Baldur's Gate 3 - Swen on Dice and Druids

Destructoid talked to Swen Vincke about Baldur's Gate 3 and how development is going.

A modern method

Click to expand...

What's helping Vincke and the team at Larian track everything is not just vocal player feedback, but the data it's collecting from playtime. An upside of Early Access is that you have a lot of players engaging with your content, and while that can help for ferreting out bugs, there are deeper problems you can root out with enough data.

It might sound strange, but story has been one area where the data has been a boon. At one point during our talk, Vincke pulls up an image to show me what's essentially a graph of the paths players have taken through a specific story dialogue, and what outcomes they received. Vincke estimated it covered about 30,000 dialogues' worth of data, with some results being clear majority and others, ones that might require use of certain mechanics or knowledge, having fewer players reach that conclusion.

Vincke describes these analytics tools, which the studio built up after the success of Divinity: Original Sin 2, as a "gold mine." It helps the studio identify what players are doing and make inferences, and potentially adapt if they're seeing a choice being taken too often or not often enough; or even, if the choice is one they want to feel special and earned, keep it that way.

"I call it a very modern way of doing game development, because you finally have the data processing to be able to do these kinds of things," Vincke says. "And it's, it improves it, literally, I mean, there's nothing to be done about it. It makes the games better. They just need time, they need to cook."

[...]

Click to expand...

More information.

Disturbing article. It really pictures Larian as a company who doesn't know where it's headed, and who may change direction depending how the wind blows.

From a company with such a vast experience, statements like "So we still have quite some work ahead of us," Vincke says. "But we do know what we're making and when we're going to be… we'll know when we're at the end, so it is visible. It's just still a lot of work." are just not right. If they had a well-defined target, the impossibility to estimate when they get there is BS, there are many techniques in Agile development (or waterfall if they're still using that).

Of course, if they don't know where they are going and if they change to please a fan base which was split from the start between D:OS and D&D fans, which seems to be the case from the rest of the article and past communications... that's something else. A company has a budget, employs developers on a payroll, and has close competitors. If they are unable to set a goal and stick to it, it's likely they won't survive, especially in game development where it's too easy to add yet another quest or tune another set of dice.

Using analytics is a great idea, however. I'm skeptical about its use in dialogs that have already been written and expensively recorded though, maybe they're using that as an example for the part that had yet to be implemented? They've also shown how they used it to detect real game problems, and that's very smart.

What I don't understand is how they can use those advanced tools, but nothing to collect and track the bugs. Perhaps they have "an army of QA testers", but they had a massive opportunity to get bug reports and suggestions, yet all the EA users had was defective forums with zero feedback from Larian and no way to know whether the reports had been seen or not. It must be a nightmare for them to make sense of those repetitive and contradictory threads - and I'd know, I've been in a similar situation. I much prefer the in-game report system of Owlcat Games or Black Crystal Games, that's the professional way to do, even if they also lack a feedback system (a simple read access to their bug tracker would be enough).

Redglyph said:

If they had a well-defined target, the impossibility to estimate when they get there is BS, there are many techniques in Agile development (or waterfall if they're still using that).

Click to expand...

I'm a software developer with 13 years of experience and estimates only work in the very narrow situation you will not encounter very often: if you are working alone, the concepts are finished, and there are zero external dependencies.
I think especially game developemnt does almost never work this way.
The bigger the project, the more people are involved, the less estimates will work - especially for longer time frames. In modern software development you can only estimate vaguely within a few weeks.

In software development there is this triangle concept of time/deadline, quality and resources where only one corner is fixed and the others need to be flexible. And Larian decided that quality is fixed and they will adapt time/deadline and resources - I don't see any problem with this.

I don't think the article/interview is disturbing at all - it is honest and does not contain the typical game development marketing BS. Their goal is very clear: They want to make the best game possible, and they don't have the illusion that you can outline and predict month and years ahead. They will adapt their way to their current circumstances.
Working for such a company would be very great I think!

About every game develops pretty much the way BG3 is doing, and changes in wild extremes from what its original concept was, to what it ends up being. Especially those with a larger budget/team. Most people here probably know that Overwatch was designed to be a mmorpg with character creation, quests, dungeons, etc, and ended up being a team PvP shooter, for example.

The mistake Vincke is doing, perhaps, is being too transparent about it, and possibly having so much data at his disposal, he may be obsessing over it. But the truth us that developers have no freaking idea of how their game will end up playing. They all have a concept and a vision, but this vision mutates during development, and the longer development takes, the more it mutates. This is why a lot of developers prefer to say they will release the game "when it's ready" instead of giving a more concrete release date. "When it's ready" = when the final "mutation" of the product has been reached and appears satisfactory enough.

So the core of the game and the concept might even remain in a similar place to how it was conceived, but all the cogs and pieces will shift dramatically. All systems, minigames, the way loot works, even the plot might shift dramatically, and that is generally just a consequence of subjecting a product to its actual development as a product.

