Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Vavrek
Mar 2, 2013

I like your style hombre, but this is no laughing matter. Assault on a police officer. Theft of police property. Illegal possession of a firearm. FIVE counts of attempted murder. That comes to... 29 dollars and 40 cents. Cash, cheque, or credit card?
On a different (I think) topic, I'm still interested in this:

Cattywampus posted:

I'm curious as what devs think of the whole 'games as service' mantra that's been coming from a certain quarter?
As well as exactly what the phrase "games as service" is intended to mean? I'm slightly familiar with the Software as a Service model, but mostly as "People call SaaS 'cloud computing' and they shouldn't because it's not what was originally meant by 'cloud computing'." Is it a reference to more subscription-oriented (or other continuous-revenue) payment models, the idea of continuing development past the official release date, both, neither? And, of course, thoughts? Good/bad/weird/irrelevant?

Adbot
ADBOT LOVES YOU

Bushmaori
Mar 8, 2009
Speaking of the cloud: What are the thoughts of you developers when it comes to cloud computing helping gaming? I remember reading a lot about it a few years ago but not so much since then.

Gerblyn
Apr 4, 2007

"TO BATTLE!"
Fun Shoe

Bushmaori posted:

Speaking of the cloud: What are the thoughts of you developers when it comes to cloud computing helping gaming? I remember reading a lot about it a few years ago but not so much since then.

Well, we use cloud saves, so I guess that counts?

Personally, I think it's risky to make a game that has an integral reliance on cloud computing, since that means it pretty much must be online only, and that is a big negative in the minds of many gamers. I believe Cloud Computing was one of the justifications that EA gave as to why the new Sim City had to be online only, for example.

I'm not even certain it would be that useful, the only real thing I can imagine is running AIs in strategy games, since 1) they can use a truly ferocious amount of processing power and 2) strategy games aren't twitch based, so players will tolerate the delay while the process is offloaded to the cloud.

That could just be down to my lack of imagination though, rather than a lack of actual uses!

MissMarple
Aug 26, 2008

:ms:

Vavrek posted:

As well as exactly what the phrase "games as service" is intended to mean? [...] Is it a reference to more subscription-oriented (or other continuous-revenue) payment models, the idea of continuing development past the official release date, both, neither? And, of course, thoughts? Good/bad/weird/irrelevant?
Yes, broadly.
GaaS is the idea that the product you launch is just the start, and that there will be a continued development of both features and content when the game is "Live" with some parallel revenue stream to support that. You could make the argument the revenue stream isn't important; outlier titles like Minecraft or Angry Birds both became insanely popular in part because they just kept adding features at the same fixed price. That improves the value for people who haven't hit the button yet, as well as giving them confidence that buying into it is an investment in future development too, but most importantly keeps veterans of the game playing it for months and years (and therefore evangelising it to others).

The earliest adventure into the money side of this was actually the DLC or Expansion Pack; selling you another chunk of content for a one off fee.
MMOs moved into a Subscription model; where "Pay to Play" funded the game continuing to be live and have live content for it.

I'm wary of steering the conversation into "costs are rising! we have to make more money!". However, the two things that are happening more and more as part of GaaS is that companies are trying to improve efficiency in two areas. Content Creation, and Willingness to Pay.

We'll get the ickier one out the way first, which is Willingness to Pay. The short version is; every player has a different amount they are willing to pay for what you are offering. If you price it at $60, you miss out on the guy who would pay $30 (although he might get it later on Sale, or second hand). You also make less money out of the guy willing to pay $200. "Limited Editions" that have higher price points and come with in-game and physical extras (that cost less than the additional dollar price) were a way that Publishers began to optimise for this. As Publishers and Developers have tried to optimise for this more and more, you can end up with practices that feel like they "nickel and dime" a player because they are trying to get every last cent you are willing to give the game out of you. You also end up with some predatory systems that aim to get a bit more than you were strictly Willing to Pay, often by obfuscating value, or having random outcomes.

The simple fact is that if you put options to spend more money in front of your most devoted players, they will. This is largely necessary to fund the second part, which is Content Creation. Games planned to be a GaaS will often deliberately design themselves to have specific types of content that can be easily expanded on, often infinitely so. For example, having a basic customisation system that you can easily deliver new content to without needing a game update. Or perhaps Multiplayer was specifically designed and built to make it easy to develop new modes and modifiers. If done well, this can be a highly efficient way to continue to keep your game fresh and engaging. The lower you can get the cost of adding quality content, the less you need to fund it with monetisation. That's good for both the developer and the player.


Being very reductionist about it, gaming is a hobby. Some people choose to spend their free time on games. As with any hobby, people are willing to spend a reasonable amount on it.
The bigger the share of that time you earn from players, the more money they are willing to give you.
GaaS is one way Developers are optimising around gaining more share of that time.
The other way I think we're increasing seeing is games making bigger offerings. More hours of gameplay. More km of terrain to explore.
If you offer people 18 quintillion planets to explore, that's going to fulfil your hobby longer than one that offers 5, right?

