What with current posts on whether PTG is balanced and the 'Gunkulator'...

[gjnoronh]

Interesting why do you say that???   I'd think what most gamers want minimally is to have most army books  be reasonably balanced against the meta.  Not every army you could devise from every book but at least a fair number of different list builds from each book.  Also not every Rock Paper Scissors match up but reasonably Rock doesn't beat Paper, Scissors, Spock and Lizard at the same time.     

You've purposely said Gunk isn't about balance.

So what is balance for you? 

 

 

Separately take a look at the complexity of calculating value in Chess which is a much simpler game (in term of 'unit' / rules interaction)  and has  a much much larger data set available then AoS.    Consider the assumptions they are calculated under (the end game of a game with a universally defined "major'  win condition, no randomness and a universally defined start state of the game board etc.)   

https://en.wikipedia.org/wiki/Chess_piece_relative_value

What would happen to that calculated point value's accuracy if we change the assumptions - even something as simple as change the board size from game to game.   Or how about saying 'build your own army' for a chess game using any pieces you like (minimum/maximum one king)  up to X point value.    It's very likely  system wouldn't work for that purpose because the core assumptions have been changed.   

[Lemon Knuckles]

37 minutes ago, gjnoronh said:

Separately take a look at the complexity of calculating value in Chess which is a much simpler game (in term of 'unit' / rules interaction)  and has  a much much larger data set available then AoS.    Consider the assumptions they are calculated under (the end game of a game with a universally defined "major'  win condition, no randomness and a universally defined start state of the game board etc.)   

https://en.wikipedia.org/wiki/Chess_piece_relative_value

What would happen to that calculated point value's accuracy if we change the assumptions - even something as simple as saying 'build your own army' for a chess game using any pieces you like (minimum/maximum one king)  up to X point value.    It's very likely  system wouldn't work for that purpose because the core assumptions have been changed.   

I think this is an excellent thought experiment.

Chess is an almost perfectly symmetrical game, except for initiative, the impact of which is effectively minimal.  I think everyone would agree that it is balanced in the sense that the chance of white winning is equal to the chance of black winning, all of the time.

If we turn Chess into an asymmetrical game in the manner you describe where each player could build their own material using points, then balance cannot possibly mean the same thing.  It cannot mean that white and black are equally likely to win regardless.  If that's what we still mean, then we are essentially saying there is no difference between a symmetrical game and an asymmetrical game.  In fact, one way to understand the difference between the two is precisely through the reframing of concepts like balance.

Now humor me as I get repetitive.  Where I see a big problem with what you are saying is again when you conflate points with value, and where you ascribe value at the unit level.  Points are the means of creating a restrictive economy wherein I must make decisions.  I must make choices.  Given the budget I have to work with, I need to decide what resources I want to assemble.  The value of those resources is best thought through in terms of their relevance to my strategy.  Units cost points, strategies generate value, units advance strategy through tactics.

Our asymmetrical chess game would not be balanced if there was one strategy that could consistently beat all other strategies most of the time, especially if that strategy benefitted from initiative.  If that were the case, then, theoretically, white would have an overwhelming advantage, regardless of what black chose to do or how black chose to spend its points.

If, however, the value of all available strategies was directly influenced by the choice of opposing strategy such that Rock>Paper>Scissors>, it would be impossible to give either white or black an advantage in general, absent a specific instance of a specific game.

What I've just written is not as precisely worded as I would like, and I'm a bit rushed for time atm, but the seed of what I want to call attention to is in there.

I also think it is very important to reflect on how much you've hedged what you were saying earlier.  You've introduced the concept of a "meta"-game in your last response, and I would challenge you to think about the implications of that, because to me it seems very, very relevant. 

[gjnoronh]

I've discussed meta earlier in the thread.  Local meta may be very different then my meta or the 'gunkulator's' meta.   Meta reshapes actual/perceived value of a unit very quickly.  The accuracy of a points value (irrespective of how derived) changes quite a bit as Meta shifts happen. I've specifically previously discussed  the value (meaning either 'points' or 'value' ) of Rend on a unit and how that changed in a local (or global) meta as Nighthaunt enters it.   A   points system that wants to be highly accurate should try and reflect projected meta as best it can.  

I've disagreed on the stark value vs points dichotomy you present. I recognize they are separate but the economy of list design specifically is using points because of perceived and projected value to our  planned strategy.   Certainly in this particular broader discussion where Fulkes is hoping to readjust points to better reflect unit profiles (read value) they are similar concepts.    We assume in wargaming they increase fairly congruently - they may vary somewhat (and that's where the gold is in list building.)  So a very very high value piece at a very low points cost would probably be either 'list gold' or 'a mistake by  the army list writer'  depending on your perspective.   A very low value piece at a high points cost generally doesn't get fielded and might be considered a 'newbie trap.'  

So I'd edit your  sentence to say something roughly like ' Units cost points, we select where we want to spend our points based on projected value,    strategies ascribe  shape the  value of a  unit, units advance strategy through tactics.' 

[Lemon Knuckles]

1 hour ago, gjnoronh said:

I've disagreed on the stark value vs points dichotomy you present.

I wouldn't call it stark, and I definitely wouldn't call it a dichotomy.  They are not opposed to each other, but they are different things, and necessarily so.  

Here's how I understand you:

Balance = Accuracy

Accuracy is when Points = Value

You have said that Value can be influenced by things like scenario, terrain, opposing army, player skill, opposing player skill, etc, etc.

