
Ah, I see. You weren’t asking an AI for the answer; you were just making another sarcastic ad hominem about my writing style. Understood. Let’s move past the personal jabs and get back to the actual game data.
- The "Bell Curve" Fallacy (Standard Deviation)
You stated: "BHVR knows what 'high MMR' is based on their quarterly stat drops; implying that there is in fact a bell curve to which a standard deviation would apply. Sorry you are incorrect."
You are mistaken in assuming that the data within the curve is correct.
If I ranked 1,000 players by flipping a coin for each of them every match, I can easily tell you that the data within the curve would perfectly represent a bell curve (Normal Distribution). I can easily tell you the top 5% of that curve and call it "High Coin-Flip MMR." However, the fact that this data follows a bell curve does not mean that the data itself is correct.
The fact that BHVR can plot their 1/0 Escapes/Kills onto a graph and find a median does not mean that the data within the curve is correct. A bell curve of corrupted data is still corrupted data.
- The "Unhook Farming" Scenario
You stated that rewarding Safe Unhooks would cause bad players to farm hooked survivors instead of doing generators while the Killer camps.
You're completely disregarding the way in which modern telemetry handles such concepts as Proximity and Negative Constraints.
If the Killer is 'Face Camping' the survivor, a proper performance system (and indeed the current system in DbD itself, the Emblem system), will take this into consideration. An unhook that immediately results in the death of the survivor will be recorded as an Unsafe Unhook, resulting in a significant penalty against the Killer attempting the unhook. The system will also take into consideration the Killer's Proximity to the Hook in the game world. The dynamic MMR system will reward players for pushing generators while the Killer is locked in place at the Hook. You're essentially arguing against the basic system that I'm not advocating for in the first place.
- The Retreat
"I'm not going to explain the situation again... so I'm going to save myself the time."
I'll admit that you're right regarding the Billy/Pallet example. The moment the actual micro-mechanics of the system come up in the form of how the telemetry system itself logs events, you immediately back down.
- On Being "Condescending"
I'm not here to make you feel condescending towards you or any other individual reading this article. I'm here to have a debate regarding the architecture of the matchmaking system in Dead by Daylight. I write in such a detailed manner because that's the only way in which I can effectively discuss the data science that goes into the system. If you take the mathematical inconsistencies in the algorithm as condescending towards you personally, then I apologize for that. But I'll certainly not dumb down the data science in order to make the current 1/0 MMR system look more favorable than it really is.
-1 - The AI Comment Clarification
-
- Coin flips are random → Escapes/Kills is not random
- Measuring every variable is unreliable
- Measuring the outcome avoids that problem
- Its really that simple. Your really just wasting your time typing in circles around those 2 simple problems with your system. You still have yet to address it. No dev on the team will read what your proposing and take it serious, its not logical.
0 -
Ah, I see. You weren’t asking an AI for the answer; you were just making another sarcastic ad hominem about my writing style. Understood. Let’s move past the personal jabs and get back to the actual game data.
This reads identically to a Chat GPT response after reprompting it for a different answer.
You are mistaken in assuming that the data within the curve is correct.
What is "correct" data? Data is neither incorrect or correct (faulty measurements notwithstanding), data is just data. The application of data is what matters. The data may not be getting "correctly" applied to YOUR liking, but there will always be a bell curve. You just don't like the bell curve it represents.
You stated that rewarding Safe Unhooks would cause bad players to farm hooked survivors instead of doing generators while the Killer camps.
You're completely disregarding the way in which modern telemetry handles such concepts as Proximity and Negative Constraints.
If the Killer is 'Face Camping' the survivor, a proper performance system (and indeed the current system in DbD itself, the Emblem system), will take this into consideration. An unhook that immediately results in the death of the survivor will be recorded as an Unsafe Unhook, resulting in a significant penalty against the Killer attempting the unhook. The system will also take into consideration the Killer's Proximity to the Hook in the game world. The dynamic MMR system will reward players for pushing generators while the Killer is locked in place at the Hook. You're essentially arguing against the basic system that I'm not advocating for in the first place.
I'll admit that you're right regarding the Billy/Pallet example. The moment the actual micro-mechanics of the system come up in the form of how the telemetry system itself logs events, you immediately back down.
You realize the system that you are asking for was the previous rank system right? It got removed for reasons I already stated. Each emblem contributed to your pips, and if you had enough pips you would rank up. If you didn't get a safety pip, you would rank down. This system did not work to the dev's liking, so they changed it. You don't have to like the new system, but to pretend that the system you are asking for wasn't already in the game for 5+ years is just factually incorrect.
