Forums / Games / Halo 5: Guardians

H5Beta vs H3Beta: Caboose wrong, CHX right

OP NNMS MXMS

  1. 1
  2. ...
  3. 2
  4. 3
  5. ...
  6. 7
I am preserving the original post in quotes below. That way, my incorrect analysis will preserved for posterity.

I based the analysis on 56 seconds being too short for a match length. If the actual H5 match length were a typical slayer match length, the analysis below would be correct. However, I no longer believe that to be the case. I don't need a reply from Josh Holmes to come to this conclusion, so I won't wait for that.

For me to be correct, the average match length for the H5 beta needs to be in the ~7min mark. Let's see if that is possible by looking at the other numbers 343i has given us:

Total kills: 181M
Total players: 1M

With 181M kills and 1M unique beta players, this means the average player achieved 181 kills. Let's just say that the average player achieved 10 kills per match. Yes, that's low . . . but it actually gets worse for the H5 beta as the kill count per match increases.

So . . . 181 kills at a 10-kill-per-match average means that the average beta player participated in 18 matches. For me to be correct, those 18 matches must each have lasted for a typical slayer match duration - or about 7 minutes. This means we can get the average per-player playtime as follows:

  • (average # matches per player) * (average match length) * (unique players) = cumulative total playtime
  • 18 * 7 min * 1M = 2.1M hours at only 2.1 hours played per person
This checks very well with the 2.5M playtime hours listed, but does not check well at all with the 20M games played. Were the games to have been typical slayer matches, there should have been ~1.4 billion kills . . . not 181M. What this means is that each ROUND of Breakout was recorded as a completed MATCH . . . and that the actual average match length was, indeed, 56 seconds. Since I can reach CHX's conclusion by looking at a different set of numbers and CHX's conclusion is consistent with all of the stats on the infographic, he is correct. My number can be achieved by looking at any 2 of the 3 numbers (kills, hours, matches), but cannot be made consistent with all three simultaneously.

In other words, CHX H8 ME was correct. In comparing the beta playtimes, the average H3 beta player logged 14.6 hours, while the average H5 beta player logged 2.5 hours. Read into that what you will.

I have posted a reply in his thread.

Quote:
So in this popular thread, the OP links to an image describing the per player differences in total hours played between the H3 and H5 betas. The OP kindly also provided the links to his source information in the image (for which I thank him). Unfortunately, based on the source information provided, the OP's comparison is incorrect.

Per the H5 Beta data source, there were 20M matches played for a total of 2.5M gameplay hours. The OP took this number to mean hours played by player. That assumption is wrong. If there were 20M matches and the 2.5M hours meant by player, then we can calculate the average match duration:

{2.5M hrs / (8 players per match * 20M matches)} * (3600 seconds / hr) = 56 seconds per match (average)

The average match length for the H5 beta was certainly not only 56 seconds. So we know now that the H5 data cannot possibly mean hours by player. If, however, we assume the 2.5M hours is not by player, but rather by match, we get a much more sensible answer:

{(2.5M hrs) / (20M matches)} * (60 min / hr) = 7.5 minutes per match (average)

So it is quite clear that the H5 Beta data is based on hours by match - not hours by player.

Looking at his provided link for the H3 Beta data source, we find Major Nelson giving 820K unique players and 12M hours of total playtime. Major Nelson further goes on to say that this is equivalent to a single player logging 1,400 continuous years of playtime. So does Major Nelson mean 12M hours by match (like H5) or 12M hours by player? We can easily find out. Let's find out what 1,400 continuous years of playtime comes out to in terms of hours by player:

1,400 yrs * (365 days / year) * (24 hrs / day) = 12.26M hours

Major Nelson does, indeed, mean hours by player. This is also consistent with his statistic for downloaded data, which was a total cumulative by player statistic.

Thus, in order to compare the H5 and H3 betas, we must either convert the H3 numbers into hours by match or convert the H5 numbers into hours by player. Let's do that and see where we come out:

Using H5 "hours by match" statistics:

H5 Guardians: 1M players with 2.5M total match hours
H3 Beta: 0.82M players with 1.5M total match hours

- or -

Using H3 "hours by player" statistics:

H5 Guardians: 1M players with an average of 20.0 hrs played per player
H3 Beta: 0.82M players with an average of 14.6 hrs played per player

In other words, the two betas had comparable statistics for matches and total hours played on a per-player basis, with the H5 beta actually beating H3 in both categories. Wild theorizing about why the H5 beta had so much less replay value need no longer apply.

One question remains: Will the OP of the other thread generate and post a corrected infographic?
Can you please explain the first calculation? I'm not saying you're wrong. Just that I'm not that good with statistics.
It's sad the community has come to providing this type of factual evidence to compare betas. While it's interesting and does correct the previous image that was presented, I'm in no way saying that what you're doing is sad, though I said prior. I meant in the communities entirety based on how divided it seems to be. Now the first argument is going to be that the Halo 3 Beta had ~200k less users and that this isn't a fair comparison, blah blah blah.