This conversation has been focused a lot on how perfect accuracy is not possible, but it is perhaps more illuminating to imagine what things would look like if it was possible.  Because as far as I can tell, such a thing would look like a game where any two players, regardless of differences in individual skill, could play any two lists, regardless of how well or how poorly constructed, from any two factions, regardless of how coherent or incoherent their respective battle-tomes, and have a equal chance of winning the game.  There's already a game that's pretty much like that.  It's called Heads or Tails.

Now I know that this is not what you mean.  But it seems to be the end result of what you are saying.  And what I am saying is that if the logical consequence of framing things like accuracy and balance and value in this way leads ultimately to an absurdity like the above, then we need to pull out the sledge-hammer and redefine our assumptions.  Because the assumptions here are not doing us any favors.

 

[Lemon Knuckles]

Three men each buy a brand new Maserati.

Man A drives his down the Autobahn and says it's great.

Man B drives his across a desert and says it's alright.

Man C drives his through a lake and says it's terrible.

Who's right?

[Ravinsild]

On 12/3/2018 at 5:32 PM, Fulkes said:

Accuracy as we know it is relative to the way the game is pointed. They are hand in hand concepts that I feel like you prioritize one part over another. By pointing the game more consistently the accuracy increases. Perfect accuracy can never be achieved in a vacuum just as a well balanced game can't be achieved in a vacuum.

Basically while I feel accuracy is important, I feel like you put too much important on how "accurate" a points system can be over all other aspects of the game. You also put too much importance, in my opinion, on the strongest possible states of a rule but ignore the possibility of their weakest states. The fact that any rule can have such a wide range of strengths or weaknesses in any game is why I feel like the potential target of the rule isn't important for "accuracy", "balance" or anything else. The rule should be pointed based on what it does (range it affects, bonus or penalty it provides) and not who it does it to (unless it has restrictions like SUMMONABLE or BONESPLITTER that weaken the rule in which case it should see a slight discount over standard versions of the rule that don't have restrictions that are as tight).

 

I'm not claiming to be the ultimate bastion of balance, and I've even argued that for balance to be achieved changing the values of specific abilities would need to change. If you find that attacks that do mortal wounds on a 6 are becoming too common or two powerful then increasing the value of those attacks and adjusting all models who have access to them by the same points cost is the thing I'd like to see done. Basically the premise is applying consistent and fair changes across the board instead of buffing some units or nerfing others without paying full attention to others who fit the same bill as the ones who are being affected in the first place.

Maybe giving an example would make it easier to see what I'm thinking. Let's take the humble Skeleton and lay out how I'd approach points costing using some of the example information I provided:

+4 for the 4+" of movement
+1 for the 1 wound
+1 for having a 6+ save
+1 for the shield save bonus
+5 for 10 bravery (+1 for every point over 5, under 5 I'd subtract 1 for each point under)
+9 for either weapon (+1 to hit and +1" of range are interchangable for points costing the way I offered before as you add +1 for every point either has) (example of the sword for a breakdown: 1 for range + 1 for attacks + 3 to hit + 3 to wound + 0 for rend + 1 for damage)
+1 for the Serve in Death ability
+3 for the full command (+1 each)
+1 for the Skeleton Legion ability
SUBTOTAL: 26 points

And then we take THAT and divide by 3 (because let's be honest for not being a monster (I'd double those points instead due to base size), hero (I'd add a tax for this as well as only divide it in half if it's not a monster) or wizard (also add a tax for this and follow the same rule as heroes) and we get 8 .666666666666666666... points per Skeleton which I'd round up to 9. So 90 for 10 in this example.

What about a Zombie?

+4 for having 4" of movement
+1 for the wounds count
+0 for having a - save
+5 for bravery
+7 for attacks (1 for range + 1 for attacks + 2 to hit + 2 to wound + 0 for rend + 1 for damage)
+1 for Dragged Down... ability
+1 for Serve in Death ability
+1 for Vigour Mortis ability
+2 for full command
SUBTOTAL: 22 points

Once again we divide by 3 and we get: 7.333333333333 points which we should call 8 (I'd only round down if it turned out that 80 for 10 was too much).

What about something tougher and from another book? Something like a Stormcast Liberator?

+5 for the 5+" of movement
+4 for the 2 wounds
+2 for having a 4+ save
+1 for the shield save bonus
+1 for 6 bravery
+11 for either Warblades or Warhammers (+1 to hit and +1 to wound are interchangable (example of the warhammer for a breakdown: 1 for range + 2 for attacks + 3 to hit + 4 to wound + 0 for rend + 1 for damage))
+1 for Grandhammer or Grandblade access (Since not every model can have them I don't add the full points I would normally for the extra rend and damage but instead divide the difference in the upgrade by half and then tax it onto every model)
+3 for the full command (+1 each)
+1 for the paired weapons ability
+1 for lay low the tyrants
SUBTOTAL: 29 points
Divide by three and we get 9.66666666... or 10 after rounding.

Now you'll noticed I applied the same method for everything (only charging 1 point for each command element, only charging 1 point per special rule, only valuing wounds at 1 point each) and that means if it comes across through playtesting that certain things are worth more than they're being charged for then they can go up in price (for example, maybe wounds go up to 2 points each) while other things (such as saves) might get cheaper if it's found that they're not worth as much (say dropping a 6+ to 0 and subtracting a point for a - save). This allows individual elements to be addressed more completely and they whole game to change because of it. For example, if we value wounds at 2 points each but drop the value of saves as mentioned above the points values would shift as follows:

Skeletons: 26/3=9ppm

Zombies: 21/3=7ppm

Liberators: 30/3=10ppm

Basically we could play with values like this all day and it's something you could even manage on a spreadsheet which would make revaluing all the models in the game a lot simplier if you're changing how you value components of the whole over everything else.