I'm not here to make you feel condescending towards you or any other individual reading this article. I'm here to have a debate regarding the architecture of the matchmaking system in Dead by Daylight. I write in such a detailed manner because that's the only way in which I can effectively discuss the data science that goes into the system. If you take the mathematical inconsistencies in the algorithm as condescending towards you personally, then I apologize for that. But I'll certainly not dumb down the data science in order to make the current 1/0 MMR system look more favorable than it really is.
Again, this reads exactly like Chat GPT apologizing to you for not being able to tell you that 2+2=5. If this is one giant Turing Test, I think now would be the time to call this thread (your only thread on your forum account by the way) a failure of the Turing Test.
1 -
Using bullet points and numbers to structure your points into boxed units does look AI-ish. I'm not saying it is, but that's how it appears. If you don't want that accusation, don't do that.
1 -
You have managed to boil down a complicated system architecture discussion into three simplistic bullet points because you have no other way to argue the actual mechanics of the game. Let’s address your last arguments:
- "Escapes/Kills are not random"
While in a pure 1v1 game such as Chess or StarCraft, game outcomes are indeed not random. Dead by Daylight is a 1v4 game. In a Solo Queue game, your own escape is heavily influenced by RNG and factors outside your control. You have no control over whether a teammate will disconnect at 5 gens, whether a teammate will kill themselves on their first hook, or whether the Killer wants to give someone a Hatch. Judging a player on the unpredictable behavior of three random strangers is statistical noise. - "Measuring every variable is unreliable"
If you believe this is true, then why does the game engine calculate Chase Time, Generator Repairs, and Safe Unhooks with perfect accuracy in order to award Bloodpoints and Emblems in every game? The game has already created a reliable system. Your argument is against how the game is currently designed. Your argument is flawed because you are arguing against how the game currently works. - "Measuring the outcome avoids that problem"
No, it ignores reality and is a fatal flaw in your argument. Your simplistic system mathematically awards a player who hides in a locker for 15 minutes and escapes via Hatch (+MMR), while punishing a game MVP who chased the Killer for 3 gens and got hooked (-MMR).
I have answered these arguments many times before, yet you still do not want to accept the fact that you are dealing with a symmetrical game design issue. Your last resort is clearly "the devs won't even bother reading this," so I am done with this conversation.
If it doesn't make sense, you don't have to write anything. And as you yourself pointed out—IT'S JUST A SUGGESTION. I don't expect anyone to take this specific suggestion and implement it right away. It’s just an outline, a rough draft. I’ve outlined the problems, I’ve outlined what needs to be improved, I’ve outlined what we have in terms of data, I’ve outlined what it COULD look like, but that doesn’t mean it has to match my suggestion exactly, because it doesn’t have to. And the suggestion itself? It seems logical, since you’ve decided to discuss it.-1 - "Escapes/Kills are not random"
-
Just make up your minds already: do you think I’m writing with AI or not? First you say I’m using ChatGPT or some other AI, but when I explained that it’s just Word’s formatting, suddenly “no one accuses me of using AI” anymore. Please make up your minds. I’ve already explained why I format this way (using Word) and why my replies are so long—read it if you want.
If your last resort in defending your arguments against mathematical facts is to accuse me of being an AI or using AI—because the text formatting annoys you—then we’ve officially reached the end of this technical discussion. What? Did ChatGPT write that too? “Yes, I confirm it, I was that ChatGPT. And yes, 2+2=5." 😁
You state: "Data is neither incorrect or correct... data is just data."
In the field of computer science, there is a fundamental concept known as GIGO, or "Garbage In, Garbage Out." If you use a defective thermometer to measure a room, the "data" exists, but it is invalid for the purpose for which it is being used. The 1/0 escape metric is a defective thermometer. It tries to measure a player's individual mechanical skill with a team-based binary outcome. The "data" collected is invalid because it is corrupted with variables outside the player's control. A bell curve constructed from invalid "data" is useless for competitive matchmaking.By admitting that the system I am referencing "was already in the game for 5+ years," you've essentially destroyed your entire case.