Both versions of the Halo series old and renewed are very fun given the mechanics they are using. I was apart of both betas and put in my good share of hours for each. While I enjoyed Halo 3 and that's my Halo of choice among all, the Halo 5 beta was very promising to how I enjoy playing Halo. I won't let a mechanic such as sprint alter my feelings toward Halo. I basically assumed that this evolution of the game was apparent at this point given the last 2 Halos had sprint.

I'm just here to say one thing. I can't wait to get my hands on Halo 5: Guardians. Not just for the multiplayer, but for campaign, and for everything that 343i hasn't announced that will be included in the game. I'm hoping for an amazing forge, theater mode (we already saw spectating so let's hope theater mode is present), fileshare for custom maps and gametypes, maybe a voting system that functions with the fileshare to have 1 of the top 3/5/10 maps and gametypes of the week to be in the public playlist. Again I'm just hoping there is a lot of 'new' features to the customization in Halo for the community to be apart of.

Have a good day everyone.
@Behold Death

If the 2.5M hours are by player (so that a match of 10 minutes with 8 players accumulates 10 * 8 = 80 minutes of total playtime) then you have to divide that 2.5M hours by the number of players in the match to get the match length.

If the 2.5M hours are by match (so that a match of 10 minutes accumulates 10 minutes of total playtime regardless of the number of players in the match), then you do not include the number of players in the divisor.

It's sad the community has come to providing this type of factual evidence to compare betas.
Like everywhere, it's just confirmation bias. Not sad . . . just human nature.
NNMS MXMS wrote:
@Behold Death

If the 2.5M hours are by player (so that a match of 10 minutes with 8 players accumulates 10 * 8 = 80 minutes of total playtime) then you have to divide that 2.5M hours by the number of players in the match to get the match length.

If the 2.5M hours are by match (so that a match of 10 minutes accumulates 10 minutes of total playtime regardless of the number of players in the match), then you do not include the number of players in the divisor.

I see. Thanks for clarifying! :)
Brilliant job OP... I've just double checked your math and came to the same conclusion!

(8 * x) * 20.000.000 = 2.500.000

8 players per match * unkown average time per match * 20.000.000 matches = 2.500.000 hours played by player

If you actually think about it... that previous numbers were crazy to begin with!

Thank you for thinking things through for yourself!
MATH

MATH IS ON OUR SIDE
MATH.
We have MATH on our side!
Double post, my bad
MATH.
We have MATH on our side!
There really shouldn't be "sides". We should all try to provide feedback to make the game enjoyable for anyone who wishes to play. Perpetuating this artificial split in the community doesn't do anyone any good; it just makes all of our feedback equally ignorable because no one at 343i will want to sift through all of the invective to find the occasional nugget of reason.

And it may surprise you to know that I do not support sprint. I did not support personal ordnance. I did not support the perks in H4, nor did I support the way loadouts were done. I still had fun playing H4 . . . but all those added things that I do not support made CE so much more enjoyable.

I made the above post because I also do not support making arguments and spinning theories where the only basis is faulty information.
A very insightful thread, and I'm glad to see that your math adds up. Thank you for posting this! It's nice to see that many other players also loved the Halo 5 Beta, it spells out good news for the future of the installment.
All of this is done under the assumption that all players A) continued to play all 3 weeks and B) stayed in the match the entire match. I'm not sure about you, but this simply wasn't my experience with the H5 beta.
NNMS MXMS wrote:
MATH.
We have MATH on our side!

There really shouldn't be "sides". We should all try to provide feedback to make the game enjoyable for anyone who wishes to play. Perpetuating this artificial split in the community doesn't do anyone any good; it just makes all of our feedback equally ignorable because no one at 343i will want to sift through all of the invective to find the occasional nugget of reason.

And it may surprise you to know that I do not support sprint. I did not support personal ordnance. I did not support the perks in H4, nor did I support the way loadouts were done. I still had fun playing H4 . . . but all those added things that I do not support made CE so much more enjoyable.

I made the above post because I also do not support making arguments and spinning theories where the only basis is faulty information.
Wow. Cool. It's nice to see partisan bickering on the forum give way to actual mathematics and research in a way devoid of agenda or forum politics, but rather, in the pursuit of knowledge and clarity. This is good to know. Slightly out of character, but nice job, Caboose.
That's cool.

H5 is still going to rocket down the xbl charts upon release though.
This is a brilliant post, and needs to comments to push it to the top of the forum. More people need to read things like this and take them as it is. Regardless of how each individual liked the beta, it still had many, many players, and showed great promise. And for 343 to come out after the beta and address the many concerns the community has only shows how far they have come as a "new" developer. Many in the forums would not have been happy with anything less than a complete removal of sprint and reducing of size in each map, but for myself at least, I am completely satisfied with the response from 343. I only have a few words for 343. Keep doing what you truly think is right, dont listen to everyone else. If you stick true to the idea that you have of Halo, as fans and caretakers of Halo, I have no doubt it will be a game for this community to rally around.
Blinky 909 wrote:
All of this is done under the assumption that all players A) continued to play all 3 weeks and B) stayed in the match the entire match. I'm not sure about you, but this simply wasn't my experience with the H5 beta.