Now I -know- that there is some heavy discrepancy between what I points costed everything out to and what they are in the books, but these are examples taken in a vacuum how a standardized system could work, not a complete 100% playtested and established method how it should work. I presented this sort of thing to try and illustrate a clearer picture how one could use such a system and even how the way adjusting the points cost of components could shift the entire game up and down at the same time by addressing the issues of certain attributes being too strong or weak.

 

It’s extremely reminiscient of the point values assigned to home brewing a D&D race some guy made up and it was tested on the PHB races and official races and they’re all roughly even with a few outliers.

Pretty cool man  

[Ravinsild]

On 12/3/2018 at 7:06 PM, Lemon Knuckles said:

I enjoy talking with you, however in this instance I feel like you've radically misunderstood what I've written.  Of course the way we choose to allocate resources in army building is based on an assumption of projected value (and those who do it especially well do so within the context of a cohesive strategy).  That's by necessity given how points and army building have to work in a universalized system with incomplete information.  My intent with what I wrote was to explore the consequences of this assumption as assumption.  The point, it seems to me, is that it is a mistake to continue this type of thinking once the actual game starts.  It is wrong to play as if value in real game terms is simply an extension of the same mode of thought as points.  They are different.  

How often do you hear someone say something along the lines of, "unit X is great... I killed 600 points with it yesterday."  Does that necessarily tell you anything about the value created by that unit in that game?

Within the pointing system, degrees of magnitude always represent degrees of projected value (to use your term).  In game, however, incremental degrees of magnitude are often irrelevant.  In game, entire vectors of force are often irrelevant if they lead to a less advantageous position.   

You seem to implicitly recognize this yourself when you mention target selection, unless you mean target selection as simply a computational exercise that projects value in the same way as points project value in army building.  I think you are way too smart to mean that, though.  If you don't mean that, then there are other principles at work in value, principles that are not in the domain of the points system.

Yeah... I had a Predator one game with my World Eaters that never got a single shot off, but my opponent kept trying to melee it to death and it was such a massive distraction that it well made its value in points simply by existing and allowing me to move freely elsewhere. 

My opponent put tons of resources into shutting down my Predator tanks, which totally worked, until my Helbrutes and Berzerkers destroyed everything else and capped all the objectives. 

Overall it didn’t earn a single point in kills, but it was exceedingly valuable indirsctly nonetheless. It had nothing to do with points (145 iirc for quad lascannons load out) and everything with what it did during the game which was be really scary and distracting. 

Put simply: it did a lot by doing nothing.

[gjnoronh]

18 hours ago, Lemon Knuckles said:

Three men each buy a brand new Maserati.

Man A drives his down the Autobahn and says it's great.

Man B drives his across a desert and says it's alright.

Man C drives his through a lake and says it's terrible.

Who's right?

Bingo that is  more or less the point I made earlier in the thread about how user skill can make it harder to determine if the  user perceived value is a correct representation of the accuracy of the point value ascribed to a unit.  Those three users might have very different perceived values of a Maserati vs a 30 year old truck.     Note some units  and armies have that lake/dessert/Autobohn learning curve and some do not.  

For a lot of units that skill variance won't matter in how we perceive value but for some it does.  That's the point I made on Frenzied units and Tomb Kings in previous editions and Stradivarius violins and my 10 year old daughter earlier.     

The majority of AoS players (total globe) are probably around  Man C (for purposes of this example even if it's not quite fair to most of us)

People who are intensively looking into rules balance and reading forums to get tips are probably around Man B.The guys consistently at the top of large GT's are Man A.  (Or you could pretend Man C is the average driver, B is someone at the peak intersection of reflex time and driving experience curves and A is a professional race car driver)

Your underlying assumption of the user skill (Man A, B, C) is going to effect how you calculate 'accurate' points.  That doesn't mean the points should change based on skill level of player   but you have to recognize that your assumptions on 'presumed  user skill' as a variable  effect what the 'right' points value of a unit should be.   

Same issue with list design and synergies.  You make assumptions in game design based on how synergystic people will make their list and how carefully they will read their local meta.   If a  list fielded  is below that presumed  level of synergy/meta awareness it's likely  going to be  less effective then one at or above the presumed level of synergy.     

On the point of Meta and Terrain and how it effects perceived value.  If the designerrs assume everyone is going to be driving on the Autobohn but local tables are actually far more often Lakes how we  should value a Maserati vs a canoe  may be different then the designers expected.    

Mind you Lemon you have said Gunk isn't about balance - but I don't think you've explained what balance is for you.  

I am not saying Accuracy = Balance.    I do think accuracy helps us achieve balance.    And again for someone like Fulkes and for me the goal   from points we ascribe to units  is probably balance. 

To be clear I'm not saying perfect accuracy is the only acceptable goal, or that perfect balance is.  I am saying (repetitively) a  math based analysis a highly chaotic and largely incomplete data set  that ignore synergy/expected utility  (and whose inputs are best guesses and playtesting) is probably not going to end up with more accurate results then  best guesses and playtesting.  If your goal is balance. If your goal is consistency sure you can get there.  

