Why Choice Sucks: the Beauty of Random Generation

[xiombarg]

Well, I will note, with reasonably freeform chargen, there's nothing to stop you from having some general random tables/lifepaths, and then making the character in question based on what you roll up using the actual system. I believe the Central Casting series of books from Task Force Games was based on this idea, though there was some contraversy over the authors injecting some of their personal beliefs into the tables, such as listing homosexuality as a "Dark Side" trait.

[Ben Lehman]

Speaking of names, that brings up another type of character generation -- decided.  For instance, when I run D&D games, I often let the players simply decide what their attributes are.  No points.  No dice.  Just whatever they want.

Right. I'd like to note in passing that Pretender does this, and quite successfully, I like to think.

BL>  Absolutely.  Pretender is totally ground-breaking in this regard.  For people interested in how purely decision based chargen looks in a non-freeform context, I urge you to check out this game.

yrs--
--Ben

[M. J. Young]

Good post, Raven. I did some articles on character generation for Game Ideas Unlimited last year, and random generation got strong marks from me for a lot of the reasons you posit here.

Bankhead's comments about the negative aspects are worth considering.

[*]Random generation creates unequal characters which are detrimental to gamist play.[*]Players often don't want or like the character they wind up rolling.[*]Some characters created are wildly inappropriate or ineffective.[/list:u]All of this is not the fault of randomized generation itself, but of flaws in early designs.

For example, D&D began with the idea of all characters being created by rolling 3d6 for each of six scores. This created characters who tended toward a defined average. It overlooked the fact that player characters were, presumably, already the outstanding members of their communities, and should have at least some above average scores in specific areas. In BD&D1, if I recall correctly, this problem was addressed by having the referee review the characters rolled and call for new characters whenever any was likely to be unsurvivable. Of course, this assumed that the referee had sufficient experience to make that call. OAD&D addressed the problem with four Methods of creating characters:[list=1][*]Roll four dice for each score and keep the best three.[*]Roll twice as many scores as you need and keep the best six.[*]Roll each score six times and keep the best roll.[*]Create twelve characters and keep the best one.[/list:o]Each of these was designed to provide an above average character of some sort, with different advantages and disadvantages to each method. With Unearthed Arcana, though, a considerably better approach to player character generation was devised, known as Method V. This approach allowed you to select your character's profession (class) first, and then gave you support for rolling that type of character. You would keep the best three of nine dice rolled for that ability score that was most important to that type of character, of eight for the second most important score, and sequentially down to a straight three dice for the score least important to that character type. This assured that your important scores would almost certainly have strongly above average values, while the unimportant scores had the normal odds of being good.

Devising a system in which the random generation is geared to produce appropriate characters answers all Bankhead's objections, while at the same time enhancing the very advantages Raven sees in randomness. Further, as the OAD&D example shows despite its age, choice can be incorporated into the randomness such that a player doesn't roll "a random character of some sort" but rather "a random character who is likely to be a superior example of a specified type"--exactly the goal Raven observes point based systems do not achieve for first time players who do not know how to create that character from the resources provided.

I should also put in a positive word for point-based systems, as well. One complaint about these is that in reality all men are not created equal, and building even a perfect system in which your strengths and weaknesses genuinely balance each other is unrealistic. A randomized method of determining available points for the character overcomes this, forcing one player to work with perhaps three quarters of the "normal" point total while another can have an extra twenty-five percent above "normal". (Of course, this should be "normal" for the unusually talented sort of character that is a player character type, however that works for this particular game.) It would also be feasible to let a player spend his points to buy random bonuses. For example, an adaptation to a D&D system would allow that the player can roll 3d6 for any ability score free, but by spending points on that score he can add a die to the roll and take the best three. Similarly, he could roll 3d6 and then spend points for the right to pick up any one of those dice and roll it again, stipulated that he doesn't get to keep the old number if the new number is worse. He could spend a lot of points to get an 18 out of that, but he is more likely to settle when at least two dice are at least five and the third is not less than four, because at that point it's not cost effective to push it.