Kylex said:

I'm a software developer with 13 years of experience and estimates only work in the very narrow situation you will not encounter very often: if you are working alone, the concepts are finished, and there are zero external dependencies.
I think especially game developemnt does almost never work this way.
The bigger the project, the more people are involved, the less estimates will work - especially for longer time frames. In modern software development you can only estimate vaguely within a few weeks.

In software development there is this triangle concept of time/deadline, quality and resources where only one corner is fixed and the others need to be flexible. And Larian decided that quality is fixed and they will adapt time/deadline and resources - I don't see any problem with this.

Click to expand...

I agree that estimations are not precise, I've rarely seen a project that met a deadline in, well, more than 20 years in development and project management in software, hardware, or much bigger projects involving many high-tech and more classic components.

But estimations still give a pretty good idea, and they help make decisions and choices. They also help keep focused on a target, and it's very important. At the end of the day, you still have to meet some deadline because it's a business, their income will have to balance the resources, and ideally leave enough benefit. That's the fixed point, I don't think anybody has the luxury of saying "oh well, we'll see when we get there", except small or secondary projects by people who have another source of income.

Maybe it's more flexible in game development because of the margin, and you're right to point that out. Although we see how bad crunches get. Aiming at quality as target is a naive approach, the one a developer would always defend - and I've done it too, but as project management goes, you quickly understand it's not viable when there's competition and money at stake.

But it's not only that. Larian gives the general impression of being undecided and trying to comply to a few vocal fans, instead of maintaining their vision. Loaded dice and less pronounced companion personality are two sticking examples. If they had to listen, perhaps it should have been at the beginning, before the initial decision of using the name "Baldur's Gate" with a modified D&D 5E, turn-based mode and a Larian style, which doesn't bother me at all but irritates a lot of D:OS fans, and even more D&D fans.

I think games are a bit of a special case in the world of development. It combines all the challenges of software development, with the artistic process of creating entertainment.

When you're making a movie, for instance, there's a great deal of assessment and adaptation that is done on the fly - this scene isn't working, let's do a quick rewrite, or try some improvisation, or a different camera setup. I think when you're doing that in the context of software, it's an unholy nightmare. You're not just trying to make software that works, but entertainment that "works", too. And if a change is needed in a game, that often pulls on a thread that unravels something somewhere else. I think it's a much bigger planning challenge.

One of the things that occurs to me with RPGs, is that you could create a very stripped down version of the game first, in its entirety. So a very basic graphical engine, using tokens like Knights of the Chalice, that you use to layout the whole game - all the planned content, the mechanics, etc. That would let you playtest and work out a lot of the fundamental issues, with minimal work to change things. Then adapt it to use a more modern graphical engine. There's obviously some wasted effort, but I wonder if it might be well worth it for the efficiency gains - you could really hit the ground running when you ramp up to the expensive part of development.

"Swen IS on The Dice With the Druids" this is just a mere technicality, when to me Swen spends his time like this:

I don't think I want a game designed by the fickle crowd. That sounds horrible. It sounds like Larian may be going badly wrong. This is not making me eager or wanting to have anticipation for future games if this is the way Larian thinks now. I think it will be very ironic if Larian starts to fall apart over one of his "dream" projects - BG 3. He needs to read some of the very good critiques and critical reviews on steam and figure out what they are doing wrong, but probably too late for that, to be honest.

In any case, BG 3 lost me and my interest a long time ago. Horrible decisions all round.
Hopefully Larian will wake up and move on with their next game, sooner rather than later.

This article doesn't ramp up my expectations for this project, to be honest. Then again, I've seen diminished returns from this company in the past three offerings, so I shouldn't be surprised. When you try to appeal to a bigger audience, you tend to put at risk perhaps the people that were amongst your most loyal fans, to whom at one time you could literally do nothing wrong. How the wheel has revolved.

Carnifex said:

This article doesn't ramp up my expectations for this project, to be honest. Then again, I've seen diminished returns from this company in the past three offerings, so I shouldn't be surprised. When you try to appeal to a bigger audience, you tend to put at risk perhaps the people that were amongst your most loyal fans, to whom at one time you could literally do nothing wrong. How the wheel has revolved.

Click to expand...

Well, the game is trying to be loyal to DnD, and this is literally what DnD has been doing for the last two decades. Appeal to a bigger audience at the potential discontent of old-school loyal fans, so no surprise there.

As everything in life, it is adapt or die.

Nereida said:

Well, the game is trying to be loyal to DnD, and this is literally what DnD has been doing for the last two decades. Appeal to a bigger audience at the potential discontent of old-school loyal fans, so no surprise there.

As everything in life, it is adapt or die.

Click to expand...

Damn straight I hate being an old man and seeing this happen though.

Mass appeal in gaming is a cancer, and the social media mob is the driver.