[Lemon Knuckles]

1 hour ago, gjnoronh said:

What's your underlying assumption of the user skill (Man A, B, C) is going to effect how you calculate 'accurate' points.  That doesn't mean the points should change based on skill level of player   but you have to recognize that your assumptions on 'presumed  user skill' as a variable  effect what the 'right' points value of a unit should be.   

Same issue with list design and synergies.  You make assumptions in game design based on how synergystic people will make their list and how carefully they will read their local meta.   It's just best guess

There is no underlying assumption about those things.  They do not belong within the domain of what points refer to.  We have all agreed that it is impossible for points to do so.  You see the impossibility as a computational problem, and your answer is to go fuzzy, squint hard and guess.  I see the impossibility as a conceptual problem borne from being slave to the notion that points = value, and therefore it exists only insofar as we are thinking about the problem incorrectly.  It is akin to trying to solve a quantum mechanical problem with classical mechanics.  It's impossible.  But it's impossible because they are different languages, and the rely on different assumptions and different conceptual understanding.  As soon as you recognize that it is not the job of classical mechanics to describe quantum phenomena, the problem disappears.  You can insist that it should, and fudge a whole lot, but that is not a problem with the math, that's a problem with being stubborn and dogmatic.

1 hour ago, gjnoronh said:

Mind you Lemon you have said Gunk isn't about balance - but I don't think you've explained what balance is for you.  

I have explained what balance is for me, in terms of value, a few times at least.

23 hours ago, Lemon Knuckles said:

In that context, balance is not about ensuring Scissors and Rock are equally viable against each other, but rather about ensuring that Rock doesn't beat everything.

20 hours ago, Lemon Knuckles said:

Our asymmetrical chess game would not be balanced if there was one strategy that could consistently beat all other strategies most of the time, especially if that strategy benefitted from initiative.  If that were the case, then, theoretically, white would have an overwhelming advantage, regardless of what black chose to do or how black chose to spend its points.

Points by necessity can only speak about what's there.  They can only speak of the unit.  Value always has context, and must always speak beyond the unit.  Balance in the pointing system occurs at the unit level.  Balance in terms of value, in terms of an actual game state, does not.

The game would be unbalanced if one strategy dominated all others.  If Alpha Strike always won, if there was no reliable counter-play, then those armies best capable of playing Alpha Strike would always be advantaged, even in the abstract absent a definite game state.  That would be an unbalanced meta.  I don't think balance means that there should never be an advantage inherent in any initial game-state... there will always be good and bad match-ups.  I also don't think it means that all armies and all strategies are equally viable across all metas... there will always be tiers.  But the meta and the tiers are fluid, and change through time as things shift.  As long as no one strategy is dominant, as long as multiple strategies and therefore multiple armies are viable, then the meta is balanced.  And as long as the meta and tiers remain fluid, then the game is balanced.

 

 

 

[gjnoronh]

Sorry I missed or forgot that point.   I think that's a lower bar for balance then I'd set - but really we're pretty close to each other.   Aside from how we think of value vs points  But I appreciate your last post on points vs value.  If points don't reflect value at all then how do players decide where to spend their points if they aren't projecting value (in the context of their planned strategy and other choices.)  

 

I'd argue for balance  it's a problem not just when Rock beats Paper Scissors Spock Lizard but also when Lizard beats no one more then 30% of the time, or when Rock beats Scissors, Spock, Lizard and only loses to Paper some of the time.  How much of a problem depends on the tolerance level  for imbalance in the player base. 

You are focusing on strategies in your Alpha Strike example.  I think the concern for players is the Army Book rather then the strategy (though some books favor certain strategies by design)  because if you can take your $1000 of minis and several hundred hours of painting and do an alternate strategy effectively you aren't going to be too worried if X strategy is better then Y.  But if you can't take your investment and be competitive that makes players angsty.  

Unlike a lot of game systems it's much harder in AoS (Because of the time/money and resulting emotional investment)  to say 'this faction isn't good I'll switch" other smaller  investment games like Blood Bowl people don't get angsty when some teams are just bad and others are very good (in fact a priori the rules are designed to created tiered win rates in Blood Bowl backed by large scale data sets from online play.)   

I'd disagree that building a points system for units don't make assumptions about user skill.   If we assume the average player is going to drive the Maserati into the lake then there is no reason to  cost the Maserati more then the rusted Jalopy.   In your analogy getting the points accurate ends up being more quantum phenomena then it is classical math.  Deciding we want to use classical math to solve a quantum problem probably isn't going to work out well.   Someone convinced they can use classical math to solve a quantum problem may be the one being "stubborn and dogmatic."   (By the way I'm not arguing for quantum phenomenon just that it's a "Big Data" type problem to get the math highly accurate for a very complex set of interactions not a limited data approach.)  

 

[JPjr]

1 hour ago, gjnoronh said:

in fact a priori the rules are designed to created tiered win rates in Blood Bowl

 

hey @gjnoronh, quick one here. Seen you mention this a couple of times now and meant to ask is there any good articles or anything online explaining this? Just so I'm clear teams in BB have been designed to be imbalanced, with certain teams made to win more. Never knew that and it's both slightly weird & very interesting. At tournaments (I assume tournaments do take place) is there some kind of levelling mechanism or does everyone (or nearly everyone) just bring the team created to be strongest?

[Lemon Knuckles]

1 hour ago, gjnoronh said:

 but really we're pretty close to each other. 

I am glad that we might be finding orbit with each other rather than just streaking past each other in the night sky.