MissMarple fucked around with this message at 14:57 on Dec 13, 2017

djkillingspree
Apr 2, 2001
make a hole with a gun perpendicular

Bushmaori posted:

Speaking of the cloud: What are the thoughts of you developers when it comes to cloud computing helping gaming? I remember reading a lot about it a few years ago but not so much since then.

My personal take was that this was a line essentially proffered by MS to support their Azure initiative and a theoretical way to shore up the original Xbox One's weaknesses vs. the PS4. I don't think there's ever been a realistic example of it being used to offload computation from client PCs in the ways it was being talked about previously.

The problem I'd see with that model of cloud computing is that the problem domain where it would be useful doesn't justify the potential risk. I could see potentially a cloud computing company developing a game that used it essentially as marketing for the idea of using cloud computing in games, but the use cases that I've heard described for it are either impractical or not worth the cost and risk. If you think about it, if you're using cloud computing to offload something that the client can't calculate in real-time, you need to be able to send the problem to the cloud, computer AND deliver the results in real-time, which is an extreme challenge. And, if you can afford NOT to deliver those results in real time, then why not just compute the results in non-real time on the client? Hard to imagine a really compelling use case.

On the other hand, from another point of view every client/server game where the server is remote from all of the players is, in a way of speaking, using cloud computing. Things like MMO servers are sorta doing cloud computing in the sense that there is computation happening on a server that would be impractical to do on each individual client. Especially as games start to leverage scalable infrastructure like AWS, there's some sense that cloud computing is helping with the server side computation of running online games, but not so much offloading traditional client loads.

foutre
Sep 4, 2011

:toot: RIP ZEEZ :toot:

Gearman posted:

A few positions that might be up your alley and are in pretty high demand now and in the near future: data scientist, data analyst, and data engineer. There are also general design positions that rely on looking at large amounts of user rest data for balancing, tuning, and general player behavior.

Cool, that does sound v well-suited to my skillset, and interesting. Thanks!

mutata posted:

You should ask this in the Games Jobs thread here: https://forums.somethingawful.com/showthread.php?threadid=3415662&pagenumber=601&perpage=40 You may get more info there since there are a lot of folks in there who don't post in here. :)

Oh, duh, thanks. Somehow missed that in forums search.

Mother
Sep 30, 2004

You are help Orz with *parties*.

djkillingspree posted:

I don't think it's necessary to assume that the direction that publishers have been moving in is inherently correct or data-driven, even though they are all doing it. You may remember when the entire industry lost a shitload of money deciding it was time to make MMOs when WoW made bank, or time to make MOBAs when LoL made bank. One problem with pointing to research is that the counterfactuals don't exist - because most companies tend to work off of the same playbook, it's hard to point to the publisher that isn't following it and how successful they were or weren't.

You're missing what the actual publisher play was.

When the big companies all decided to make MMOs, the executives didn't get in their boardrooms and say "Guys! MMOs are THE thing! They are sure to work! There is NO WAY we can lose!"

They said "Our competitors have a MMO. It is doing very well. We have a silly pile of money sitting in the bank. For a relatively minor amount, we can spin up a MMO team and try to get in on the trend too. Best case, we make a LOT of money. Worst case, it fails -- the stock price dips a point (unless one of our other games has a good year and covers the loss), we F2P the thing to recoup some of the dev cost, and we blame the dev so none of us get fired."

(Then they cackled while rubbing their hands together and ate some babies or something.)

No game is certain. These companies make calculated bets. Every one of the many publishers scrambling for a PUBG entrant right now knows that 90% of these are going to fail. But to be part of the 10% that doesn't fail, you have to take a shot. So, they take shots.

Every major publisher had failed MMOs (including Blizzard if you count canceled projects) -- which one was killed by the failure?

djkillingspree
Apr 2, 2001
make a hole with a gun perpendicular
If you think a small amount of money was lost on those MMOs or that small teams were involved in the big publisher MMOs, I don't buy that. And while EA didn't go out of business, their stock did go down by *over 33%* in the year after SWTOR launched, and ultimately one of the main reasons John Riccitello was ousted was the poor financial performance of EA over his tenure, much of which was defined by the insanely expensive development of SWTOR and its ultimate failure to succeed financially as a subscription MMO.

I would point to the shuttering/sale of SOE, 38 Studios, etc. as examples of why these were, in fact, bad bets.

The fact that some of those MMOs have been able to be put on life support through F2P doesn't imply that the companies that developed them weren't heavily injured by their failure, or that they were smart gambles.

The extent to which a major publicly owned company needs to fail to actually stop existing is pretty spectacular. A large, publicly held company can undergo financial disaster and still survive.