As someone who has had an active Pathfinder (which is probably closer to the Baldur's Gate D&D than 5th edition is) tabletop campaign for the last 8+ years, I have to say that I don't find the idea of loaded dice that offensive. There is a concept in Pathfinder of "taking a 10" for a skill check instead of rolling. Also when a character fails a skill check in tabletop, the DM can improvise, which is something thats not so easy to do in a computer game.

forgottenlor said:

As someone who has had an active Pathfinder (which is probably closer to the Baldur's Gate D&D than 5th edition is) tabletop campaign for the last 8+ years, I have to say that I don't find the idea of loaded dice that offensive. There is a concept in Pathfinder of "taking a 10" for a skill check instead of rolling. Also when a character fails a skill check in tabletop, the DM can improvise, which is something thats not so easy to do in a computer game.

Click to expand...

That's interesting Yes, a human DM is hard to replace...

I think there are different issues with the dice.

One that is annoying many people is the use of randomness in dialogs, especially when it has an impact on the story. Larian already addressed that with the (limited) possibility to throw the dice again.

Another is when a roll doesn't bring anything to the game, for example trying to bash a door when there is no time constraint. You know you'll succeed eventually, but you need to attack the door many times with random results. It's pointless and the DM may say "OK, let's pretend it's done" and skip the rolls. Maybe that's your example?

Then there's the feeling by some that they are cursed, and that they always have low rolls while the enemy has good rolls. That's what Larian tried to "fix" here. Personally, instead of loading the dice, I would just keep track of the stats to show them the average value, so they can check if the average is 10.5

Solasta has a new alternative method too, it's a difficulty option.

Yeah as far as I know take10 is usually only utilised in situations when the skill check can be done in calm, without pressure or time constrains. It can be used for things that are less trivial than bashing a door, like for example, if you want to set a trap on the path you know your target will walk next day, you can just cautiously do so and take 10 to guarantee a success, provided your character has the time and ease of mind to do so.

There is also the Reliable talent that some classes get, for example Rogue gets it at level 11, which makes it so that whenever the rogue rolls 9 or less in a skill check she's proficient with, she can replace the result with a 10. But that is a high-ish level perk, not a standard blanket rule.

In general, I like die rolls in tabletop DnD because failure can lead to the player having to improvise and face fun situations, but whether that is something a videogame can do or not, is still up in the air.

I see why Larian is trying to fix the dice roll situation. I have started the game on 5 different occasions. I have only gotten the Intellect Devourer once. Bad rolls.

Lue

I have gotten the intellect devourer 5 out of 7 times so some people just have bad rolls. That is how random works. Humans mentally weight things.

Anyhow the article is neutral for me, with no red flags or green ones really. Seems open and honest. I worry they may dumb the game down too much, as I saw a little of that when they added weighted dice .. but its an option so no hard done. But they also make the druid/child quest much easier and changed the whole thing with the other druid you go for healing. It made it more fail safe, i.e. dumbed it down, based on feedback. But they want to sell games to more then a small niche so can also see why.

Anyhow on my 4th play thru with over 112 hours played and couldn't be more happy with this game. Love where it is going and if the final game is anything like the first chapter its going to be a golden paradise to play.

"I see why Larian is trying to fix the dice roll situation. I have started the game on 5 different occasions. I have only gotten the Intellect Devourer once. Bad rolls."

It's not a "situation." Sometimes you fail rolls, sometimes you succeed. That's how the game works and should work. The only thing Larian need concern themselves with is making sure failure in some roll doesn't have outsized repercussions. A real life DM wouldn't say "You walk into a cave. Roll 1d100. Oh, you got a 12? You fall in a pit you didn't see in the dark and you die." And neither will Larian. They also won't lock huge amounts of content behind a random die roll result. But this idea that it's broken if you don't pass every roll is so bizarre. What do you think dice rolling is? What do you think random chance is?

I think it's probably a good idea to include an option for massaged results. Not loaded dice exactly, but something that adds a modifier that considers past results, and provides some with a version of what they think probability should be, rather than the reality of random chance. And then true RNG for the rest of us.

Luethar said:

I see why Larian is trying to fix the dice roll situation. I have started the game on 5 different occasions. I have only gotten the Intellect Devourer once. Bad rolls.

Lue

Click to expand...

Are you sure it's bad rolls and not a less than optimal build for this specific roll?

I want to make noncombat rolls so I sacrifice combat ability to get better modifiers for non combat rolls. I've made most of my rolls in my several playthroughs and since combat is fairly easy I don't feel too bad about gimping my combat attributes a bit. I do get the occasional terrible roll but that's to be expected when dealing with chance.

Having to make choices like this is one of my favorite parts of DND vs games where you can be great at everything.

Arkadia7 said:

I don't think I want a game designed by the fickle crowd. That sounds horrible. It sounds like Larian may be going badly wrong. This is not making me eager or wanting to have anticipation for future games if this is the way Larian thinks now. I think it will be very ironic if Larian starts to fall apart over one of his "dream" projects - BG 3. He needs to read some of the very good critiques and critical reviews on steam and figure out what they are doing wrong, but probably too late for that, to be honest.

Click to expand...

It will only be bad if they're being dumb about it. Data is great, it's how you use it that's important. For instance, if they want a particular choice to be hard and 90% of people pick one of the options, then they missed something. They can use the data to give more interesting options for the players, for example.