1 hour ago, gjnoronh said:

 If points don't reflect value at all then how do players decide where to spend their points if they aren't projecting value (in the context of their planned strategy and other choices.)  

Perhaps it is reasonable to say that points reflect value the way chemistry reflects biology?   I'm not entirely sure that lines up with what I have in mind, but good enough for now.

1 hour ago, gjnoronh said:

You are focusing on strategies in your Alpha Strike example.  I think the concern for players is the Army Book rather then the strategy (though some books favor certain strategies by design)  because if you can take your $1000 of minis and several hundred hours of painting and do an alternate strategy effectively you aren't going to be too worried if X strategy is better then Y.  But if you can't take your investment and be competitive that makes players angsty.  

The only thing different there versus a game like say MtG is the hobby aspect, the investment of time and emotion.  I agree that it is a difference, but I'm not sure how relevant it is to competitive play.  The minute we start talking about points, we are focusing on one type of gameplay only.  And although these feelings exist, I'm not sure they are relevant in principle.  They may be relevant because we want and agree for them to be so, but they are not relevant by necessity.  When I played MtG, I was a mostly a Johnny, and favored the homebrew rogue deck over any of the established decks.  My decks were rarely as competitive, but that's my choice and not a question of the balance of the game. 

Similarly, and touching on an example I used way back, the NHL is a competitive hockey league that creates "balance" through the implementation of a salary cap, and through a structured draft process.  I think it is wrong to say that the league is unbalanced given that some teams are better than other teams, when that difference is due to differences in scouting, player evaluation, contracting, talent development, and coaching strategy.  Indeed, the "meta" and "tiers" of the NHL are fluid, and change over time.  There are no real dynasties anymore.

1 hour ago, gjnoronh said:

Unlike a lot of game systems it's much harder in AoS (Because of the time/money and resulting emotional investment)  to say 'this faction isn't good I'll switch" other smaller  investment games like Blood Bowl people don't get angsty when some teams are just bad and others are very good (in fact a priori the rules are designed to created tiered win rates in Blood Bowl backed by large scale data sets from online play.)   

That is definitely true, and opens the door for us to make normative claims about what we want points to be, rather than what they are, if we so choose.      

1 hour ago, gjnoronh said:

If we assume the average player is going to drive the Maserati into the lake then there is no reason to  cost the Maserati more then the rusted Jalopy. 

On what basis do we make that assumption?  And what happens to balance the moment players realize that the can buy a Maserati for the same price as a rusted Jalopy?  Players are smart enough to find and exploit that market efficiency.  They will seize on it.  Take the example of the most broken MtG card that was around back when I was playing standard... Skullclamp.  It seems innocent enough on the surface as none of its effects seem particularly overpowering.  But it was capital B broken precisely because it was priced wrong, because of the assumptions they made in playtesting about what people would do with it, instead of pricing it simply for what it was.  

2 hours ago, gjnoronh said:

In your analogy getting the points accurate ends up being more quantum phenomena then it is classical math. 

But only because of the way you are defining accuracy.  The Maserati gives me everything I need to price it.  Labor, material, infrastructure and supply-chain costs.  The price can't be effected by what I do with it.  Why would a Maserati cost me less if I decided to park it on my front lawn and grow daises out the sunroof, or indeed because someone, somewhere might decide to do so?   The value of the Maserati is emergent.  The cost of the Maserati is self-contained.

[gjnoronh]

Sure here's a very brief overview of the history

https://www.thenaf.net/2017/05/tiers/

Tournaments can either choose to incentivize lower tiers by providing various bonuses to them (an approach used in the WFB world at times) or choose not to ('unadultered rules' blood bowl) really up to the TO.  

GW has taken back rules writing from the player base (though there may be input we're unaware of) so I don't know if they are using objective win rates to help  guide the minor recent rules tweaks they have done in the BB 2016 era.  They also used data from Cyanide (see below) for their BB 2016 rules set I believe.  

There are three large data sets I'm aware of

1. FUMBBL a longstanding free online platform to play Blood Bowl used by the Blood Bowl Rules Committee particularly to my understanding.  Allowed creation of test team rules that they could watch how it plays.  

2. NAF tournament results - almost all tournaments in the world submit standardized results to the NAF these can provide rankings of players and some fancy data for in depth analysis as below.  https://www.thenaf.net/rankings/

3. Cyanide which runs two (maybe 3 depending on how you count it) of online Blood Bowl in their proprietary software with many many many matches.  There current season IIRC has 30,000 registered teams.  That data is largely non public except when they choose to publish something.

 

In the post BBRC era there is a lot of pretty nice statistical analysis still being done particularly off the NAF data set. Here is a 17 page multiyear thread (starts in 2013) with ongoing math analysis of tournament results.    http://www.talkfantasyfootball.org/viewtopic.php?f=81&t=40863&start=240

Here's a nice overview :https://public.tableau.com/profile/mike.sann0638.davies#!/vizhome/NAFGames_0/RaceGridNumbers

You can do some pretty cool things playing with the data set Mike who builds those analyses puts out various analyses as the mood strikes him (specific rules sets for example he published one just looking at games played under the rules set being used for the 2019 world championship) 

An overview of a lot of his recent publications is here:

https://public.tableau.com/profile/mike.sann0638.davies#!/

That being said one interesting thing that's understood in Blood Bowl data analysis is that at different levels of Team Value  (in the AoS world points) it's very clear some teams become much much better or much much worse.   What hits a certain desired 'power goal' at the standard game level doesn't work as initially intended at higher levels.  Because Blood Bowl has a team 'growth' system built into the rules set (for both league and to an extent tournament play)  it's not unusual to have games at  high Team Values.