Assumption (A) is not required. Whether someone played a single match (or even part of a match) or played hundreds of matches throughout the duration of the beta affects nothing for either H3 or H5. The beta statistics are aggregate numbers and are independent of when any particular player played.

Assumption (B) can be tested. In Major Nelson's case, whether a player stayed the entire match or quit is irrelevant. Major Nelson simply logged the total hours played. So no adjustment would be needed to the H3 per-player-hours figure (the match-hour number would need to be corrected, but I'll stick with per-player-hours for this). In H5, it is relevant, since the statistic given was match hours, and we estimated per-player-hours assuming 8 players per match.

If what we want to see is how often players must quit for the H5 beta hours per player to equal - rather than exceed - the H3 beta hours, we must solve the following:

  • (# H5 finishers) / (starting players per match) = (H3 hours per player) / (H5 hours per player)
  • (X) / (8) = (14.6) / (20)
  • X = 5.856
So unless the average number of players that finished an H5 match was less than 5.856 players, the per-player time spent on the H5 beta still exceeds that of the H3 beta. Additionally, for simplicity, I assumed that the H5 players quit instantaneously after the match started. If I instead used a more realistic assumption - that the average time-to-quit was around 1 minute - you get the following:

  • {(# H5 finishers) / (starting players per match)} / {1 - (avg time to quit) / (avg match time)} = (H3 hours per player) / (H5 hours per player)
  • {(X) / (8)} / {1 - (1min) / (7.5min)} = (14.6) / (20)
  • X = 5.08
This means that an average H5 beta match would had to have concluded with 5 or fewer players (with all of the quits occurring at the 1 minute mark) in order for the H3 beta per-player time to have been greater than the H5 beta per-player time. If the average quit time were longer, then more players would have had to quit. This is physically possible, but highly unlikely.

Regardless of how this data is sliced, under any reasonable assumptions, the per-player-hours for the H5 beta exceeded the per-player-hours for the H3 beta. Whether that says anything about the future success of H5 remains to be seen. My personal feeling is that the attempt to tie per-player-hours in a beta to sustained MP success is dubious at best.
Definitely sounds right. 7 minutes was about the average time per game. Some were shooter some were longer.
you should post a link of this on the other post. Good thinking btw.
NNMS MXMS wrote:
So in this popular thread, the OP links to an image describing the per player differences in total hours played between the H3 and H5 betas. The OP kindly also provided the links to his source information in the image (for which I thank him). Unfortunately, based on the source information provided, the OP's comparison is incorrect.

Per the H5 Beta data source, there were 20M matches played for a total of 2.5M gameplay hours. The OP took this number to mean hours played by player. That assumption is wrong. If there were 20M matches and the 2.5M hours meant by player, then we can calculate the average match duration:

  • {2.5M hrs / (8 players per match * 20M matches)} * (3600 seconds / hr) = 56 seconds per match (average)
The average match length for the H5 beta was certainly not only 56 seconds. So we know now that the H5 data cannot possibly mean hours by player. If, however, we assume the 2.5M hours is not by player, but rather by match, we get a much more sensible answer:

One question remains: Will the OP of the other thread generate and post a corrected infographic?
Original OP here. Caboose, fascinating take on the data provided, I applaud your efforts here. Yes I took the numbers to mean hours played by player. That is my assumption. However you have an assumption yourself, that 56 seconds per match (average) is incorrect?

56 seconds per match sounds correct to me, and forgive me, but I'm going to make more assumptions to help prove my original theory correct. Breakout, Breakout is the reason why 56 seconds average per MATCH is correct. Breakout matches end rapidly cause it only takes 4 kills and done, the MATCH has ended in well under a minute. However the Breakout GAME consist of about anywhere from 4-9 MATCH's.

The average time it takes to finish 1 Slayer game is 7.5 mins.
However, in that same timeframe you might have played 8 Matches of Breakout (1 full Game of Breakout) which would average out to 56 seconds per Match. I rarely played a Breakout Match that lasted a full min, the majority I would say were done with'n 15-25 seconds.

So considering that 1 GAME of Breakout is broken down into 4-9 MATCH's each lasting about 20 seconds, the average seconds per match could easily be 56 seconds across the H5 Beta. Again, we both have assumption that help fuel our own personal argument. One of us is right, the other has made a small/large calculation mistake. I can not with 100% accuracy say who is correct right now. Perhaps it's me, and it's possible it's you.
If it's you, will I make a new info graph? Negative, but I will issue an apology on my original thread with an assumption that you will do the same if my theory proves correct?
  1. 1
  2. ...
  3. 2
  4. 3
  5. ...
  6. 7