I will say that MOBAs involve a massively lower amount of risk than MMOs, so to that extent, I think they are less of a reckless investment on the part of publishers. Though I still think publishers chasing them in a massive way are missing out on the reality of what people want in that segment, personally.

djkillingspree
Apr 2, 2001
make a hole with a gun perpendicular

MissMarple posted:

a bunch of gaas stuff

This is a really good and accurate breakdown of GaaS. The extent to which the industry has discovered that there are customers willing to pay a lot more than they used to for games is a big part of how the industry has moved in the past like, 5-10 years.

Whistling Asshole
Nov 18, 2005

djkillingspree posted:

I would point to the shuttering/sale of SOE, 38 Studios, etc. as examples of why these were, in fact, bad bets.

38 is probably not a good example since it was destined to fail not necessarily because of the genre (though that definitely didn't help) but because it was run by Bobby Baseball-Hero who had 0 experience running a business or developing games.

Related anecdote: A friend of mine is one of the best game programmers I've ever met, and he was recruited heavily by 38 including multiple personal calls by Curt Schilling to come out and interview with them. He accepted, and they put him up in a 4 star hotel for a week, comped all his expenses, and basically rolled out the red carpet. A few hours into the interview, my friend asked them if they had full access to the source code for one of the key pieces of middleware they were using because he had just finished working on a project that used the same middleware and discovered it had major issues with the game engine they had chosen. The company that made the middleware wouldn't give them the full source and it effectively crippled the project. It was at this point that the people interviewing him looked at each other nervously and said they'd have to investigate further. My buddy was offered the job but declined based on a bunch of red flags he saw with the company. It turns out that him asking this question resulted in them either renegotiating their deal on this piece of middleware (or dropping it entirely and starting over, I forget).

In short, every story I've ever heard about 38 makes it sound hilariously mismanaged despite having a wealth of talented developers.

Whistling Asshole fucked around with this message at 05:41 on Dec 14, 2017

Mother
Sep 30, 2004

You are help Orz with *parties*.

djkillingspree posted:

If you think a small amount of money was lost on those MMOs or that small teams were involved in the big publisher MMOs, I don't buy that. And while EA didn't go out of business, their stock did go down by *over 33%* in the year after SWTOR launched, and ultimately one of the main reasons John Riccitello was ousted was the poor financial performance of EA over his tenure, much of which was defined by the insanely expensive development of SWTOR and its ultimate failure to succeed financially as a subscription MMO.

I would point to the shuttering/sale of SOE, 38 Studios, etc. as examples of why these were, in fact, bad bets.

The fact that some of those MMOs have been able to be put on life support through F2P doesn't imply that the companies that developed them weren't heavily injured by their failure, or that they were smart gambles.

The extent to which a major publicly owned company needs to fail to actually stop existing is pretty spectacular. A large, publicly held company can undergo financial disaster and still survive.

I will say that MOBAs involve a massively lower amount of risk than MMOs, so to that extent, I think they are less of a reckless investment on the part of publishers. Though I still think publishers chasing them in a massive way are missing out on the reality of what people want in that segment, personally.

I thought the argument here was “they aren’t correct / data-driven, they’re all just copying one another and chasing MMOs (etc.)”? That assertion is not supported by “here is an expensive game that failed.”

I never said a small amount of money was lost. I said the amount of money lost is expected. They want to bet on what they identify as good risks. A good risk is still a risk. A good risk can still fail. When it does, that doesn’t mean that it wasn’t a risk worth taking or that the people who decided to take it just threw darts or said “I hear WoW is making money, let’s copy that.”

They expect some of their bets to fail. Riccitiello didn’t get ejected because he said “everyone is chasing WOW, let’s (seven years after WoW came out?) get on that bandwagon with SWOTR” -- but then that game failed.

He got the axe because of the end product of his overall strategy, a direction that had not performed for five straight years. Rapid and aggressive expansion was expensive, BioWare / Pandemic, Playfish were not cheap and didn’t ROI, the SimCity fiasco, Mass Effect 3, SWOTR, the press from being the worst company ever, constant layoffs, very public high-level departures, a 70% drop in stock price while he was in charge, the silliness of an MMO group scattered between Austin and Edmonton and Quebec and Virginia, etc. He didn’t go because the one SWOTR bet failed, he went because the board thought *none* of the bets he was making had paid off and he wasn't doing a good job of making EA "work." They were done waiting.

My overall point is simply that these guys are very data-driven and calculating (far too much so, IMO). They are not a bumbling group of goofs cluelessly chasing whatever they see everyone else doing. It often makes financial sense for them to attempt to get in on a trend, even when the chances that it will fail are high. (Which, in the case of SWOTR, hasn’t been the case longer-term, given that the thing makes money for them today.)

Willie Tomg
Feb 2, 2006

Gerblyn posted:

Sure, but the problem is you guys aren't actually providing much in the way of counter arguments. It's all very well to say things like "You can't prove the data is correct" and "Companies have made mistakes in the past", in fact it's perfectly valid, but what is your evidence that this data in particular is wrong?