[gjnoronh]

42 minutes ago, Lemon Knuckles said:

       

On what basis do we make that assumption?  And what happens to balance the moment players realize that the can buy a Maserati for the same price as a rusted Jalopy?  Players are smart enough to find and exploit that market efficiency.  They will seize on it.  Take the example of the most broken MtG card that was around back when I was playing standard... Skullclamp.  It seems innocent enough on the surface as none of its effects seem particularly overpowering.  But it was capital B broken precisely because it was priced wrong, because of the assumptions they made in playtesting about what people would do with it, instead of pricing it simply for what it was.  

But only because of the way you are defining accuracy.  The Maserati gives me everything I need to price it.  Labor, material, infrastructure and supply-chain costs.  The price can't be effected by what I do with it.  Why would a Maserati cost me less if I decided to park it on my front lawn and grow daises out the sunroof, or indeed because someone, somewhere might decide to do so?   The value of the Maserati is emergent.  The cost of the Maserati is self-contained.

When you make a decision as a player on where to spend your points you are doing your best guesses on what you think something will be worth you (value)

The Maserati in real life and particularly in this case has it's cost driven in particular by perceived value to the customer.    I could charge my daughter $3000 US for a Maserati (good value for someone with a license) to her it's worth nothing. Car brands (and other brands) charge above their cost to produce how high depends on their perception of brand value to the customer.

In AoS    the + 1 to hit for khorne keyword only models in 8 inches  has some perceived value to a Khorne player and almost none to a Wanderer's player (A wanderers keyword+1 to hit might have a different value).   The intrinsic components used to build it (stat line the presence of a AoE buff) doesn't  matter to the user as much as the perceived value in the context of their army list and strategy. How that value compares to the points (resources) they have to allocate to obtain it is how folks decide what to put in their list (amongst other factors in the AoS world like cash cost, emotional investment,  appearance factor, theming, emotional attachment to the story line etc.)   Back to the Fulkes discussion - how that unit choice synergizes in the context of the army (outside of highly variant Open play we we don't expect Khorne models in a Wanderers list!) matters.  

 

As you are clearly heavily influenced by Magic I read this earlier today and thought it was pertinent to our discussion about Man A, Man B, Man C and whom the designers are designing for.  (Yes we could design for all three, but in Magic they appear to have made a choice.)

https://adjameson.wordpress.com/2018/12/04/an-open-letter-to-cedric-phillips-gerry-thompson-and-the-pro-magic-community-at-large/

 

 

[Lemon Knuckles]

3 minutes ago, gjnoronh said:

When you make a decision as a player on where to spend your points you are doing your best guesses on what you think something will be worth you (value)

Completely agree with you

3 minutes ago, gjnoronh said:

The Maserati in real life and particularly in this case has it's cost driven in particular by perceived value to the customer.   

It has its price driven by those things, not its cost.  Its cost is its cost, regardless of perceived value.  Its price is a reflection of the fact that we have a free economy, unlike points which is a closed economy.  I did hold up the notion of a free market for points as perhaps as good a model as we could get to have points work the way you want them to, but it does seem impossible for practical reasons.

6 minutes ago, gjnoronh said:

The + 1 to hit for khorne keyword only models in 8 inches  has some perceived value to a Khorne player and almost none to a Wanderer's player.   The intrinsic components used to build it (stat line the presence of a AoE buff) doesn't  matter to the user as much as the perceived value in the context of their army list and strategy. How that value compares to the points (resources) they have to allocate to obtain it is how folks decide what to put in their list

Again, I completely agree.  It remains a mystery to me how we agree on so much but follow those beliefs to two very different places.

8 minutes ago, gjnoronh said:

(amongst other factors in the AoS world like cash cost, emotional investment,  appearance factor, theming, emotional attachment to the story line etc.)   

Although all these other factors are doubtlessly real, I'd like to continue to ignore them as relevant to the points system.  Do see any problem with that?

9 minutes ago, gjnoronh said:

As you are clearly heavily influenced by Magic I read this earlier today and thought it was pertinent.   It's pertinent our discussion about Man A, Man B, Man C and whom the designers are designing for.  (Yes we could design for all three, but in Magic they appear to have made a choice.)

https://adjameson.wordpress.com/2018/12/04/an-open-letter-to-cedric-phillips-gerry-thompson-and-the-pro-magic-community-at-large/

Thanks, I'll read it and let you know if I get anything from it.  I also read this that you wrote earlier, and find it very, very relevant to our conversation.

37 minutes ago, gjnoronh said:

That being said one interesting thing that's understood in Blood Bowl data analysis is that at different levels of Team Value  (in the AoS world points) it's very clear some teams become much much better or much much worse.   

  

[JPjr]

@gjnoronh thanks! Will dig into all that.

[gjnoronh]

Yeah there is a lot of accumulated  remarkably well analyzed data in the Blood Bowl world in a much more limited complexity game system. And there is a test environment where you can do tweaks and collect playtest data from thousands or more games in a well defined environment.      There is a wealth of data on how player skill level (ELO and GLICKO ratings) effect outcomes and how that is effected by the list they are playing in any given head to head match up.   

That's a pretty amazing data basis to do math from and even then it's still best guesses and playtests that drive point cost adjustments rather then mathematically derived formulae.    