You are confusing a static progression ladder with a dynamic mathematical matchmaking rating system, such as Elo or MMR. The static system was flawed because it used hard-coded "brackets" and forced unnatural gameplay on players to obtain Iridescent badges. I am not asking for the return of Ranks 1 through 20. I am asking you to recognize that the games own data, its ability to correlate multiple variables and weigh them based on context, already exists in the engine BHVR chose to ignore all that rich data and take the lowest math possible: win/loss (1/0) when implementing their hidden SBMM.
If your next response is going to be another digging into how I formatted this and whether I am human, I will concede your entire argument on the 1/0 system.
And this is my only thread on this forum, because only now have I had the time, the motivation, and the boredom of playing against people who are clearly better than me. Don’t you think a bot or AI would be more likely to spam posts and comments? By the way, if this is my first post, so what? Can’t I post if I haven’t written anything since the forum started? How am I supposed to start posting on the forum if I could never write a single post because I haven’t written anything before? It’s a vicious circle, illogical, and physically impossible. Unless you have a time machine. But I would really ask you to stay on topic, because arguments like “I'm writing this for you,” “you don't have any other posts,” have absolutely no connection to this specific thread and aren't relevant to the topic. So, since “the big forum boss is letting me post,” even though this is my first and only post on this forum, should I just pack up and run?
-1 -
So I can’t use Word after all? You’d rather have a plain, huge block of text? And you won’t accuse me later of “going to all that trouble for nothing—no one’s going to read a block of text like that”? OK, so starting with this comment, I promise to post a blank block of text—and please, no complaints.
Sorry for the downvote—it was a misclick.-1 -
No one said it needed to be poorly presented. Indents and punctuation have made for ease of reading since time immemorial. Breaking things into segments with bullets and numbers isn't the typical, average writing format. It looks like AI. In a time when everyone is slowly forgetting how to write for themselves (you can see both AI slop and people of substandard writing ability in the forum at any given moment), if you don't want to be accused of it, then don't do.
1 -
Reading through the thread I don't particularly care about the AI issue, though saying Microsoft Word is a really strange defense because since when does Word force bulletpoints and large spacing into posts, but its when you say you want a technical discussion and then make posts like this:
By admitting that the system I am referencing "was already in the game for 5+ years," you've essentially destroyed your entire case.You are confusing a static progression ladder with a dynamic mathematical matchmaking rating system, such as Elo or MMR. The static system was flawed because it used hard-coded "brackets" and forced unnatural gameplay on players to obtain Iridescent badges. I am not asking for the return of Ranks 1 through 20.The first sentence acts like a gotcha on @Valuetown, but then two sentences later you admit it was flawed and then say you don't want that system.
You're trying to strawman the argument by making it seem like people are arguing that these things could not be coded, when what people have been pointing out is that they can not be realistically coded in a way that could more accurately determine skill than the current system. If what you want is a discussion instead of doing a gotcha, if you think that system was bad you should have agreed it was, then moved onto how it could have been better.
I am asking you to recognize that the games own data,itsability to correlate multiple variables and weigh them based on context, already exists in the engine BHVR chose to ignore all that rich data and take the lowest math possible: win/loss (1/0) when implementing their hidden SBMM.And people have been asking how do you practically do that in a meaningful way given all of the nuance and subjectivity in decision making within the game? If you have a better system, great, let's run through it, but so far you seem to want to imagine there could be a better system and don't like that people are pointing out the realistic flaws that such a system encounters.
5 -
You state: "Data is neither incorrect or correct... data is just data."
In the field of computer science, there is a fundamental concept known as GIGO, or "Garbage In, Garbage Out." If you use a defective thermometer to measure a room, the "data" exists, but it is invalid for the purpose for which it is being used. The 1/0 escape metric is a defective thermometer. It tries to measure a player's individual mechanical skill with a team-based binary outcome. The "data" collected is invalid because it is corrupted with variables outside the player's control. A bell curve constructed from invalid "data" is useless for competitive matchmaking.
What is "correct" data? Data is neither incorrect or correct (faulty measurements notwithstanding), data is just data.
So I exclude the very instance you are talking about, yet you bring it up anyway. There is no faulty data when determining if someone escaped or not. There is no "faulty thermometer" in the game's code. YOUR claim is that the data gathered does not have a high enough resolution. Those are two different things. One deals with genuine false data. One is dealing with the ambiguity. Your misunderstanding of the two is how you arrive at this conclusion.
The 1/0 escape metric is a defective thermometer. It tries to measure a player's individual mechanical skill with a team-based binary outcome. The "data" collected is invalid because it is corrupted with variables outside the player's control.