I'm not trying to say that you guys are necessarily wrong, or that we're necessarily right, I'm just saying that you guys need to provide your own arguments because it feels all you're doing is attacking ours.

That the data is nonfalsifiable and of questionable utility is the counterargument. That you stipulate these things is a problem!

berenzen
Jan 23, 2012

What gain would they get from falsifying that information though? Particularly because it would come back to bite the market research company in the vein of extremely large lawsuits as companies like EA/Activision/etc. would act on said research. If it came out that those companies lost money because of fraudulent data, do you think that they wouldn't sue the poo poo out of said market research company?

OneEightHundred
Feb 28, 2008

Soon, we will be unstoppable!

MissMarple posted:

However, the two things that are happening more and more as part of GaaS is that companies are trying to improve efficiency in two areas. Content Creation, and Willingness to Pay.
A big thing that's underappreciated is that obsolescence has slowed a lot too. "We've achieved photorealism!" is still a punchline, but technological advancement is deep in diminishing returns. Hardware advances have lost their ability to revolutionize games every 5 years and the bottleneck has been shifting to production costs. One of the big risks of maintaining a game 2-3 years after launch was that it would start to look and feel dated, but that process has slowed significantly, to the point that there are games porting to the next hardware generation and continuing their life. The expectation is reversing into expecting that games GAIN value over time.

OneEightHundred fucked around with this message at 04:48 on Dec 18, 2017

ShadowHawk
Jun 25, 2000

CERTIFIED PRE OWNED TESLA OWNER

berenzen posted:

What gain would they get from falsifying that information though? Particularly because it would come back to bite the market research company in the vein of extremely large lawsuits as companies like EA/Activision/etc. would act on said research. If it came out that those companies lost money because of fraudulent data, do you think that they wouldn't sue the poo poo out of said market research company?
Expensive consultants telling management what they want to hear is a bit of a cliche for a reason.

berenzen
Jan 23, 2012

ShadowHawk posted:

Expensive consultants telling management what they want to hear is a bit of a cliche for a reason.

A lot of market research companies are 3rd party, them faking data would literally break their companies.

ShadowHawk
Jun 25, 2000

CERTIFIED PRE OWNED TESLA OWNER

berenzen posted:

A lot of market research companies are 3rd party, them faking data would literally break their companies.
It's not about faking data, it's about painting an incomplete picture by looking only at the stuff you already think you want to look at. There are many ways to be fooled by correct data.

mutata
Mar 1, 2003

Well no one in here is providing actual data about their side of the debate and quite frankly that's because we unfortunately don't have access to it and even if we did, market research is expensive and is bought and therefore counts within companies as valuable trade secrets so we're not likely to see any manifest here.

I've lost sight of the thesis statements on either side here, frankly. If the position is that market data can lead to poor decisions, then that's absolutely correct, but not all the time. In my opinion Disney Infinity was market tested into oblivion and we did far too little just listening to our customers (there are other issues behind this though. Everything is more complex than we think it is). If the position is that companies should not rely on market data to make creative decisions, then that's fair to say but not realistic to expect from AAA aside from maybe a couple outliers. Mid-tier and indie are much more able to sluff huge data trends and go for a niche, but even then it's a gamble. All companies want to get as close to being able to tell the future as possible and that's not limited just to games industry companies either.

This is all pretty obvious, though, so, like I said, I may be missing people's end game points here.

mutata fucked around with this message at 09:33 on Dec 18, 2017

Tricky Ed
Aug 18, 2010

It is important to avoid confusion. This is the one that's okay to lick.


Yeah, ultimately any decision just comes down to 5-10 people in a room doing what they think is right. The difference is, of course, what factors in to "right." The bigger the company, the more likely they're leaning towards following a trend, minimizing risk, and maximizing profits, and the less likely they're considering fun or mechanics. Companies where they do both are rare.

Mother
Sep 30, 2004

You are help Orz with *parties*.
Market research isn’t what many people think it is. I know that when I came into the business, I definitely believed someone did a bunch of math, cooked up a report, and told you “this will sell 10 million copies” – and they ended up being right. That doesn’t exist. Market research is just data. Some good, some bad, some hilarious. It can be very useful, but you still have to read it and make decisions.

A lot of places (not just games) are silly with data. You’d be surprised (or maybe not) at how many people in high-level positions don’t ask questions about numbers they are handed if the presentation looks pro.

I had a day-long argument with a PM who angrily insisted my questions about his data were me trying to deflect the problems shown by the numbers. His data showed that the “average time to complete the first tutorial” on a game we were working on was something like 90 hours. Almost anyone who has ever done any work on a game (or played a game or paid attention to time or been awake) is going to see something like “90 hours to complete the first tutorial” and think “that cannot be right.” But, there we were….

mbt
Aug 13, 2012

Have you ever had a great idea for a game but gotten scooped on it?
As in someone beat you to the punch and made something beyond what you had imagined. Would you still pursue the project or try to refactor it into something different?