@Lemon Knuckles that comment on Blood Bowl  says to me context matters and experienced gamers in a stable well analyzed system know it.  Some teams also get better in different skill package formats even at the same TV or under specific rules sets for the events.   

If points were a universal mathematically derived certainty that stuff shouldn't matter.  The ups and downs of whats good at different Blood Bowl Team Values argues those points relationship to actual value is actually context dependent.    The game is largely balanced (intentionally unbalanced mind you!)  to work  as intended around low TV teams in a league environment and not other environments like online matchmaking play where Team Value similarities drive matchmaking.   And again variance in matchmaking has defined data on how that effects performance.  

 

[Lemon Knuckles]

41 minutes ago, gjnoronh said:

@Lemon Knuckles that comment on Blood Bowl  says to me context matters and experienced gamers in a stable well analyzed system know it.  Some teams also get better in different skill package formats even at the same TV or under specific rules sets for the events.   

I think everybody knows it.  Every single person who has participated in this discussion readily acknowledges it, and no one has ever even hinted at suggesting otherwise.

42 minutes ago, gjnoronh said:

If points were a universal mathematically derived certainty that stuff shouldn't matter. 

Points can be any way you design them to be.  It is certainly possible to derive them mathematically.  Yes, they will not accurately reflect value, but there is no methodology of deriving points that will.  No matter what way you choose to design them, context will always matter, since value emerges from context.  In fact, I'd rephrase what you said as "If points equaled value, then context wouldn't matter.  Since context always matters, points do not equal value."  By logical necessity, points exist outside of context. 

Now you might say, hey, we can analyze lots and lots of context and try to abstract something from that and then import that abstraction back into the points system.  But there are at least two insurmountable challenges with this:

1. Given the variety of people, strategies, context, etc., it will never be accurate on an individual level (maybe you could argue that it is accurate to some statistical abstraction)
2. And more importantly, points themselves are part of the context.  The second you import that abstraction back into the points system, you've changed the economy, and thus you've changed the context, and thus it is no longer even accurate to a statistical abstraction of what the game will be like going forward.  New strategies and new valuations will emerge.  It is infinite regress.  

[JPjr]

16 hours ago, gjnoronh said:

As you are clearly heavily influenced by Magic I read this earlier today and thought it was pertinent to our discussion about Man A, Man B, Man C and whom the designers are designing for.  (Yes we could design for all three, but in Magic they appear to have made a choice.)

https://adjameson.wordpress.com/2018/12/04/an-open-letter-to-cedric-phillips-gerry-thompson-and-the-pro-magic-community-at-large/

this was really interesting, thanks! (and also chimes with my uneducated, un-backed up belief that despite making the most noise online and being the most visible tournament/competitive AoS players probably make up just a minuscule fraction of the overall community).

[gjnoronh]

23 hours ago, Lemon Knuckles said:

I think everybody knows it.  Every single person who has participated in this discussion readily acknowledges it, and no one has ever even hinted at suggesting otherwise. 

My friend you may know it and I may know it.   "Context matters in evaluating how effective something  is in a wargame. " 

There are others who have been vocal in this thread who think we can just  do rough math based on unit stat lines to get  more accurate points then  GW  is currently publishing while specifically  ignoring easily identified contextual factors like in army synergy.   

There is no way of getting perfectly accurate value to reflect all individual tabletop situations. There are ways manufacturers choose to determine  points that are roughly accurate for most table top situations based on their assumptions on how the game will probably be played  (and sometimes clearly roughly inaccurate) what size, what skill, what terrain, etc.       For GW (and AFAIK other manufacturers I am aware of) it appears to be  rough guess and playtesting tweaks and then community feedback. That's not a perfect model because say a typical level of 100 playtest games by playtesters (specific  skill levels,  degree of competitive/inventive personalities, and terrain/point levels) pre army release won't pick up everything that thousands  of real world games  in diverse environments  will.  GW has the volume of players to have many many many times the playtest number of games played in the real world  the first weekend after an army book is released.      Best guess and playtest isn't perfect. Community feedback adds accuracy  but is imperfect as well as loud but factually incorrect groups may unduly influence the process.  

I am simply arguing I don't believe we can use simplistic math to get to high degrees of accuracy on points.    More explicitly to get a more accurate mathematical derivation of "correct" points then the current  system you are going to have to come with a model that includes more then unit stat lines and flat values for a given buff in all armies irrespective of expected synergy.   Particularly if your mathematical model  starts with best guess and is adjusted by playtesting.   Both of  which data sources the model maker has already deemed   intrinsically unacceptably  inaccurate to give us points values.     You can get rough math at best which probably isn't more accurate then what we have now.  

You aren't arguing (I believe)  for a  simplistic math model  to determine 'accurate' points (and in fact have said your goal for math has nothing to do with balance/accuracy) but that has been the argument I've been responding to in this conversation.

So why are we disagreeing?

[JPjr]

Just tossing this in here, hail eris!

https://en.wikipedia.org/wiki/Chess_piece_relative_value

[gjnoronh]

4 hours ago, JPjr said:

this was really interesting, thanks! (and also chimes with my uneducated, un-backed up belief that despite making the most noise online and being the most visible tournament/competitive AoS players probably make up just a minuscule fraction of the overall community).