The point, I think, is to develop a character generation system that combines some amount of random development with some amount of choice. Some lifepaths, some choice systems, some dice-based, and some point-based systems do this effectively.

--M. J. Young

[jerry]

Lastly, and what always annoyed me the most, some systems randomly determine factors that aren't actually used in the game, but help to define the character.  E.g., AD&D had random generation of character height, weight, and age.  Original Top Secret randomly determined height, age, and whether or not the character needed glasses/contacts.

Men & Supermen also allows for random generation of blood type, a feature that is not used anywhere else in the game rules.

What I went with in M&S was complete player choice--no points at all--vetted by the rest of the players and GM. But if you don't know what you want, there is always a random roll to help you choose.

When I first started writing it, I tried to create a more choice-oriented in-rules approach. It ended up becoming too much a part of the game, as others have noted about other point-based systems. Once I decided to chuck the whole thing and say "you can choose whatever you want, as long as the other players agree, and if you don't want to choose, you can roll randomly", I haven't seriously looked back.

Jerry

[Sydney Freedberg]

The most satisfying single character I ever played was generated by me rolling dice for various attributes and the GM (a Brit named Alden Speiss) giving me back totally unrelated values for those attributes -- to the extent that the stat I said was most important didn't end up being my highest one. This was, in theory, AD&D.

At the same time, I had come in with a very clear character concept (elf who hates the whole elven superiority schtick, is in rebellion against her Elf Lord father, and cut off the points of her own ears) that fit perfectly with GM's campaign (rogues on the run from a dictatorship ruled by racist elves). So this wasn't an example of a random creation method challenging me to come up with a character I wouldn't have thought of -- although being tossed a 17 Charisma which I had neither rolled nor asked for was in the "pleasant surprise" category and may've changed how I played the character somewhat.

I suspect this was actually a variant of "just pick your number" (i.e. Drama) where the GM did most of the picking in accordance with (a) his understanding of my individual character concept and (b) what was needed to balance out the group as a whole. It worked, astoundingly well, but I'd say, "don't try this at home."

[John Kim]

Bankhead's comments about the negative aspects are worth considering.

[*]Random generation creates unequal characters which are detrimental to gamist play.[*]Players often don't want or like the character they wind up rolling.[*]Some characters created are wildly inappropriate or ineffective.[/list:u]All of this is not the fault of randomized generation itself, but of flaws in early designs.

While varant mechanics can mitigate some of these, I think that they are central to many people's objections.  

Unequal Characters

That random generation is bad for all Gamism, I think that is wrong.  Gamist play can and does accept randomness -- and randomness in generation is no different in principle than randomness anywhere else in the game.  Rolling a low Constitution at the start is no different than, say, losing 4 points of Constitution from a missed poison saving throw.  Players can distinguish poor rolls from poor play, and they can give greater respect to someone who succeeds despite poor rolls.  Randomness isn't everyone's cup of tea -- i.e. some people prefer chess to poker -- but it isn't inherently anti-Gamist.  

On the other hand, I think it is true that many types of Gamist play don't want these sorts of random long-term effects.  For many, some randomness is OK, but it should have short-term effects and quickly average out from a large number of rolls.  If this is so, then you probably want to minimize or eliminate randomness in generation.  

With the "whiff" factor, there are really two issues here:  the average and the variation in character effectiveness.  If the average is low, that's no different from a design system which gives people only enough points to be ordinary.  For example, you could have a GURPS game which gives 15 points instead of 100.  If you want the PCs to be more heroic, then the random generation has to be weighted to be at the upper end.  I guess a problem with early generation systems is that they were designed to have the average be an average person in the world -- PC or NPC -- and the weighting was an afterthought rather than a part of the design.  But that's really the same issue as, say, D&D characters starting at first level (say).  It's not an issue with random generation overall.  