I imagine there were a lot of people making harvest moon farming sim type of games who had their lunch completely eaten by Stardew Valley, there's almost no point in releasing another one of those games for a few years due to how great it is.

Sub question - how do you view other games in the same genre during development? If you're making an FPS and a game like Doom 2016 comes out, do you play it and frantically take notes trying to integrate the best parts of it, or do you try to ignore its influence to maintain your original vision?

Bushmaori
Mar 8, 2009

Gerblyn posted:

I'm not even certain it would be that useful, the only real thing I can imagine is running AIs in strategy games, since 1) they can use a truly ferocious amount of processing power and 2) strategy games aren't twitch based, so players will tolerate the delay while the process is offloaded to the cloud.

That could just be down to my lack of imagination though, rather than a lack of actual uses!

That makes a lot of sense and is something I completely overlooked. You're saying that in this way it would be a case of send state, compute, receive state, which is a perfect fit for the technology?

djkillingspree posted:

The problem I'd see with that model of cloud computing is that the problem domain where it would be useful doesn't justify the potential risk. I could see potentially a cloud computing company developing a game that used it essentially as marketing for the idea of using cloud computing in games, but the use cases that I've heard described for it are either impractical or not worth the cost and risk. If you think about it, if you're using cloud computing to offload something that the client can't calculate in real-time, you need to be able to send the problem to the cloud, computer AND deliver the results in real-time, which is an extreme challenge. And, if you can afford NOT to deliver those results in real time, then why not just compute the results in non-real time on the client? Hard to imagine a really compelling use case.

This here was what confused me so much about the idea. I remember seeing examples of huge scale physics simulations apparently only being possible through cloud-computing. In this case, I had to wonder exactly how viable this would be. In terms of computing, the cloud would be great. After that, you would have to factor in the transmission of potentially thousands of individual physics pieces, in a constantly delivered method which also would have to take into account data loss. Maybe forms of compression could reduce the overhead in terms of online bandwidth, but that would seem to bring its own issues of real-time decompression back on the home console?

So for my next question, forgive me if this has already been asked, I would like to know where you guys see the next generation of consoles going? I'm especially interested when it comes to the areas of mobility, like the Switch, backward compatibility (apparently easier now that nobody is using weird-rear end cell-processor technology?), and whether you see an increase in CPU power to aid in reaching a more common 60fps frame-rate.

Pixelante
Mar 16, 2006

You people will by God act like a team, or at least like people who know each other, or I'll incinerate the bunch of you here and now.
Are there any useful Christmas presents to buy for a teen who is pursuing a career in game dev? He's done a Digipen high school thing and really liked it.

leper khan
Dec 28, 2010
Honest to god thinks Half Life 2 is a bad game. But at least he likes Monster Hunter.

Meyers-Briggs Testicle posted:

Have you ever had a great idea for a game but gotten scooped on it?
As in someone beat you to the punch and made something beyond what you had imagined. Would you still pursue the project or try to refactor it into something different?

I imagine there were a lot of people making harvest moon farming sim type of games who had their lunch completely eaten by Stardew Valley, there's almost no point in releasing another one of those games for a few years due to how great it is.

Sub question - how do you view other games in the same genre during development? If you're making an FPS and a game like Doom 2016 comes out, do you play it and frantically take notes trying to integrate the best parts of it, or do you try to ignore its influence to maintain your original vision?

I expect very few people were working on harvest moon clones. And surely a successful one launching is a reason to /finish production/ not cancel.

It’s hard to know what question you’re asking, as you refer to both an independent project with effectively no budget and also a well financed project from a large corporation. Generally, everyone plays everything (to some practical limit). If a game does something with its UI or controls that are really good, those things will find their way to other games because they work. You may even hear: “you should check out the [UI, art, procedural generation, pacing] in <game>” from friends/colleagues. This is essentially how the medium has progressed for decades. See also music, art, film, or any other creative endeavor.

Things that involve a lot of tech are a bit different. Indies don’t generally have the budget to build stuff like the animation system in DooM (both from a tech and art perspective) and large projects are slower to adapt to fast changes by their nature.

It’s definitely happened that games get rushed to beat a competitor to market, but that decision typically gets made at the exec level. Most of the people in industry here aren’t making those decisions. Or if they are, they’re probably on a relatively small team.

leper khan
Dec 28, 2010
Honest to god thinks Half Life 2 is a bad game. But at least he likes Monster Hunter.

Pixelante posted:

Are there any useful Christmas presents to buy for a teen who is pursuing a career in game dev? He's done a Digipen high school thing and really liked it.

If they’re into programming, the unity course from Ben tristem on Coursera is pretty good and routinely excessively cheap. It’s project based. Did I mention you can probably pick it up for :10bux:

No idea on the art side..?

mbt
Aug 13, 2012

leper khan posted:

I expect very few people were working on harvest moon clones. And surely a successful one launching is a reason to /finish production/ not cancel.