Yeah that article has started an interesting discussion amongst some event organizers for AoS.   To what extent should  grand tournament rules sets cater to the guys who are 'serious players' who are travelling the nation/globe competing in the largest events and to what extent should it be for the guy who is going to his first large tournament.    I'm of the mind where AoS is now our primary focus should be the latter new player group while giving  the highly competitive group plenty of room to show off their skills and compete against other top players meaningfully.   In context of this discussion it's expect a lot of attendees to be  Man B/C, make sure Man C doesn't get scared off and quit or drive his car into the wall and cause a pile up,   and give Man A   room  to try and outrace the best of the best on a  fast  but safe track.   

I know tournament players is a very very small portion of my personal local community.  And high end serious tournament players is an even smaller number.  For my local commmunity I would guess it's probably 5-6/100 for the first group and 3/100 for the second.   Those number derived from my experiences running  a large scale tournament that draws from people driving 10-12 hours to get here but out  of my locals I get a very limited section of the guys I know are playing.  

 

[gjnoronh]

5 hours ago, JPjr said:

Just tossing this in here, hail eris!

https://en.wikipedia.org/wiki/Chess_piece_relative_value

Yeah I shared it at the top of the page.   It's a good read and suggests if you read carefully the math is complex and context dependent.  

[JPjr]

@gjnoronh ah ******. so you did. too many links to keep up with and more maths than I've contemplated seriously for about 25 years...

 

[Lemon Knuckles]

6 hours ago, gjnoronh said:

So why are we disagreeing?

We're both wondering that a lot, aren't we?  To be clear, my "disagreement" is academic and not political.  I don't have a stake in the outcome, I'm not advocating to change anything.  I just want to understand and explore things.  In that context, you've been fantastic to talk with.  Any disagreement you sense on my part comes down to the question of Why?  If points can never be accurate to value in the way you define (I agree) then it's logically clear that points cannot equal value.  But it seems important to you to maintain that points equal value, and I guess I'm just trying to grasp what's at stake for you.  Why is it important?  What are the negative consequences to assuming the more logical view, or what benefit is lost if we abandon the less logical view.  They Why? to me is the source of confusion, and disagreement such as it is is really just a desire to understand.

On 12/5/2018 at 8:19 AM, gjnoronh said:

Mind you Lemon you have said Gunk isn't about balance

I think the conversation has left Gunk far behind, but if it's at all helpful, I can tell the story.  Gunk started, as most of these kinds of things do, as a little crunch I did to try and help me understand what I was buying and help shed light on list-building ideas.  It originally had nothing to do with points at all.  I just wanted to quickly calculate how much force I could bring to bear with different units, under different conditions.  What the effect of this kind of buff or artefact might bring, etc.  It then led me to comparisons, and then comparisons of different aspects, and finally to points efficiencies.  If I needed "something" in my list, I could compare how much of that something I could get on a per point basis.  It was one of those things that I imagine a lot of people do in different ways.

An interesting thing happened when people started posting in the GW Wants Feedback for GHB 19 points discussion.  People would say things like, "I feel Unit X should cost 20 points more/less," etc.  And when out of curiosity I ran the unit through gunk, they were often "right" (whatever that word means in this context).  Which is to say, people's feelings on what a unit should cost overlapped in a large degree with what gunk happened to say it should cost.  Not always, and not exactly, but in the majority of cases, and directionally so.  It made me curious the way people felt about units seemed to converge with what an objective math process said about those units.  Which brought up the idea to focus on gunk and see if it couldn't be used to crack the code and reverse engineer the GW points system.  If it couldn't help "explain" the system.  The methodological premise was always that the GW system was (mostly) correct, but that there were outliers (as evidenced by people giving opinions on points updates) that could perhaps be better understood and contextualized once we understood how points currently work.  So, in that sense, it was never about balance, but that doesn't mean it was against balance.

23 hours ago, gjnoronh said:

https://adjameson.wordpress.com/2018/12/04/an-open-letter-to-cedric-phillips-gerry-thompson-and-the-pro-magic-community-at-large/

Had a chance to read it today.  Good, well-written article, and credible.  Reminded me of my own experiences getting into AoS.  I am late to the tabletop world.  Never played WHFB or any other tt game.  My initial experiences were completely overwhelming.  I remember my first game.  It was against a Slyvaneth list, and I had my hands more than full just trying to remember the rules and sequence of actions, and trying to remember what my guys did and when they should do it, and where they should be relative to other units in order for things to work.  After deployment, my opponent had some allegiance ability that let him basically move all his guys around before the game even started.  Blew my mind.  And my mind kept being blown.  You guy can do that?  You have an ability that lets you do that?  etc, etc.  So much time spent looking at the app, at my books, at his books, at FAQs, etc.  I think we played for 3.5 hours and barely managed to finish round 2.  I get that complexity is a barrier to entry, and I think I've always advocated in other conversations for less complexity and less book-keeping.  That's why I thought the Swedish Comp thing you shared was interesting but misguided.  It's also why I said in my first theory blog that although thinking about the game theoretically can be creative, interesting and even fun, it will rarely be valuable in the sense that we'd be much better served from a "play better" standpoint by spending our time on just staying informed and current with the knowledge of what things exist and how things work, and by studying tactics.  (I do hold out hope that theory can ultimately be applied to simplify the game and simplify in-game decision making, although the process of getting there might not be simple).

Anyway, I get the sense from your sharing that article in particular, and also form the way you've used the Man A/B/C example in a way that wasn't how I was thinking it when I wrote it, that your Why? has something to do with balance, and community.  I'm not sure I connect the dots to the points discussion though, as I think those kinds of concerns are much better addressed in rules design, battle-tome design, FAQs, rules overhead, etc.  But I take those kinds of concerns seriously, even if I don't yet see the connection.