While I have a great deal of respect for the author, this doesn't help quite a few of the companies (and PC gamers) out there.
Basically, the position 'we will only attempt to sell to people who would prefer to buy over download' doesn't scale to big budget titles. There are a lot of gamers out there who like AAA, content rich games. These are the games that need to sell a million+ units just to break even. Ignore the programming - some of these games have dozens of artists and designers working for multiple years.
The 'make niche games' position doesn't help these developers (or the gamers who love their games). We're talking about shops like Valve and Relic here.
The game industry is certainly eyeing whats going on in the movie and music industries. The basic truth is that most people would rather download for free than vote on what they want to see in the future by buying it.
Consumers demands for content rich games is exceeding sales. This means that big blockbuster titles are likely take a hit similar to flight sims several years back. For some gamers, this is probably great. There are plenty out there who would love to see the death of the FPS/action genre if it means a few smaller games come out in their place.
Its going to be a rough few years as big devs figure out how to stay in buisness. Its likely to drive the 'big/blockbuster' titles even more towards the consoles which big markets and lower piracy rates so far this generation.
Can you expand on the pieces you view as biased? Most of it looks rather straight forward. They explained that bsd style licenses grants a freedom, explained why this freedom can be beneficial, and explained what you lost by (downstream openness) by selecting the license.
That seems like a pretty simple and blunt explaination. Each license has pros and cons depending on need.
Thats what happens when one company signs a deal with another - they get to set up their own licensing rules. Valve probably paid id a whole bunch of money back when they started re-licensing HalfLife tech. At that point, they probably removed the license notices as if they were reselling it, they must have had rights to do so.
If you link against LGPL, you need to release object files. This allows people to relink your app against new library versions or alternate lib implementations. GPL requires releasing your source.
Some of these companies are stuck in a hard spot. With GPL 3, they are beening told to give back more than they were previously.
Sure, they shared their modifications and allowed others to make use of their code, they just protected their hardware. In many cases, this protection is either legally required or is required by their buisness model. For example, I'm the MPAA was rather concerned about TIVO. As they were a closed platform, they could work with MPAA to find a balance between freedom for the customer and content protection.
The PS3 is in a similar position. Sonys buisness model is to sell hardware at either a loss or little profit and then make money selling games. If the platform must be open, they are no longer in control of what is released on it.
And yes, it could be argued that their buisness model is invalid as it relies on restricting end user freedom. If they are forced to sell the consoles for a profit, they could easily be priced out of the market. Closed source competitors would have a significant buisness model advantage as they could sell their hardware could be much cheaper.
So the argument moves to 'well, they don't have to use GPL'. And again, thats valid. But it does mean there is one less company potentially contributing changes back to the community, instead likely funding a closed source alternative.
A single data point isn't statistically useful, but I graduated from a Bay Area high school in 1998. I didn't see any of the nonsense in the news at school. Most of the teachers were high quality, cared about education, and weren't afraid to fail students.
The only stereotypical 'California' issues I recall involved racial incidents. There a continuous paranoia about gangs and 'bad' teens from neighboring cities.
Overall, education is always what the students make of it. Students who pushed themselves opened doors.
I'm not arguing that used games are bad buisness for GameStop. Its definitely lucrative. Long term though, I don't know if it is good for anyone.
Publishers will figure out how to break the model (via DRM, contact-the-server, strong arming, etc). This will EB/GameStop who depend on this as a major income stream. If Gamefly gets much bigger, they could potentially become a target as well.
Gamers will be hurt as stores close and they have to deal with worsening anti-resale measures. They are already getting hit by decreasing game variety. The game industry isn't heading for a crash, but a re-adjustment looks likely. Game creation is simply too expensive for publishers to ignore missed sales.
In the end, its a short term win for GameStop but the long term impact could be messy.
No, their secret weapon is game resales. They buy games back for a few dollars and resell them for just under full price. They actively push used copies, putting them at the front of the store and offering used copies to people buying new.