It’s hard to know what question you’re asking, as you refer to both an independent project with effectively no budget and also a well financed project from a large corporation. Generally, everyone plays everything (to some practical limit). If a game does something with its UI or controls that are really good, those things will find their way to other games because they work. You may even hear: “you should check out the [UI, art, procedural generation, pacing] in <game>” from friends/colleagues. This is essentially how the medium has progressed for decades. See also music, art, film, or any other creative endeavor.

Things that involve a lot of tech are a bit different. Indies don’t generally have the budget to build stuff like the animation system in DooM (both from a tech and art perspective) and large projects are slower to adapt to fast changes by their nature.

It’s definitely happened that games get rushed to beat a competitor to market, but that decision typically gets made at the exec level. Most of the people in industry here aren’t making those decisions. Or if they are, they’re probably on a relatively small team.

I'll rephrase it: have you ever been working on a game and had another game come out in the same genre that completely blew you away? Is it more ':smith: now we're going to be compared to this' or "that's awesome and doesn't affect my morale at all" or something else entirely?

leper khan
Dec 28, 2010
Honest to god thinks Half Life 2 is a bad game. But at least he likes Monster Hunter.

Meyers-Briggs Testicle posted:

I'll rephrase it: have you ever been working on a game and had another game come out in the same genre that completely blew you away? Is it more ':smith: now we're going to be compared to this' or "that's awesome and doesn't affect my morale at all" or something else entirely?

Personally, I get a feeling similar to this: :capitalism:

The interpretation of which may be good or bad depending.

Star Warrior X
Jul 14, 2004

Meyers-Briggs Testicle posted:

I'll rephrase it: have you ever been working on a game and had another game come out in the same genre that completely blew you away? Is it more ':smith: now we're going to be compared to this' or "that's awesome and doesn't affect my morale at all" or something else entirely?

Nah, this tends to not happen. There are a few reasons. First of all, most games aren't really unique anyway, so you can focus on what makes your game different from the thing that just came out. It's really unlikely that the other game did everything you were planning for yours.

Even if it did, though, games aren't like tools, where you only want one of each kind, and only the best one you can get. Games are like cake. If you really like ice cream cake, and you eat a good ice cream cake, it's usually not too long before seeing another ice cream cake doesn't make you think 'oh, but I already ate ice cream cake,' but instead, 'yay! More cake!'

mutata
Mar 1, 2003

Pixelante posted:

Are there any useful Christmas presents to buy for a teen who is pursuing a career in game dev? He's done a Digipen high school thing and really liked it.

Game Maker can also be a great piece of hobbyist software for people starting out. If they're an artist, then a monoprice or huion graphics tablet is standard kit. Wacom's are best but super pricey.

Mother
Sep 30, 2004

You are help Orz with *parties*.

leper khan posted:

If they’re into programming, the unity course from Ben tristem on Coursera is pretty good and routinely excessively cheap. It’s project based. Did I mention you can probably pick it up for :10bux:

No idea on the art side..?

Second this. I did his Unity course on Udemy. I have had good luck with Udemy tutorials in general and (if you poke around) you can gift tutorials to people from the site -- I just did a few for Christmas.

Gerblyn
Apr 4, 2007

"TO BATTLE!"
Fun Shoe

Bushmaori posted:

That makes a lot of sense and is something I completely overlooked. You're saying that in this way it would be a case of send state, compute, receive state, which is a perfect fit for the technology?

I don't know about perfect, but it would be suitable yeah. As you say, you'd need to upload the game world, which is essentially your save file, let the cloud run its turn, then send a save file back to the player of the world after the AI took it's turn, along with data so that the player could see the moves the AI took. The whole process would probably take a few minutes, but if you're playing a round-robin type turn based game, that's fairly normal.

If the game supports Play By Email, you could also have a cloud based AI take part in that as well, since PBEM games essentially involve emailing save games around.

GeeCee
Dec 16, 2004

:scotland::glomp:

"You're going to be...amazing."

Bushmaori posted:

So for my next question, forgive me if this has already been asked, I would like to know where you guys see the next generation of consoles going? I'm especially interested when it comes to the areas of mobility, like the Switch, backward compatibility (apparently easier now that nobody is using weird-rear end cell-processor technology?), and whether you see an increase in CPU power to aid in reaching a more common 60fps frame-rate.

The next consoles will be powered by the A-HAB Engine™ which will give developers even more terafloopies in their games to dedicate to more elaborate sources of monetisation.

GeeCee fucked around with this message at 02:21 on Dec 20, 2017

djkillingspree
Apr 2, 2001
make a hole with a gun perpendicular

Gerblyn posted:

I don't know about perfect, but it would be suitable yeah. As you say, you'd need to upload the game world, which is essentially your save file, let the cloud run its turn, then send a save file back to the player of the world after the AI took it's turn, along with data so that the player could see the moves the AI took. The whole process would probably take a few minutes, but if you're playing a round-robin type turn based game, that's fairly normal.

