Ghost In The Machine

David Kennerly wrote an article a while back on why MMOG players don't really want randomness in the sense as it is usually discussed colloquially ("a die-roll").  Specifically he suggests that we would feel better if virtual worlds adopted mechanisms whose random samples were based on replacement (akin to pulling cards from a deck, "Randomness without Replacement").  One way of looking at this is through an analogy.  Damion Schubert a while back wrote an excellent piece about the the state of design of AI characters in MMOGs - namely their stupidity is their greatest virtue:  they present a pattern, a "puzzle the player needs to solve..."  And too a puzzle the player can solve."

If it is really unfettered chance and serendipity that we despise, is card counting the obvious successor?   Perhaps what we really want is visibility into the breadth of the function of the world...

Brian Green recently penned an essay about Why the Monthly Subscription is Doomed.  The conundrum he crisply states as (my emphasis):

The main problem I've run into as a small game operator is that this business model doesn't scale down very well. Let me illustrate: if I have 1,000 subscribers and charge them $10/month, I have a yearly income of $120,000. If another game has 1,000,000 subscribers and charges the same $10/month, then they make $120 million. Obviously, the bigger game has more funding to do more development work. Yet, to the individual players, it appears that they are paying the exact same price for a different amount of development. 

He hints that the MMOG machine through whatever design or implementation fault may not communicate the value of component player investments.  Is it that players look for patterns of developer input suggesting world vitality (via the proxy of perceived developer investment in the world)?

The distinction between pattern and predictable result may be an important one.  Another way of thinking about it might be that we seek the illusion of randomness and variability, but that in fact we really want our toys and drops predictably.  We want to be able to touch the gears of the game world machine.

A recent Bear Stearns analysis of the Asian Online Games sector suggests that game server piracy stems from players wanting to avoid paying subscriptions, but the kicker is that they may be quite willing to "pay for perks" (phrase from here).  Perhaps the explanation is that paying to touch the gears of the machine directly is a whole lot more fun than the mere opportunity.

Comments

[magicback (Frank)]

1.

Nate,

Just to state the obvious as a basis: the Internet is an enabling technology and most view it as such. The Net enables the consumer to consume the specific type of entertainment at the price point they mark as their price points.

Thus, the pay-for-peformance model that Google as mastered to a tune of $4b+ revenue is the current model. This is also the underlying factor to the pricing of Cable services and also digital music downloads.

And thus, a flat fee subscription is doomed as model as Brian Green explained and also in similar ways as hourly fees was doomed.

The pay-for-performance model appeals to consumption behaviours that we have adopted over the years. I'm not an expert on the underlying behaviors, but I do see a trend in behavior adoption.

Here is a behavior trend I see based on some research on micropayments:
1) 10 $1 purchases is much easier to decide on than 1 $10 purchase.
2) But, 100 $0.10 purchases is tougher to manage than 1 $10 pruchase.
3) This is do to our mental accounting. $10 purchase one time is a big decision. $1 purchases 10 times is easier to manage for many as each decide how many $1 purchases they can accept. $0.10 purchase over 100 times is just too fussy.

So in terms of patterns, it is really about adoption of a puchase pattern that people have gotten used to: Google Ads, Music downloads, ebaying, paypal, the probability inherent in a 52 card, etc.

Problem solving is one element, but I think the neglected element is the desire to find familar and comforting patterns: Hero's journey, narratives in 3 acts, gravity, cause and effect, randomness based on reducing probability space, etc.

This applys to gameplay and also revenue models.

Frank

[Dan S]

2.

Runescape is already doing rather well with a pay-for-perks model. AO is also running in a similar manner. You don't HAVE to pay for the fuller games in these cases, and I'm sure many third-world players don't. But you do get a much richer experience, a much larger game universe if you do.

I find myself gravitating back to smaller games though, in contrast to the seeming market trend. The paradox of large development budgets is that they rarely support more real content, more dynamic games. They ossify and dumb down to try to capture more of the median, thus larger market share. If you want cutting edge, you do not play large games (witness the wholesale dumbing-down of SWG over the last couple of years...)

That isn't so surprising considering that most real innovation comes from entrepreneurial endeavor, not established companies, but somehow it seems to get lost in the press on games lately.

But to drag back onto topic, the evil RNGs are a constant complaint in games. Streaks of the sort we see from "real" RNGs just don't FEEL right. We love them when they are positive, but when they are negative we suspect the developers are somehow playing with the numbers.