The variation is a different issue.  A high variation in power level means that you can have parties whose PCs vary highly in effectiveness.  Again, it's a question of what you want.  For some people, always having a group of exactly equivalent peers is a flaw.  It's more interesting to have heroes and sidekicks, or crack agents and comic relief.  

Disliking Rolled Characters

Random generation is going to give you different results than if you designed a character exactly as you wanted it.  That's really the point -- i.e. attempts to put more choice in the process are simply making the system less random.  It doesn't justify the randomness in the first place.  If you have a concept already formed and want to create exactly that character, then random generation is the wrong tool to use.  Random generation should be used to give you something different than what you would have come up with on your own.  

That said, I think random generation can be fun and interesting.  I don't usually use it, but I did use it for the HarnMaster game I last played, and for an upcoming campaign.  For the last one I just used it straight.  The upcoming campaign that I will be playing in is a little higher power.  The GM's initial suggestion put in some rerolls at particular points.  I preferred instead to do a complete random roll, but to be more on-par, I rolled three complete characters to pick from.  The link below shows my rolling results...
http://www.darkshire.net/~jhkim/rpg/harn/city/pcs.txt

It's not for everyone, but I like my prior character (Baraud) and my upcoming character (#2 in that list, Girard).  

PC Groups

That said, I think there is a fair point that a strongly random system can have problems because the randomly-generated PCs won't form a coherent group.  Not only might they have conflicting niches for skills/powers, but they may have backgrounds or other traits which make them unlikely to work together.  This can be easier if the random system is designed to create characters in a certain range -- i.e. if you make secret agents, they can be assigned together.  Alternately, the campaign might be about characters who are together by accident rather than by choice.  Still, it's a real issue.  This is generally dealt with by reducing the amount of randomness in games, such that the choices can outweigh them.  Other approaches (like weighted rolls) should be possible, but I haven't seen them.  

The point, I think, is to develop a character generation system that combines some amount of random development with some amount of choice. Some lifepaths, some choice systems, some dice-based, and some point-based systems do this effectively.

I don't think there is any "ideal" method here.  Different approaches will appeal differently.  All generation will have some choice involved.  But I don't think there is any particular level of choice or area of choice which is "best".

[simon_hibbs]

Interesting topic, but a lot of the problems that are being bandied about that random or non-random systems 'solve' are realy fundamentaly down to bad system design, and aren't anything particularly to do with randomness or determinism per se. From the orriginal post:

Rather, having not played TROS before, I did not know what I should be placed where, how the various scores I could place would function in play, and thus which scores would best serve my purposes in which particular categories.

IMHO this is clearly a fault in TROS. After all, surely in this situation you can emulate a random system by just assigning arbitrary values. Your ignorance of the significance of those values equates to a random system - problem solved!

Consider that if I want to play a particular type of character (which is the usual argument for the use of choice-based character generation) I need to place my points "correctly", ...

If you want to play a particular character type, then surely what you want is a transparent set of rules that allow you to do that. Randomness can do nothing to help a player with this goal to achieve it. See below*

However, I fail to see any compelling reason why one would, by default, be less involved in the playing out of a character over whom they had little or even no initial design input. Actors do this constantly --

This is a realy old argument, I've seen it a hundred times and it's complete rubbish. Casting directors do not assign roles to actors randomly by lot, they chose the actor to fit the role whenever they can.

In many ways, that can be the best sort of challenge, and the most rewarding.

Well yes, if that's the kind of challenge you like. Obviously you do, other's don't. My take on all of this is that that's what it comes down to - personal preference.

[/quote]This also has the added benefit of making each new character a new challenge.[/quote]

If you want a new challenge, why not deliberately design a new kind of character? Why force people into a particular kind of challenge becase it's "better".

The system might be known after the first character, and with point/choice-based character creation, you know how to work the system to produce the various effects you desire, and often to the best effect.