If the game supports Play By Email, you could also have a cloud based AI take part in that as well, since PBEM games essentially involve emailing save games around.

I wonder how much value there is to the cloud in this case, given the fact that it would mandate an online connection, which is likely to piss some people off. I do wonder how much strategy game AI is limited by processing power vs. more design issues of how to design an AI that is fun to play against. It feels like the latter is a design challenge nobody's cracked moreso than a technical challenge?

On the subject of cloud computing, something that I'd heard suggested as a potential use case, which sounded kinda neat and plausible, was delivering your character model to the cloud and then rendering out pre-rendered cutscenes with your character inserted in the cloud and streaming them back. That's a use case that's actually pretty plausible (it's just sending a model out and streaming video back, no input is required), is actually pretty cool, and could actually have a fallback case for offline games (either a generic fallback movie, or falling back to in-engine quality for the movie if no connection is present).

Gerblyn
Apr 4, 2007

"TO BATTLE!"
Fun Shoe

djkillingspree posted:

I wonder how much value there is to the cloud in this case, given the fact that it would mandate an online connection, which is likely to piss some people off. I do wonder how much strategy game AI is limited by processing power vs. more design issues of how to design an AI that is fun to play against. It feels like the latter is a design challenge nobody's cracked moreso than a technical challenge?

The AI in Age of Wonders 3 is CPU limited, given more power we could make some simple tweaks and you’d get a moderately smarter opponent. There are also some techniques and processes we never attempted due to the high CPU cost, which would certainly make the AI smarter, like letting the AI dynamically customize it’s army compositions to try and counter yours, for example.

As for the Design question, that’s not really the case. Although design can inform how the AI should act, actually making it act that way in an intelligent and dynamic way is a programming problem and extra CPU power would be immensely helpful. You can design around the problem, but then you end up with something assymetric, like Sorceror King, where the AI is actually playing a completely different game to you.

Red Mike
Jul 11, 2011
Smarter AI does not equal a more fun AI to play against, that's something that's been known for well over a decade. Depending on the game, perfectly smart AI range from easily possible to very hard to do but very easy to fake.

The most fun AI to play against most likely isn't going to be one of those. In an FPS, players tend to look for sheer difficulty that is almost identifiable as human (realistic mistakes or delays, and lack of easy predictability to abuse). In a strategy game, players tend to look for varying difficulty (not even nearly optimised like a hardcore player would) and reactions to your actions that seem to be justifiable if it were a human.

Ironically, you can take the perfect AI that can play the game perfectly, add an "if Random(0, 100) > 50 then do something dumb" statement running every so often, and suddenly you've got a more fun AI to play against.

Of course, cooperative AI on the other hand benefits a lot more from being smart, but similarly should never be smart enough that you're made redundant as a player because of it.

e: and re: justifiable reactions, most players read patterns in noise. This is part of why having AI that quasi-randomly decides to do dumb things makes players read intent into it like "oh it decided to build a three layer wall around that one pass because it thought I'd attack from that direction, so the rest of the base has barely any walls".

That one article on the AI of the Pacman ghosts would be relevant here if anyone has the link.

Red Mike fucked around with this message at 14:35 on Dec 21, 2017

AG3
Feb 4, 2004

Ask me about spending hundreds of dollars on Mass Effect 2 emoticons and Avatars.

Oven Wrangler
I don't know which article that is, but I saw a video about the Pacman AI just the other day: https://www.youtube.com/watch?v=l7-SHTktjJc

Gerblyn
Apr 4, 2007

"TO BATTLE!"
Fun Shoe

Red Mike posted:

Smarter AI does not equal a more fun AI to play against, that's something that's been known for well over a decade.

I’m not disagreeing with what you’re saying at all, I think I expressed myself badly. When I used words like Smart, I didn’t mean Smart like Napoleon destroying the player with searing strategic insight, I meant Smart like another human playing a reasonably competent game.

I think you’re underestimating how difficult it is write a 4X game AI which isn’t pants-on-head incompetent, yet alone one which is so smart that the player hates trying to beat it. Managing production queues, empire expansion, diplomacy, economics, research and military deployment, all with something that even roughly ressembles a human’s playstyle is a hugely complex task requiring a ton of prvoessing power. There’s a reason people complain that 4X AIs only provide a reasonable challenge by cheating in extra resources and units.

In the end, faking it only gets you so far with something as complex as a strategy game AI, and getting it
to play reasonably well falls squarely into the domain of the AI programmer. Most designers are only really qualified to provide the highest levels of direction, they can’t design around these problems except by fundamentally changing the rules of the game the AI is playing. That’s why I mentioned Sorceror King, which I personally think is going the right direction for people who want a fun, single player 4X game.