They could be normalized, but that would yield an experience where whittling down things was the standard play; there would be no critical success (or failure) to vary results in a dramatic fashion. The only answer I can see to the trade-off is more player-skill based systems, and that tends towards twitch where non-player factors (latency, system performance, networks) too often play a determining role still. But if/when those factors fall significantly in import, a system based on a smaller range of cummulative results could work well with player-skill playing a larger role. Randomness could be mostly factored out.

[nate combs]

3.

>RNGs (Random Number Generators)

Additional (technical) discussion here:

http://www.vanguardsoh.com/forums/showthread.php?threadid=11093

As one comment noted, the issue seems less about the details of technique as much as 'nature of randomness' and its role in a virtual world...

[nate combs]

4.

>"familar and comforting patterns"

Also too, given the suspicions that often underlie RNG discussions (per Dan's point): a (perceived) fair pattern too.

[Franek]

5.

A lot of folks have exposed the relationship of the frustration/boredom dimension in game system design. At the end of this post I put an example of my understanding of how it may work.

The question seems to be how, once you’ve identified a target point along that dimension for some aspect of your design you maintain it for particular players. To what extent do you build your solution into the underlying mechanics of the game (precompiled tables of probabilities applied to combat events) vs. the extent you special case situations to accommodate the balance you are trying to achieve (Apply a “miss counter” to probabilistic events and when a player approaches the “frustration level” of misses intervene). Do folks have any insight on the balance between one-off tweaking and generic solutions?

How some psychological considerations may bear on satisfaction with game mechanics:

1) the role of understanding in satisfaction (pattern) : To what depth do people have to go to feel they understand something to be satisfied? At one end of the scale is superficiality (apples fall from trees) at the other end is insight into first principles (GUT). Are the patterns exposed by greater depth of understanding extendable to other aspects of the world? For game systems then, how do you discern your target population’s place along those dimensions?

2) the role of predictability in satisfaction: are people more satisfied if they perceive the environment to be stable or changing? At one end of the scale are deterministic environments (I can kill that MOB every time) and at the other end is continuous upheaval (every time I run into that creature it uses a different attack and it wastes me!)

3) the tension between frustration and boredom: for both understanding and predictability, how do you design your game to be sensitive to each player’s potential for boredom or frustration along dimensions of understanding and predictability?

For my part, I suspect that the frustration-boredom dimension needs to be deliberately balanced along the continua suggested by the post. Those continua can be applied to a variety of game mechanics – like combat systems or crafting.

Using combat systems as an example: we can gain something from looking at damage, for example, on the basis of how understandable it is. Will players be bored by a completely superficial system (A shot from a level 2 gun kills a level 1 mob with one shot every time) or frustrated by one that needs more effort to understand (With enough observation I see that a MOB wearing type A armor can take a shot to the leg with a Swerd Pistol and be immobilized – no matter what level they are)? For predictability, will players get bored by a completely deterministic system (Get a target lock, shoot, and the shot hits) or frustrated by a completely random system (pull the trigger and the shot has a chance of hitting no matter which direction you point it – an area-effect weapon).

I don’t for a minute want to minimize how difficult it is do decide where those lines are drawn or how. Moreover, where we make decisions for one game mechanic they may well change the balance for others. And the combinatorial effects rapidly become hugely complex. When that happens we are temped to “adjust” our ideal mechanisms with tweaks. Since those tweaks are by their nature situation specific their cumulative effect on the system become increasingly difficult to predict and any semblance of design rapidly deteriorates.

This suggests there is a tension between principled simulation and tying a design to goals for balance. Can we reconcile that tension? If not, is it the simulation that has failed or our statement of inherently conflicting goals? Does this run up against our ability as humans to handle contradicting realities and our inability as engineers to model it?

[Fitzhume]

6.

I agree that the strict monthly subscription business model may not be the sole income source for a MMOG business model, but I really doubt it will "go away" entirely. Monthly fees provide a very predictable and secure income for a business and are very valuable to have as a way to forecast expenses for the future. In this example, forecasting for staffing developers and support.

Personally I believe that MMOG companies are going to shy away from full dependence on subscription fees, but will always have them in some form to provide a secure income from operations. I predict $15 subscription fees are replaced by $5 fees, but then there may be a 2.99 forum subscription fee, an in-game auction site where the company may receive anther fee, and possibly even (I feel limited) in-game items offered from the company for real world dollars.