Well blow me. Points based systems let players design character that are realy effective at what they want to be effective at. This is a flaw how?

Random character generation also forces an individual to play a role, a role they didn't choose, so they have to make it work and make it their own.

The operative word being "Forces". Why do you want to force people to play the way you want? See below again*

This also puts a stop to those (honestly annoying) players who play the same, eventually tiresome character in every game, such as the guy who only plays dwarves who all act alike, or the girl who only plays the bitchy, snooty paladin.

*See here!

Now we're getting to the bottom of it. You want everyone to roll characters randomly, because you don't like the choices other people make when they're given the chance. Your previous rhetoric ("Suppose I want to create a certain kind of character") is smoke and mirrors. You don't want anyone to be able to genereate a certain kind of character because you don't like the choices they make, and therefore any system that give other people a choice must be faulty. What on earth gives you the right to trash other poeple's preffered mode of fun in this way?

In such cases, randomized generation can be quite liberating, as one is not constrained to pre-conceived notions of "what the character should be like" during the design process, one simply enters play and discovers what the character they have recieved actually is like, unbound by the preconceptions that would have been necessary in the character's original design choices.

Again, nothing to do with randomness. HeroQuest solves this by letting you literaly design the character during play in response to events and demands that occur in the game. No randomness required. Again it's down to apropriate game design, not randomness at all.

Some individuals may be selfishly inclined towards complete control of their avatar, to the detriment of their own enjoyment;

So we must force them to have more fun your way [draws gun, shoots ground betyween the other player's feet]. Pow! Pow! Dance you fools! Dance!

What gives you the right or authority to tell other people that your brand of fun is better than theirs? It looks like some players are into selfishly denying other players any choice over their type of character to suit their own agenda.

IMHO good game design can resolve all these issues. HeroQuest is entirely points-buy, but suffers none of the system-based problems mentioned. Characters are as effective as you want them to be because any ability is by definition the best possible ability at what it's for. If you're not sure how to develop the character, you can reserve some or all your abilities and define what they are during play as you prefer, assigning values from your pool of unspent points. Characters can gain new abilities and flaws during play, allowing considerable character growth and adaptation during the course of a game. All contests use the same rules, so there are no hidden nooks and crannies in the rules that can give experienced players unreasonable advantages.

Perhaps I overstate my case a little, I'd be lying if I said the HQ garden was all a bed of perfect roses, but it's clean core design does show that many of the problems raised here arrise from game mechanical issues that have nothing to do with randomness or determinism.


Simon Hibbs

[John Kim]

This also puts a stop to those (honestly annoying) players who play the same, eventually tiresome character in every game, such as the guy who only plays dwarves who all act alike, or the girl who only plays the bitchy, snooty paladin.

Now we're getting to the bottom of it. You want everyone to roll characters randomly, because you don't like the choices other people make when they're given the chance. Your previous rhetoric ("Suppose I want to create a certain kind of character") is smoke and mirrors. You don't want anyone to be able to genereate a certain kind of character because you don't like the choices they make, and therefore any system that give other people a choice must be faulty. What on earth gives you the right to trash other poeple's preffered mode of fun in this way?

First of all, quoting line-by-line and picking apart each one is against Forge etiquette.  I'm certain that greyorm isn't advocating putting guns to people's heads to force them to use random generation.  People always have and always will use systems by choice.  

But to address the point, I don't think that forcing restrictions is the same thing as trashing people's preferences.  Following this logic means that all rules are bad, because they prevent people from doing what they really want to.  The flaw in this is that people can and often do find that they enjoy playing under restrictions even if those restrictions make them play in ways they hadn't originally conceived or would have come up with on their own.

[M. J. Young]

I find it curious that Raven (of all people) has not responded to any of the comments generated on this thread which he initiated.

Are you still reading, Raven? Are we talking about your subject?