It is not a defective thermometer. You are asking a weather rock to be a calibrated, industry grade thermometer. Neither is defective. They read different resolutions of the same data.
By admitting that the system I am referencing "was already in the game for 5+ years," you've essentially destroyed your entire case.
You are confusing a static progression ladder with a dynamic mathematical matchmaking rating system, such as Elo or MMR. The static system was flawed because it used hard-coded "brackets" and forced unnatural gameplay on players to obtain Iridescent badges. I am not asking for the return of Ranks 1 through 20. I am asking you to recognize that the games own data, its ability to correlate multiple variables and weigh them based on context, already exists in the engine BHVR chose to ignore all that rich data and take the lowest math possible: win/loss (1/0) when implementing their hidden SBMM.
You are asking to apply the same (drum roll please, your favorite word is coming!) TELEMETRY data in the same exact manner as the old rank system did, which is exactly the same as the grade system for awarding pips. In the old system, a player could have died to the killer, but ranked up due to their contribution in the match. In the current system, that is no longer the case. This system was removed for a variety of reasons, all of which are only known by the dev team.
And this is my only thread on this forum, because only now have I had the time, the motivation, and the boredom of playing against people who are clearly better than me. Don’t you think a bot or AI would be more likely to spam posts and comments? By the way, if this is my first post, so what? Can’t I post if I haven’t written anything since the forum started? How am I supposed to start posting on the forum if I could never write a single post because I haven’t written anything before? It’s a vicious circle, illogical, and physically impossible. Unless you have a time machine. But I would really ask you to stay on topic, because arguments like “I'm writing this for you,” “you don't have any other posts,” have absolutely no connection to this specific thread and aren't relevant to the topic. So, since “the big forum boss is letting me post,” even though this is my first and only post on this forum, should I just pack up and run?
I genuinely don't know how you meander through that thought process while also having the hubris (it's more than just a dbd perk name) to simultaneously say this in the thread:
They wouldn’t understand the mental shortcuts. An example? I asked a question in a certain Facebook group about DBD. I wrote briefly, concisely, and to the point. The result? People completely misunderstood what I was getting at and interpreted the topic literally the opposite of what I wrote and what I meant.
How is it that you are not addressing the main point of that comment (the fact that the majority of your posts read as AI generated in nature), but rather addressing a supplemental point (which doesn't clear the idea that your posts may be AI generated, it is simply a potential point of evidence) in a way that completely misses the point?
2 -
- That didn't prove escapes/kills are random. There still not.
2. Thats not measuring every variable. The emblem system only tracks activities, not the why. The why is the variables.
3. A player who hides and goes for hatch every game will lose MMR overtime. So no there not rewarded.
2 -
Your example of “weather rock” is a great illustration of what I mean.
A weather rock can tell you if it is raining. It cannot tell you if it is cold outside. If you use a weather rock to measure the temperature of a room, you are not getting low-resolution data; you are using the wrong tool to do so.
The 1/0 system measures a team-based outcome (did your team get an escape for you?). It does not measure an individual mechanical skill set. Using a survival "weather rock" to measure an individual skill set in an asymmetrical game is not a problem of resolution; it is a problem of architecture. Furthermore, BHVR has access to the thermometer (i.e., the engine's detailed telemetry), but are actively choosing to use the weather rock anyway. You state that the original system was changed due to "a variety of reasons, all of which are only known by the dev team." This is a textbook fallacy known as Appeal to Authority. We are in a feedback forum discussing the mathematics involved in matchmaking. Shaking your shoulders and saying, "Well, we have no idea why; it is a dev team decision," is an admission that you cannot mathematically defend this system to begin with. Again, you're mixing up a Monthly Progress Ladder, where you go from rank 20 to rank 1, with a matchmaking system, such as SBMM or Elo. These are two completely different animals. My original point still stands, which you've already admitted yourself: the system is fully capable of tracking all that data. You're complaining I missed your "main point," which is that my posts "read as AI generated." I've already dealt with this issue multiple times. I use Microsoft Word to type up my posts because I'm a bit of a neat freak and I like to organize my thoughts using proper syntax, paragraphs, and bullet points, so people can actually read what I'm saying. I even made a joke about it last time I responded. If you're shocked that your posts don't look like a jumbled mess of words, with proper grammar, syntax, and a logical flow, then perhaps you're not used to reading posts written by humans. If you'd rather discuss why this game uses a weather rock instead of a thermometer, I'm happy to discuss that with you. If you're still upset about how I'm writing, then perhaps you shouldn't bother writing at all. What's your purpose here, then? Is it because you're intimidated by Word formatting that we're done here?0 -
Microsoft Word doesn't enforce line breaks and wide spacing in posts. Who told you that? Did I write something like that? No? Then please don't accuse me of something I didn't write, and please read carefully. I set up that formatting in Word for YOUR convenience, but no, that's not it. Another thing is that you’re yet another person who sees the discussion but doesn’t read it carefully, yet really wants to chime in. This is JUST A SUGGESTION. I listed what the system collects, what YOU CAN do, BUT DON’T HAVE TO do. I gave my suggestion. And what did I get? People nitpicking the text formatting, nitpicking the spacing, nitpicking when I respond, and nitpicking that this is my first and only post. Well, a very mature, substantive, and technical discussion...