It is hitting certain game styles enough (primarily single player, plot driven games) that fewer are being made. Offline games pushed to build in 'contact the server before you play' model, requiring an internet connection to play single player games. Or publisher based online distrobution, where transfering ownership costs money if it is possible at all.
So, they are estimating they wrote 120 million lines of new code for the NVidia drivers?
Fine, so I am sure they were exaggerating at 6 drivers, 20 million lines each. At the same time, how the heck does a driver take 20 million lines in the first place? Most next gen game and tools combined, running on several platforms, are less than 2 million lines.
A huge percent of game bugs are due to sound driver issues. The first suggestion for most crashes/framerate shuttering/sound stuttering in XP games was to reduce or disable hardware sound settings.
In addition, hardware sound frequently increases the load on the CPU. No idea why, but you can go from 5% of the CPU dedicated to sound without acceleration to 25% dedicated with it. If the sound card driver authors got built reliable, stable drivers (and hardware - the difference between the quality of the SB Live hardware revs is stunning), I doubt MS would have dropped hardware sound support.
Unfortunately, I think MS had to do this. Or they could have pushed for open source audio drivers.;)
Because we want to run our game on the 360? Because the PS3, while it supports a version of GL, works better with you use the layer under it? Because DX has a more consistent interface while GL depends more heavily on vendor specific extensions? Because DX has a greater critical mass (ie simpler to hire expert, better vendor support, better OS support on the primary gaming OS)
Frankly, porting a modern game to GL is cheap compared to game dev. Making a game might cost 10M, not including marketing, etc. Publishers want to make lots of money. If the potential revenue was worth it, more games would be ported.
I think this clause causes issues with LGPL for many software devs:
If you link other code with the library, you must provide complete object files to the recipients, so that they can relink them with the library after making changes to the library and recompiling it.
Most closed source shops aren't willing (or able) to provide object files. This has pushed many libraries that want wide spread adoption to less restrictive licenses. Or it requires the LGPL'd portion to be a dynamic library instead of a static library.
For example, Ogre is LGPL. They recommend not using it as a static library to avoid this clause. Unfortunately, console games historically can't use dlls (no OS support, though the 360 now allows it). They also can't release object files as the compilers are proprietary. Its all a dead issues anyway, as people can't update the executable. A BSD license avoids the issue entirely.
1) AA is a 360 cert requirement. Yes, a title or two may be able to get that waived by having something else special/that does the same thing. In general though, MS won't let 360 titles with jaggies ship. I am not sure what you are talking about here.
2) Yes, the EDRAM takes extra effort to work with. Sure, it would be nice if there was an infinite amount. In the end though, the tiling work generally impacts the rendering programmer for a short period and no one else. There is plenty of info for 360 devs on how to use it. If this is a devs main 360 complaint, MS is making devs lives easy - providing a good system, tools and documentation.
3) Marketing will always use renders. Their job is to get people excited about the title in a cost/time effective way - staged screenshots and renders are the fastest way to do that. This generation, difference in quality is much more visible in movies. A few marketing departments will use rendered movies here, but the cost is basically quite high relative to a rendered model.
The main 1080p challenge is the performance hit you get for drawing that many more pixels. For games that are GPU bound, going from 720p to 1080p hits the pixel shaders hard. The other issue is memory (same on both platforms) as many games have multiple full screen render targets. A 1080p render target is much larger than a 720p target.
Funding game development is a gamble for publishers. Publishers make most of their money on the mega-hits which sell millions of units. There are only a few such games a year. To put this in perspective, most games probably sell less than 200,000 units across multiple platforms. Adding minor platforms adds cost (internal and external QA, code maintenance, potential design comprimises, marketing) but won't have a huge impact on sales.
Sim-ship with consoles can easily triple sales (or more). I doubt the linux and mac sales would match the PC sales. Publishers and devs focus on where the money is. Generally, linux and mac ports are done by other companies with lower burn rates (if your team is 3/4 artists, you need them to move on to the next title asap to keep money coming in - holding up for engineers for a few months while porting to alternate platforms can mess up the next projects schedule).
As a result, even if an engine is completely cross platform, mac and linux releases are simpler more difficult to financially justify. Thats part of the reason why PC games are hurting right now - Windows PC game sales are frequently dwarfed by console sales. Publishers put funding where they have the chance to see the highest payoff.
C++ performance is nowhere near that of C.Yes, if you write C++ while ignoring performance.
If you pay attention, it may be higher performance - a few C-only game dev shops switched to C++ as the compiler optimizations are more advanced. Yes, you can then write C in a C++ compiler. Guess what this suggests about the comparitive performance of the languages?;)
I'd love to see an open source solution, but it is a rather massive undertaking as you would be competing with packages that can cost up to a million a pop. Normally, open source has an advantage here.
As the publisher generally buys/provides funds to license the engine, low risk is much more important than cost. Marketing alone may cost 10 times as much as the engine. On the dev side, content creation is much bigger bottleneck than tech. Tools are massively important - potentially much more so than the engine.
Here are a few notes whoever is looking to compete in the engine market. You should consider:
- a cross platform engine that runs on all current consoles as well as windows and linux - exporters (from max, maya, photoshop, etc) - shader construction tools - world construction tools - sound tools - a cross platform dedicated server app - animation/behavior constructions tools - a game framework (object serialization, reflection framework, etc) - content packers - an unified asset pipeline to handle all platforms - an automated build system - a localization pipeline and associated tools - a test pipeline - support libraries (physics, input, networking, etc) that work on all platforms. - a major shipped title to demonstrate functionality - high quality in game artwork to demo functionality - hooks for a web portal/online stats tracking database (abstracted for each platform) - a runtime optimized for minimal memory allocation (ie don't hit malloc each frame), cache efficiency (ie one of the biggest performance hits on newer platforms), and seek times (ie don't seek on a DVD, it can slow your loads by 200ms to a second) - good use of multithreading (including on the PS3, where you have 6 'threads' with access to 256k data at a time)...and if all of this is MIT style license (or something other than GPL/LGPL: - it will have to be linked with closed source code to run on some platforms - most publishers are nervous about legal issues with GPL unless someone external handles legal protection - devs are hesitant to release all engine code - multiplayer frequently send too much information to the client/relying on encryption/other techniques to minimize cheating - use of third party libraries that can't be distributed/etc)
The lighting model selection was a design call, not crappy coding. Someone at id (or Activision) decided that requiring high end machines was acceptable to be one of the first games with per-pixel lighting. id games traditionally sold depite having high requirements - people upgraded their machines to play them. Benefit? Their games sell partly because they push the tech. It give the game a marketing and design hook.
Valve choose to avoid high tech lighting solutions and go with a traditional that scaled better. Again, a game design/marketing call, not a code quality issue.
Or perhaps Microsoft didn't want to support two separate APIs? If someone really wanted GL on the 360, they could write a wrapper layer over the top of directx.
Of course, Sony went a different route and put out a custom graphics API (someone said they had a GL layer sitting on top of it, but most devs I've talked with are using the lower level API due to efficiency concerns).
Even if the game was ported voluntarily and distributed over the internet, the publisher would still have to provide some degree of tech support. This could easily burn through 10-20k the publisher would see, at which point shipping potentially costs them money (continuing to provide support) or reputation (dropping support for the platform).
In addition, this could count as a separate sku, meaning significant costs licensing third party libraries.
Mod up the parent - this is dead on. It is a strong chip for very specific types of processing. It is much more difficult to leverage the SPU power when working with large data sets or large amounts of branching.
Just rumours, but I've heard that the PS3 cell SPUs aren't IEEE 754. Assuming this is true, would that make them less useful for the academic world, or is this mostly a portability issue?
Most devs are focused on what they can control - ie making quality titles. Protecting IP value is more of a big-picture strategic issue.
I am not saying devs are naive/don't care about their futures. Trust me, I am a paranoid game dev.:) We just have more immediate, concrete issues to worry about like hitting the next milestone/shipping the title/keeping up with tech changes/etc.
Hmm. Perhaps someone should make a Starcraft FPS mod. Starcraft is an RTS after all, no way an FPS would impact them. Starcraft: Ghost has a nice ring to it.;) With Bungies RTS background, this may well consider making an RTS in the future (if RTS sales start increasing again some time).
I agree, the risk/reward separation can be messy. Fan work can definitely support a product/IP. At the same time, the Battle Field 'Desert Combat' mod was in a position to undercut BF2. Slightly different situation, but really good mods can make it difficult to sell a competing product. This isn't a bad thing for the consumer, but for the who is responsible to shareholders to protect their future potential...
Regardless, my initial point is that modders have known for a long time that using licensed IP for a mod is dangerous. This shouldn't have come as a surprise to anyone on the team making it.
The mod community has aware of the risks of this type of mod since the infamous 'Foxing' of the Aliens vs. Predator mod in 94. If you use someones elses IP and they get wind of it, you have a 90% chance of getting shut down. Best case scenario, you get to finish the product but they own it/have massive say in what happens.
To be fair, these mods do undercut the future marketability of franchise products. Lets say they did pull off an extremely popular, successful Halo RTS game - now Bungie just has more competition. Going the other route, if make a notorious bomb of a mod, Halo could end up with a black mark in the RTS arena.
The ironic thing is, chances are decent that the people who actually work on the game day to day probably don't care, as devs aren't typically focused on the long term/5-10 year value of the IP.
Slashdot Mirror
While I have a great deal of respect for the author, this doesn't help quite a few of the companies (and PC gamers) out there.
Basically, the position 'we will only attempt to sell to people who would prefer to buy over download' doesn't scale to big budget titles. There are a lot of gamers out there who like AAA, content rich games. These are the games that need to sell a million+ units just to break even. Ignore the programming - some of these games have dozens of artists and designers working for multiple years.
The 'make niche games' position doesn't help these developers (or the gamers who love their games). We're talking about shops like Valve and Relic here.
The game industry is certainly eyeing whats going on in the movie and music industries. The basic truth is that most people would rather download for free than vote on what they want to see in the future by buying it.
Consumers demands for content rich games is exceeding sales. This means that big blockbuster titles are likely take a hit similar to flight sims several years back. For some gamers, this is probably great. There are plenty out there who would love to see the death of the FPS/action genre if it means a few smaller games come out in their place.
Its going to be a rough few years as big devs figure out how to stay in buisness. Its likely to drive the 'big/blockbuster' titles even more towards the consoles which big markets and lower piracy rates so far this generation.
Can you expand on the pieces you view as biased? Most of it looks rather straight forward. They explained that bsd style licenses grants a freedom, explained why this freedom can be beneficial, and explained what you lost by (downstream openness) by selecting the license.
That seems like a pretty simple and blunt explaination. Each license has pros and cons depending on need.
Thats what happens when one company signs a deal with another - they get to set up their own licensing rules. Valve probably paid id a whole bunch of money back when they started re-licensing HalfLife tech. At that point, they probably removed the license notices as if they were reselling it, they must have had rights to do so.
If you link against LGPL, you need to release object files. This allows people to relink your app against new library versions or alternate lib implementations. GPL requires releasing your source.
Please tell me the government has tools to automate something like this.
Some of these companies are stuck in a hard spot. With GPL 3, they are beening told to give back more than they were previously.
Sure, they shared their modifications and allowed others to make use of their code, they just protected their hardware. In many cases, this protection is either legally required or is required by their buisness model. For example, I'm the MPAA was rather concerned about TIVO. As they were a closed platform, they could work with MPAA to find a balance between freedom for the customer and content protection.
The PS3 is in a similar position. Sonys buisness model is to sell hardware at either a loss or little profit and then make money selling games. If the platform must be open, they are no longer in control of what is released on it.
And yes, it could be argued that their buisness model is invalid as it relies on restricting end user freedom. If they are forced to sell the consoles for a profit, they could easily be priced out of the market. Closed source competitors would have a significant buisness model advantage as they could sell their hardware could be much cheaper.
So the argument moves to 'well, they don't have to use GPL'. And again, thats valid. But it does mean there is one less company potentially contributing changes back to the community, instead likely funding a closed source alternative.
Feel the same way about digital books?
Whats the difference between paper books and software anyway? Books cost a bit more to duplicate, but can be made easily these days.
A single data point isn't statistically useful, but I graduated from a Bay Area high school in 1998. I didn't see any of the nonsense in the news at school. Most of the teachers were high quality, cared about education, and weren't afraid to fail students.
The only stereotypical 'California' issues I recall involved racial incidents. There a continuous paranoia about gangs and 'bad' teens from neighboring cities.
Overall, education is always what the students make of it. Students who pushed themselves opened doors.
I'm not arguing that used games are bad buisness for GameStop. Its definitely lucrative. Long term though, I don't know if it is good for anyone.
Publishers will figure out how to break the model (via DRM, contact-the-server, strong arming, etc). This will EB/GameStop who depend on this as a major income stream. If Gamefly gets much bigger, they could potentially become a target as well.
Gamers will be hurt as stores close and they have to deal with worsening anti-resale measures. They are already getting hit by decreasing game variety. The game industry isn't heading for a crash, but a re-adjustment looks likely. Game creation is simply too expensive for publishers to ignore missed sales.
In the end, its a short term win for GameStop but the long term impact could be messy.
No, their secret weapon is game resales. They buy games back for a few dollars and resell them for just under full price. They actively push used copies, putting them at the front of the store and offering used copies to people buying new.
It is hitting certain game styles enough (primarily single player, plot driven games) that fewer are being made. Offline games pushed to build in 'contact the server before you play' model, requiring an internet connection to play single player games. Or publisher based online distrobution, where transfering ownership costs money if it is possible at all.
So, they are estimating they wrote 120 million lines of new code for the NVidia drivers?
Fine, so I am sure they were exaggerating at 6 drivers, 20 million lines each. At the same time, how the heck does a driver take 20 million lines in the first place? Most next gen game and tools combined, running on several platforms, are less than 2 million lines.
I'd love to know why their drivers are that huge.
A huge percent of game bugs are due to sound driver issues. The first suggestion for most crashes/framerate shuttering/sound stuttering in XP games was to reduce or disable hardware sound settings.
;)
In addition, hardware sound frequently increases the load on the CPU. No idea why, but you can go from 5% of the CPU dedicated to sound without acceleration to 25% dedicated with it. If the sound card driver authors got built reliable, stable drivers (and hardware - the difference between the quality of the SB Live hardware revs is stunning), I doubt MS would have dropped hardware sound support.
Unfortunately, I think MS had to do this. Or they could have pushed for open source audio drivers.
Because we want to run our game on the 360?
Because the PS3, while it supports a version of GL, works better with you use the layer under it?
Because DX has a more consistent interface while GL depends more heavily on vendor specific extensions?
Because DX has a greater critical mass (ie simpler to hire expert, better vendor support, better OS support on the primary gaming OS)
Frankly, porting a modern game to GL is cheap compared to game dev. Making a game might cost 10M, not including marketing, etc. Publishers want to make lots of money. If the potential revenue was worth it, more games would be ported.
I think this clause causes issues with LGPL for many software devs:
If you link other code with the library, you must provide complete object files to the recipients, so that they can relink them with the library after making changes to the library and recompiling it.
Most closed source shops aren't willing (or able) to provide object files. This has pushed many libraries that want wide spread adoption to less restrictive licenses. Or it requires the LGPL'd portion to be a dynamic library instead of a static library.
For example, Ogre is LGPL. They recommend not using it as a static library to avoid this clause. Unfortunately, console games historically can't use dlls (no OS support, though the 360 now allows it). They also can't release object files as the compilers are proprietary. Its all a dead issues anyway, as people can't update the executable. A BSD license avoids the issue entirely.
1) AA is a 360 cert requirement. Yes, a title or two may be able to get that waived by having something else special/that does the same thing. In general though, MS won't let 360 titles with jaggies ship. I am not sure what you are talking about here.
2) Yes, the EDRAM takes extra effort to work with. Sure, it would be nice if there was an infinite amount. In the end though, the tiling work generally impacts the rendering programmer for a short period and no one else. There is plenty of info for 360 devs on how to use it. If this is a devs main 360 complaint, MS is making devs lives easy - providing a good system, tools and documentation.
3) Marketing will always use renders. Their job is to get people excited about the title in a cost/time effective way - staged screenshots and renders are the fastest way to do that. This generation, difference in quality is much more visible in movies. A few marketing departments will use rendered movies here, but the cost is basically quite high relative to a rendered model.
The main 1080p challenge is the performance hit you get for drawing that many more pixels. For games that are GPU bound, going from 720p to 1080p hits the pixel shaders hard. The other issue is memory (same on both platforms) as many games have multiple full screen render targets. A 1080p render target is much larger than a 720p target.
Funding game development is a gamble for publishers. Publishers make most of their money on the mega-hits which sell millions of units. There are only a few such games a year. To put this in perspective, most games probably sell less than 200,000 units across multiple platforms. Adding minor platforms adds cost (internal and external QA, code maintenance, potential design comprimises, marketing) but won't have a huge impact on sales.
Sim-ship with consoles can easily triple sales (or more). I doubt the linux and mac sales would match the PC sales. Publishers and devs focus on where the money is. Generally, linux and mac ports are done by other companies with lower burn rates (if your team is 3/4 artists, you need them to move on to the next title asap to keep money coming in - holding up for engineers for a few months while porting to alternate platforms can mess up the next projects schedule).
As a result, even if an engine is completely cross platform, mac and linux releases are simpler more difficult to financially justify. Thats part of the reason why PC games are hurting right now - Windows PC game sales are frequently dwarfed by console sales. Publishers put funding where they have the chance to see the highest payoff.
C++ performance is nowhere near that of C.Yes, if you write C++ while ignoring performance.
;)
If you pay attention, it may be higher performance - a few C-only game dev shops switched to C++ as the compiler optimizations are more advanced. Yes, you can then write C in a C++ compiler. Guess what this suggests about the comparitive performance of the languages?
I'd love to see an open source solution, but it is a rather massive undertaking as you would be competing with packages that can cost up to a million a pop. Normally, open source has an advantage here.
...and if all of this is MIT style license (or something other than GPL/LGPL:
As the publisher generally buys/provides funds to license the engine, low risk is much more important than cost. Marketing alone may cost 10 times as much as the engine. On the dev side, content creation is much bigger bottleneck than tech. Tools are massively important - potentially much more so than the engine.
Here are a few notes whoever is looking to compete in the engine market. You should consider:
- a cross platform engine that runs on all current consoles as well as windows and linux
- exporters (from max, maya, photoshop, etc)
- shader construction tools
- world construction tools
- sound tools
- a cross platform dedicated server app
- animation/behavior constructions tools
- a game framework (object serialization, reflection framework, etc)
- content packers
- an unified asset pipeline to handle all platforms
- an automated build system
- a localization pipeline and associated tools
- a test pipeline
- support libraries (physics, input, networking, etc) that work on all platforms.
- a major shipped title to demonstrate functionality
- high quality in game artwork to demo functionality
- hooks for a web portal/online stats tracking database (abstracted for each platform)
- a runtime optimized for minimal memory allocation (ie don't hit malloc each frame), cache efficiency (ie one of the biggest performance hits on newer platforms), and seek times (ie don't seek on a DVD, it can slow your loads by 200ms to a second)
- good use of multithreading (including on the PS3, where you have 6 'threads' with access to 256k data at a time)
- it will have to be linked with closed source code to run on some platforms
- most publishers are nervous about legal issues with GPL unless someone external handles legal protection
- devs are hesitant to release all engine code - multiplayer frequently send too much information to the client/relying on encryption/other techniques to minimize cheating
- use of third party libraries that can't be distributed/etc)
The lighting model selection was a design call, not crappy coding. Someone at id (or Activision) decided that requiring high end machines was acceptable to be one of the first games with per-pixel lighting. id games traditionally sold depite having high requirements - people upgraded their machines to play them. Benefit? Their games sell partly because they push the tech. It give the game a marketing and design hook.
Valve choose to avoid high tech lighting solutions and go with a traditional that scaled better. Again, a game design/marketing call, not a code quality issue.
Or perhaps Microsoft didn't want to support two separate APIs? If someone really wanted GL on the 360, they could write a wrapper layer over the top of directx.
Of course, Sony went a different route and put out a custom graphics API (someone said they had a GL layer sitting on top of it, but most devs I've talked with are using the lower level API due to efficiency concerns).
Even if the game was ported voluntarily and distributed over the internet, the publisher would still have to provide some degree of tech support. This could easily burn through 10-20k the publisher would see, at which point shipping potentially costs them money (continuing to provide support) or reputation (dropping support for the platform).
In addition, this could count as a separate sku, meaning significant costs licensing third party libraries.
Mod up the parent - this is dead on. It is a strong chip for very specific types of processing. It is much more difficult to leverage the SPU power when working with large data sets or large amounts of branching.
Just rumours, but I've heard that the PS3 cell SPUs aren't IEEE 754. Assuming this is true, would that make them less useful for the academic world, or is this mostly a portability issue?
Most devs are focused on what they can control - ie making quality titles. Protecting IP value is more of a big-picture strategic issue.
:) We just have more immediate, concrete issues to worry about like hitting the next milestone/shipping the title/keeping up with tech changes/etc.
I am not saying devs are naive/don't care about their futures. Trust me, I am a paranoid game dev.
Hmm. Perhaps someone should make a Starcraft FPS mod. Starcraft is an RTS after all, no way an FPS would impact them. Starcraft: Ghost has a nice ring to it. ;) With Bungies RTS background, this may well consider making an RTS in the future (if RTS sales start increasing again some time).
I agree, the risk/reward separation can be messy. Fan work can definitely support a product/IP. At the same time, the Battle Field 'Desert Combat' mod was in a position to undercut BF2. Slightly different situation, but really good mods can make it difficult to sell a competing product. This isn't a bad thing for the consumer, but for the who is responsible to shareholders to protect their future potential...
Regardless, my initial point is that modders have known for a long time that using licensed IP for a mod is dangerous. This shouldn't have come as a surprise to anyone on the team making it.
The mod community has aware of the risks of this type of mod since the infamous 'Foxing' of the Aliens vs. Predator mod in 94. If you use someones elses IP and they get wind of it, you have a 90% chance of getting shut down. Best case scenario, you get to finish the product but they own it/have massive say in what happens.
To be fair, these mods do undercut the future marketability of franchise products. Lets say they did pull off an extremely popular, successful Halo RTS game - now Bungie just has more competition. Going the other route, if make a notorious bomb of a mod, Halo could end up with a black mark in the RTS arena.
The ironic thing is, chances are decent that the people who actually work on the game day to day probably don't care, as devs aren't typically focused on the long term/5-10 year value of the IP.