--M. J. Young

[Bankuei]

Hi Raven,

I think there are uses for any type of character creation- part of the question is what kinds of character generation work for what purpose?  That is, does it fit?   With that, there are many ways random generation can be either useful, or counterproductive, depending on what you're trying to do.

That all said, another solution to the problem you've mentioned about not being familar enough to proficiently design characters, is to have archetypes of mostly or completely defined characters.   Feng Shui is the first game that comes to mind.  Want to do Kung-fu?  Pick the Martial Artist.  Want to do magic?  Pick the Sorcerer.   Classes typically are similar ideas, but most games layer it with tons of other options, making it a complex affair.

I think the common aspect between "mostly done" character archetypes and random creation is that the players aren't being asked to make a lot of decisions.  I find a lot of players are at a loss when presented with a system that allows them to completely define their own traits, such as Over the Edge.  It's easier when folks have a list to choose from.  It gives them directions and ideas.  Likewise, with random generation, a lot of the dice make the decisions.

Thoughts?

Chris

[greyorm]

I have to apologize to everyone, I have been reading the thread, and trying to get a response together in my limited spare time -- family and school have been eating up more than their regular share of time this past week.

The baby ate the lengthy response I had written on Sunday (ie: I wrote it up, and then she hit the "Power" button...{imagine me weeping}), but I will be putting a new response together late tonight or tomorrow evening sometime.

So, again, my apologies for my lack of input. I would like to thank everyone for participating thus far, even without my feedback! I am glad to see the topic has generated much good discussion.

Thanks for your patience!

[simon_hibbs]

JKohn's probably right, perhaps I could have been more terse and polite in my post.

But to address the point, I don't think that forcing restrictions is the same thing as trashing people's preferences.  Following this logic means that all rules are bad, because they prevent people from doing what they really want to.  The flaw in this is that people can and often do find that they enjoy playing under restrictions even if those restrictions make them play in ways they hadn't originally conceived or would have come up with on their own.

But that's not the argument that was offered, the argument was that the choices other players make are bad, so therefore it's best not to let them choose even if (actualy especialy if) they want to. That's quite different from saying that rules structure in the pursuit of creative agenda is good, which I wouldn't deny. The offered argument isn't aimed at any particular creative agenda, it's aimed at restricting the creative agenda of others because they choose bad creative agenda. That's a pretty brutal judgement call.

Now ok, maybe I personaly don't have a whole lot of personal time for people who always choose to play transgendered dwarven ninjas or whatever in every goddamn game, year after year, but at the end of the day who am I to deny them their fun? Nobody's forcing me to play with them, after all.

Simon Hibbs

[John Kim]

But to address the point, I don't think that forcing restrictions is the same thing as trashing people's preferences.  Following this logic means that all rules are bad, because they prevent people from doing what they really want to.  The flaw in this is that people can and often do find that they enjoy playing under restrictions even if those restrictions make them play in ways they hadn't originally conceived or would have come up with on their own.

But that's not the argument that was offered, the argument was that the choices other players make are bad, so therefore it's best not to let them choose even if (actualy especialy if) they want to. That's quite different from saying that rules structure in the pursuit of creative agenda is good, which I wouldn't deny. The offered argument isn't aimed at any particular creative agenda, it's aimed at restricting the creative agenda of others because they choose bad creative agenda. That's a pretty brutal judgement call.

Now ok, maybe I personaly don't have a whole lot of personal time for people who always choose to play transgendered dwarven ninjas or whatever in every goddamn game, year after year, but at the end of the day who am I to deny them their fun? Nobody's forcing me to play with them, after all.

But I think this is the point I was addressing.  Let me take a more exact parallel:  let's compare randomized conflict resolution and randomized character generation.  The same logic seems to apply.  What if a player wants to succeed in the conflict?  Why shouldn't they be allowed to get what they want, instead of leaving it up to a die roll?  Instead, many games force down their throats that they don't get to succeed.  