0 -
OK, this is going to be a long block of text in the wrong format. Happy now?
0 -
@Valuetown @top500spiderman Let's try this another way... It seems to me that this whole recent discussion is pointless, because no matter what I do, it's always wrong—it'll always be “ChatGPT!!!”, “You're giving me downvotes!”, “You're typing/responding too fast!”. So let's try a different approach. What do you think BHVR should do about this? Raise the soft cap? Do nothing? Is it bad? Is it good? What should I do about it? Should I suffer through a pathetic 10-minute match, being treated like a punching bag by a team of noobs and tryhards? Should I DC as soon as I figure out what kind of team this is?Go to a corner of the map, stand facing the wall, and go AFK until they finish the generators and leave? Or just stop playing? (After all, the “golden advice” is “if you don’t like it, don’t play and don’t criticize”—but what about freedom of speech?). I'm looking forward to your suggestions and solutions. And honestly, after these discussions, I'm hoping for something wise, serious, logical, and sensible—not like that “golden advice.”
0 -
My original point still stands, which you've already admitted yourself: the system is fully capable of tracking all that data
It doesn't, The system only tracks activities not the variables of why those activities happen. Your point crumbled a long time ago. But your just fighting with pride and refusing to acknowledge the large flaws that were pointed out in what you’re proposing. Instead it looks like you’re trying to force whatever LLM you’re using to smooth over those flaws, but it can’t.
IT'S JUST A SUGGESTION. I don't expect anyone to take this specific suggestion and implement it right away
You saying this tells me you understand the system you propose may have flaws. So why are you defending the flaws that have been pointed out about it? is it because it crumbles your entire system? Its ok if it does. Very accomplished scientist write papers all the time that get trashed, criticized and rejected all the time. Dont let your pride make you try to turn a dead end into a highway.
0 -
I’m not defending those flaws (the ones you’re probably referring to). It’s not about pride or anything like that; it’s about logic and the realization that it’s impossible to create a 100% perfect system—I can admit that, and I understand it. What I am defending, however, are the points I defended a long time ago. You’re trying hard to make CERTAIN aspects of my proposal seem difficult or impossible, as if it couldn’t be done and as if it were some kind of magic. But it’s not magic—it’s math, algorithms, programming, and functions. I’ve given examples of how this is already happening—there are plays, goals in a game, and they’re being tracked. You insist that they can’t be transferred to the matchmaking system. Why? Because you misunderstood my entire suggestion and now you’re acting like the winner. I wrote earlier in the comments about emblems. It was said that BHVR has already BASED matchmaking on emblems at some point. Yes, that’s true, that system was also flawed. And that’s where the misunderstanding comes in. I’m still not talking about transferring 1:1 what’s collected for emblems. That was just an example to counter the argument “data isn’t collected during the match.” Data is collected during the match because there are emblems, there are blood points, and there are conditions for activating perks. Now, we need to figure out how to feed some of that data into 2 functions—or maybe even 3—related to MMR. This does not equate to a return to the old emblem-based system. And ever since this misunderstanding, there have been, let’s call them “attacks” on me regarding AI and response speed. Just as you yourself wrote—as if pride prevents you from grasping that you don’t understand. And ever since that misunderstanding, there have been—let’s call them “attacks”—on me regarding AI and response times. Just as you yourself wrote—as if pride prevents you from realizing that you didn’t understand, as if pride is forcing you to try to turn a dead end into a highway. All right, let's move on. Take a look at my comment above, read it carefully, and show me what you've got.
0