There are also glaring problems in the no subscription model as well. I agree that the revenue per customer may be more, but it may be at the expense of having a large customer base. That sense of "equality" that a standard monthly fee offers consumers has more value in my opinion than David Kennerly believes. You have to consider how many subscribers opted not to become a part of a MMOG because of the pay-for-perks model, where I firmly believe you won't find that many people who will not play a game because it has a subscription fee. (Obviously as long as it is within reason)

Alot of this comes down to opinions, but I feel that the pay-for-perks model clearly requires the developers to reward those who pay more than those who do not. And if that holds true, then unless you are marketing to a small audience of people who have significant disposable income, you will doom yourself to a model that does not maximize the potential playerbase, and an income that is extremely susceptible to drastic monthly fluctuations. How can you accurately predict future cash flows from this type of business model to make expense forecasts for the future?

Ultimately, I agree that you get more dollars per consumer with the free model, but there are significant trade-offs to holding strictly to that business model like Project Entropia. The higher gains require significantly higher risks. Blending these different types of models (subscription fees, forum fees, auction fees, and pay-for-perks) will be the future for MMOG business models in my opinion. In a sense, you are hedging the risk of the downsides to each type of business model while getting the benefits of each as well.

[Greg]

7.

"Die rolls" are not, in fact, about randomness; it's about variability of outcome. Utterly predictable outcomes are often boring--and in the real world, there are many unpredictable or uncontrollable factors, so that a degree of randomness is in fact a better simulation than perfect predictability.

Die rolls of course are not completely random--they are random within a range of predictable outcomes. And most games that rely heavily on die-rolls, such as board wargames, employ many frequent rolls, few if any of which individually have a major impact on the course of play. Consequently, there is regression to the mean--in multiple playings, only a few outliers will be heavily impacted by a string of unusually good or bad die-rolls.

[magicback (Frank)]

8.

I view the RNG issue from the Vegas perspective: many player-choosen choices of variability within an finite space. The 'Randomness without Replacement' system provides that variability within an finite space and players can chose which odds they want to take.

In essence, most people don't like bad luck, flukes, accidents, and other mishaps. So when designing a variability system I would prefer to shape the probability distribution to be fatty-tail on the positive side and flat-tail on the negative side.

Also, as part of the game design I prefer that designers allow players to put forth something extra to overcome string of bad luck. In many Heroic PnP games systems, players can earn Luck points (also known as Action, Fate, Divine points among other names) and use it to mitigate, but not overcome, bad odds.

So instead of full success with the use of the points, some of the mitigating actions are:
1) reroll
2) add bonuses to the role

Thus, given the players the ability to purchase or earn 5 non-trasferable action points per month (which is the character cap) can be both a gameplay and profit enhancing.

Frank

[Johnicholas Hines]

9.

"Perhaps what we really want is visibility into the breadth of the function of the world..."

Nate seems to be assuming that there is breadth and functionality underneath or beyond the RNG. I believe the primary function of randomness is to hide the actual mechanics.

Creating 'breadth and functionality' of physics is a nontrivial task for developers. When there are rich game mechanics, they are a pleasure to observe and manipulate. When the game physics are impoverished, can you blame the developers for hiding it?

[Bart Stewart]

10.

I also suspect that a person's reaction to randomness in gameplay will be filtered through his or her psychology.

So I found Franek's breakdown of satisfaction with game mechanics fascinating. In particular, it seems to me that the two axes of satisfaction:

1) the role of understanding in satisfaction (pattern)
2) the role of predictability in satisfaction

line up remarkably well with the two axes I identified for general personality types, and which I believe align with Richard's original four "Players Who Suit MUDs" types. (We discussed this back in Will The Real Explorer Please Stand Up?.)

I called my axes "internal vs. external" and "change vs. structure." At one end of the first axis is a concern for how things work, for what's inside a thing, for comprehension of behaviors, for identifying abstract principles. That sounds to me quite like "understanding." At the other end of this axis is a concern for using things, for manipulating things, for owning things, and for appreciating the tangible and sensory forms of things. In terms of randomness, this axis relates to variation in the outcomes of encounters.

My other axis of change vs. structure also seems pretty closely related to "predictability." On the "change" side I include preferences for freedom of action, for novelty, and for surprise. And on the other side I see preferences for security, for structure, and for stability. In terms of randomness, this axis relates to variation in the number and kinds of encounters.

Based on these two axes, I wound up seeing four quadrants of preferred behaviors -- in the context of MMOGs, four preferred styles of gameplay based on the player's innate motivations. If there's anything to this theory, maybe there's a way to use it to think about likely reactions to perceived randomness.

External Change: These players are looking for new sensations. They want to be free to experience anything and everything in the game world; they don't like any limits to what they can do. Understanding isn't necessary... but control is. These players don't mind variation in the number and kinds of of encounters (in fact, they demand it), but the outcomes of encounters had better be reliable. Anything (whether an object or another player) that can change internally to hide its true capabilities threatens the ability of these players to reliably manipulate that thing.

External Structure: These players want to define their world in terms of stable relationships to things. They care about accumulating more and better possessions, whether those are money, gear, kill stats, or group membership. These players would be the least likely to accept randomness in any aspect of the game world. Static outcomes in a well-defined space of potential encounters may lead to boring and repetitive gameplay, but that's less problematic than gameplay that can't be trusted to produce reliable results for effort expended. In short, the game world had better be comprehensible and consistent because the alternatives -- complexity and variability -- threaten the player's ability to possess and accumulate.

Internal Structure: These players are interested in identifying the rules of the world. Security is obtained through increasing knowledge of why the game world's features work the way they do, and of how the pieces fit together. They don't mind being recognized or compensated for their knowledge, but real satisfaction comes from having the knowledge, not from using it. For these players, random encounters are frustrating because they increase the difficulty of repeating experiments; a large space of knowable encounters is preferable. The outcomes of encounters, on the other hand, should be variable within a reasonable range -- in fact, both perfect randomness and perfect consistency of outcome are results too trivial to be interesting. "Somewhat surprising" is what these players want. They'll keep playing in order to ferret out what appear to be the complex (but comprehensible) rules of the universe.

Internal Change: These players are attracted to the new and different in people, and are interested in how the meanings of things change in different contexts. Ultimately their satisfaction is determined by how well the game world enables and promotes their personal growth. In a way, it doesn't matter to these players what the game world does. Whatever its form (random or not), it's just a space within which it's possible to meet people and form personal relationships with them. The world will be appreciated to the degree that it allows these players to see new reflections of themselves through finding and interacting with interesting people.

So: for encounters, social and PvP gameplay should offer some randomness in interactions, while PvE, economic, and crafting features should allow players to define the interactions they want to have. In terms of outcomes, combat and economic interactions should tend to have stable and non-random outcomes, while crafting and social features should allow some amount of surprise in their results.

Theories are wonderful things, aren't they? *g*

--Bart

[Matt Mihaly]

11.

Fitzhume wrote:

How can you accurately predict future cash flows from this type of business model to make expense forecasts for the future?

Experience. We've been running MMOs on a virtual item sales business model for a long time and it is not particularly difficult to make annual predictions. The monthly fluctuations are larger than in a subscription model, but annual revenue remains fairly predictable for us.

--matt

[Matt Mihaly]

12.

Just as an example, btw, revenue for 2005 is going to be within 3% of my prediction.

--matt

[Keybounce]

13.

> Obviously, the bigger game has more funding to do more development work. Yet, to the individual players, it appears that they are paying the exact same price for a different amount of development.

If this were true, then WoW, with LOTS of money, would have high development standards, and high customer satisfaction, while tiny little games like PuzzlePIrates (www.puzzlepirates.com) would be unloved, and people would complain about low developer response.

In fact, the opposite seems to be true -- while developers are commonly seen on the YPP forums, with developers reading everything, and responding frequently, the WoW forums are constantly lacking in developer feedback.

As much as the "more expensive game, with more subscribers, has more money" is concerned, it is accurate.

That doesn't mean that you get a better play environment. If you go for a bigger, more demanding world -- which may be essential to attract more people -- then you are also getting into more developer costs, and potentially a less satisfying game.

Perhaps the ultimate example/test will come when Vanguard comes out. It looks to be one of the most complex/comprehensive game environments, and may in turn have the hardest maintenence/support/developer need.

As a subscriber, I think a better thing to look at is "How much developer time and attention do I get for my subscription?". If a game has twice as many subscribers, and twice as many developers, then I'm getting the same relative developer attention. If the game has 1,000 times as many players, and a 50% more expensive subscription fee, but I'm only getting 20 times as many developers instead of 1500 times as many developers, then it will feel like I'm just not getting as much for my buck.

Incidently, I only stayed with WoW for the first month, while I've been on YPP for about a year and a half.

[nate combs]

14.

>> Yet, to the individual players, it appears that they are paying the exact same price for a different amount of development.

>> I think a better thing to look at is "How much developer time and attention do I get for my subscription?".

>> Obviously, the bigger game has more funding to do more development work. Yet, to the individual players, it appears that they are paying the exact same price for a different amount of development.

>>> I think a better thing to look at is "How much developer time and attention do I get for my subscription?".

Perhaps the reconciling theme here is accountability: if players understand what the pieces are and what they are paying for... all is well that ends well?