This argument is false, because it presupposes that players don't like having choice taken away.  Having rules is always having choices taken away.  Despite this, many people enjoy playing within restrictions, because it gives focus and challenge to the activity.  For example, with conflict resolution, many players enjoy having failures forced on them even if they wanted to succeed.  

Yes, a game can work if the players choose their successes and failures rather than rolling for them.  But that doesn't mean that using rolls for resolution is an insulting judgement of the players.

[greyorm]

Heya all,

Just had time to type up some quick responses to various statements and issues that have been raised; I have tried to cover everything I felt immediately relevant to the direction of the conversation. Also, a great deal of what I feel has been said already by Mike, John, and others, so I won't be going into great detail where they have already.

I did want to reinforce one of the issues John touched on, one of the main oppositions voiced regarding randomization (that of the preservation of game balance). Though he has already adaquately covered it, let me say I am in full agreement with him, as some may recall that such is even my battlecry: "Game balance is a red herring!"

Everyone should immediately note and adjust their understanding of Gamism, for as noted prior, Gamist play is not necessarily impacted by a difference in power levels between characters. The only real impact such differences have is in the Step On Up for each individual player, whether they have a more difficult or easier time in relation to one another is not an issue of the Gamist mode, or even of "fairness" (necessarily -- the word is far too vague to use as a defense, or to object to).

As such, a D&D character with a 5 Strength is not an impediment to Gamist (or tactical) play. A 5 STR has far more to do with expectation and desire than fairness or tactics, for which it simply changes the tactics required in play of that player.

As to realism and starship captains: it seems obvious to me that if a game is about starship captains, then the randomizing system would take that into account in the first place, and not generate anything except realistic starship captains (even if they are random). In fact, a choice-based system had better take this factor of play into account as well, and also not allow choices that would produce non-starship captains.

Regarding actors and characters...with rare exception, no, actors do not choose their roles. They are given to them, thus I fail to see how this is "oft repeated rubbish." Yes, directors may choose actors, but who was talking about directors? Unfortunately, such mistakes are characteristic of your response in this instance, Simon (hibbs), so rather than spend more time uselessly refuting arguments I had never made, I am going to invite you to reread my post without predjudice.

You do make one good point I want to respond to, however. That in my example I could produce randomness by simply assigning the points I am given willy-nilly, without knowledge of what effect they will produce. I believe the hurdle here is more psychological than systemic: that being the idea that with a presented choice-based generation methodology, the player will go into "Must...create...character!" mode, ignoring completely the idea that he could just slap things together.

Related to this, Iron asks what I mean by "expected effectiveness"...well, just what it says! What I (the player) expect the character will be effective at in the game -- what I expect him to be like, be able to do, and behave as. To return to the TROS character example, if I create a character who I describe to be a fencing master then I expect that he will win fencing matches against the majority of his opponents, know the uses and care of his weapons and armor, be acquainted with various other fencers and organizations. Even if I can pick the (supposedly) "correct" skills to represent all this during generation, as I do not know what the scores will actually produce in play I cannot truly say any of the above is true, and the character may very well end up being only a moderately good fencer, rather than a master.

This also came up in our D&D game: the player expected his elven warrior to be good at melee combat and archery, as that is the way he built the character. This has not happened yet, and has become something of a running joke (if still frustrating for the player) that his expectations about the character are not being met, that the framework he built the character around is incongruent with his in-game experience of the same character. Simply, "He's been a warrior all his life, and suddenly he goes adventuring and can't hit the broad side of a barn from two paces!"

This is what is meant by "expected effectiveness": that the character will perform, in general, as the player expected them to prior to and during creation.

Ultimately, random generation is not the only way to solve this problem (or any other), but it is one way to do so, by removing "expectation" from the equasion. As Chris notes, another method is to utilize character archetypes players can choose (though in this case, they also choose the expectations!).

Also, a couple of folks have noted that no one has said anything about "degrees of randomness" thus far, but you'll note that very thing was one of the points made towards the end of my post. One of the main points regarding random generation versus choice is not that one should have no choice, but that many players have been crippled by choice, for example consistently choosing the same sorts of character, for reasons of effectiveness or for reasons of ego, ultimately using system at generation to tip the game's odds in their own favor or playing the same character over-and-over, in either case stunting their own creativity and experience in the process.

Finally, even considering the strong language used in my original post, I find that I am fascinated by some of what I see as kneejerk reactions to the argument posted. For all the GNS-inspired play and self-as-gamer enlightenment encouraged on the Forge, it seems that we haven't gotten very far away from defensive -- if not reactionary -- stances about other sacred cows of gaming?

I'll post more later about the various ideas for randomization in generation that support the above benefits, and/or problems caused by current systems that utilize it (and how those might be mitigated?), but in the meanwhile, if anyone has any more input along those lines, feel free to add to what has been started by MJ and others.

[inky]

Everyone should immediately note and adjust their understanding of Gamism, for as noted prior, Gamist play is not necessarily impacted by a difference in power levels between characters. The only real impact such differences have is in the Step On Up for each individual player, whether they have a more difficult or easier time in relation to one another is not an issue of the Gamist mode

I don't think this is a realistic analysis of most Gamist play. I think the majority of the time people do end up comparing their progress against each other -- "Hey, I killed three zombies in the time it took Bob to kill two" in addition to comparing against their own past performance. And, right, the former case is where differing power levels pose a problem -- if I'm fighting goblins and you're fighting dragons, it's not really clear if you taking out two with a single blow is better, worse, or the same as me doing the same. And it's even more complicated if we end up with different results -- if you do pretty well against the dragon but ultimately have to run away, and I duke it out with the goblin guards and kill their leader, who stepped further up?

As to realism and starship captains: it seems obvious to me that if a game is about starship captains, then the randomizing system would take that into account in the first place, and not generate anything except realistic starship captains (even if they are random). In fact, a choice-based system had better take this factor of play into account as well, and also not allow choices that would produce non-starship captains.

The difference is that if somebody picks a non-starship captain, presumably they're happy with doing so, whereas if they get randomly assigned one, they may not be in the mood for something so unusual. It seems like if you want a random generation system to produce reasonable results, it inevitably has to get rid of the more extreme possibilities, even though certain combinations of those might be fun to play, especially if you knew what you were getting into.

This also came up in our D&D game: the player expected his elven warrior to be good at melee combat and archery, as that is the way he built the character. This has not happened yet, and has become something of a running joke (if still frustrating for the player) that his expectations about the character are not being met, that the framework he built the character around is incongruent with his in-game experience of the same character.

I can't quite tell what the problem is for your player. Did he design his character poorly? In that case, it seems like the issue is more about learning the system (and/or making the system more transparent during character design). Or is it just that he's rolling poorly? In that case it seems like this is more a flaw of the system than the character design process -- whether I choose a character who's an archer with 18 dex or randomly roll one, it'll be incongruous regardless if later I can't hit anything with a bow.

Finally, even considering the strong language used in my original post, I find that I am fascinated by some of what I see as kneejerk reactions to the argument posted. For all the GNS-inspired play and self-as-gamer enlightenment encouraged on the Forge, it seems that we haven't gotten very far away from defensive -- if not reactionary -- stances about other sacred cows of gaming?

Hmm, I assume your point here is that everyone loves point-based characterization? Personally, the reason why I had a negative reaction to your initial post wasn't so much because I don't like randomness as because most of my rpg experience has had far too much randomness in it already. Obviously everyone's history differs, but for me a lot of it's been rolling up a d&d character with so-so stats, then going into combat and getting so-so die rolls, and then making a couple skill checks and getting so-so results. And, enh.

I guess it's that I don't feel so burnt out on any system that I'm out of ideas for it, so I don't feel the need for random rolls for inspiration, and I worry that adding randomness inevitably means deleting options to make the randomness "come out right".