In SK, the Big Bad Guy AI doesn’t need to deal with all the complex stuff that makes the game fun for a player, all it needs to do is spawn armies and things which are tailored to challenge the player. If a designer says “Fighting this kind of army while trying to invade is fun”, the AI can just create one in the right place, which is far simple than what an AI in a traditional 4X would need to do to get the same result.

Wallet
Jun 19, 2006

Gerblyn posted:

There’s a reason people complain that 4X AIs only provide a reasonable challenge by cheating in extra resources and units.

I always wonder if people actually care about the cheating, or about the way that the cheating allows the game to get away with really poor AI. Civ's combat AI gets bagged on because it is incredibly bad at actually engaging in combat and makes up for this by having far more units than a player could ever produce, which warps the entire game from a player perspective, but if the AI managed units reasonably well and only cheated in reasonably sized armies, I'm not sure anyone would be bothered.

ninjewtsu
Oct 9, 2012

Wallet posted:

I always wonder if people actually care about the cheating, or about the way that the cheating allows the game to get away with really poor AI. Civ's combat AI gets bagged on because it is incredibly bad at actually engaging in combat and makes up for this by having far more units than a player could ever produce, which warps the entire game from a player perspective, but if the AI managed units reasonably well and only cheated in reasonably sized armies, I'm not sure anyone would be bothered.

i mean anyone who's aware that strategy AIs cheat like nuts and ever plays a strategy game above easy difficulty has to be ok with it on some level

something of note is that making an AI that can win an FPS is pretty easy, you just make it an aimbot and it can probably wipe the floor with pretty much any human player, but obviously no one would find that fun. a strategy AI is a much different problem space, and making an AI that can win without cheating period against a competent player is near impossible, forget about trying to make it "fun" as well. it seems like it'd be best if genre was specified in AI discussions.

Adbot
ADBOT LOVES YOU

Red Mike
Jul 11, 2011

Gerblyn posted:

I’m not disagreeing with what you’re saying at all, I think I expressed myself badly. When I used words like Smart, I didn’t mean Smart like Napoleon destroying the player with searing strategic insight, I meant Smart like another human playing a reasonably competent game.

I think you’re underestimating how difficult it is write a 4X game AI which isn’t pants-on-head incompetent, yet alone one which is so smart that the player hates trying to beat it. Managing production queues, empire expansion, diplomacy, economics, research and military deployment, all with something that even roughly ressembles a human’s playstyle is a hugely complex task requiring a ton of prvoessing power. There’s a reason people complain that 4X AIs only provide a reasonable challenge by cheating in extra resources and units.

In the end, faking it only gets you so far with something as complex as a strategy game AI, and getting it
to play reasonably well falls squarely into the domain of the AI programmer. Most designers are only really qualified to provide the highest levels of direction, they can’t design around these problems except by fundamentally changing the rules of the game the AI is playing. That’s why I mentioned Sorceror King, which I personally think is going the right direction for people who want a fun, single player 4X game.

In SK, the Big Bad Guy AI doesn’t need to deal with all the complex stuff that makes the game fun for a player, all it needs to do is spawn armies and things which are tailored to challenge the player. If a designer says “Fighting this kind of army while trying to invade is fun”, the AI can just create one in the right place, which is far simple than what an AI in a traditional 4X would need to do to get the same result.

Oh if you mean 'smart as in a realistic challenge' then, sure. My point still mostly stands though, in that faking it tends to have the same end effect for much less effort. Even better, in some cases it ends up being more predictable and malleable design-wise.

Example:

An AI developer decides to use next-gen top of the line convolutional neural networks (or whatever) to train up a network to serve as an enemy. It works great and is really fun. To use it in-game, it doesn't even take a lot of CPU really. It's released and everyone is excited at how good the AI is.

Then for the next update, the designer decides to add a new game mechanic that the AI needs to take into account. Worse, he wants the AI to act in a specific way (very defensive, very aggressive) for X minutes when something happens.

Suddenly that network needs to be retrained, have similar enough results that the players won't revolt at how 'different' the AI acts, and has to take into account an entirely different mode correctly in certain cases.

This is obviously a very specific example, but my point is this problem is very complicated, and 'smarter' is a really complicated thing to aspire to, and I'm not sure more CPU necessarily helps with that. Good design (of the game mechanics) and good design (of how you want the AI to act) is what helps most, in my opinion. If you have to fake something to make it work, you just have to make it so the average player (not you specifically) can either rationalise or not realise that it's been faked. You'd be surprised how many players don't even realise the AI isn't playing by the rules at all.



And yes, I'm well aware how insane it is to try and write 4X AIs. That was basically the 'impossible to do but easy to fake' one. Which is an exaggeration yes, it's still hard to fake. I don't think more CPU will help with that very much though (in the long run), for the above reasons.

e: This is the Pacman AI article I meant here.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply