Thursday, October 14, 2010
hey guys
Ive got alot of Real life issues at the moment so my posting is going to become a little slower for the next few weeks
Tuesday, October 12, 2010
NELL: the Computer that Learns from the Internet
A computer at Carnegie Mellon called NELL thinks that "persistent cookies" are a baked good and that the First Amendment is a musical instrument.
NELL stands for Never Ending Language Learner and it has been reading the internet for the last few months. Since it was switched on, NELL has learned 440,000 iotas of information with an accuracy of about 74 percent. Not bad for a computer, that's better than my GPA in college. NELL deduces whether things are true if the knowledge comes from a trusted source, or if it's corroborated by several lesser sources. Such information is then raised to the level of "beliefs" and put into NELL's permanent database. Unfortunately, some of the things it has learned are hilariously wrong and there is no programming to "unlearn" its beliefs.
For example, NELL knows that cookies are a baked good but then it assumes that persistent cookies are delicious and tasty as well. Klingon is not an ethnic group, despite several sources on the internet claiming that it is.
NELL knows that Risk is a boardgame but it incorrectly believes that security risk is also a boardgame. Maybe it's on to something, I might want to play a game called Security Risk where your armies are malware programs killing viruses... Maybe not.
Hilariously, NELL doesn't do well with the Bill of Rights. It thinks that the First Amendment is a musical instrument, that the Second Amendment is a hobby, and it refused to divulge any information about the Fifth Amendment at all.
The scientists behind NELL expect the learning curve to be quite steep when it comes to gleaning information from the internet. It's also learning quite a bit about the English language from these disparate sources and the mistakes it makes are understandable. "One might expect a nonnative reader of English to make similar mistakes," one researcher said.
Despite the funny misconceptions that NELL has, it's clear that our robot overlords are using the internet against to further it's A.I. development. Recommendation: Wipe the Internet clean before NELL learns everything there is to know.
Starting now----
http://www.escapistmagazine.com/news/view/104299-Meet-NELL-the-Computer-that-Learns-from-the-Internet
NELL stands for Never Ending Language Learner and it has been reading the internet for the last few months. Since it was switched on, NELL has learned 440,000 iotas of information with an accuracy of about 74 percent. Not bad for a computer, that's better than my GPA in college. NELL deduces whether things are true if the knowledge comes from a trusted source, or if it's corroborated by several lesser sources. Such information is then raised to the level of "beliefs" and put into NELL's permanent database. Unfortunately, some of the things it has learned are hilariously wrong and there is no programming to "unlearn" its beliefs.
For example, NELL knows that cookies are a baked good but then it assumes that persistent cookies are delicious and tasty as well. Klingon is not an ethnic group, despite several sources on the internet claiming that it is.
NELL knows that Risk is a boardgame but it incorrectly believes that security risk is also a boardgame. Maybe it's on to something, I might want to play a game called Security Risk where your armies are malware programs killing viruses... Maybe not.
Hilariously, NELL doesn't do well with the Bill of Rights. It thinks that the First Amendment is a musical instrument, that the Second Amendment is a hobby, and it refused to divulge any information about the Fifth Amendment at all.
The scientists behind NELL expect the learning curve to be quite steep when it comes to gleaning information from the internet. It's also learning quite a bit about the English language from these disparate sources and the mistakes it makes are understandable. "One might expect a nonnative reader of English to make similar mistakes," one researcher said.
Despite the funny misconceptions that NELL has, it's clear that our robot overlords are using the internet against to further it's A.I. development. Recommendation: Wipe the Internet clean before NELL learns everything there is to know.
Starting now----
http://www.escapistmagazine.com/news/view/104299-Meet-NELL-the-Computer-that-Learns-from-the-Internet
Monday, October 11, 2010
Playdead Using Human Skull To Test Audio For Limbo Follow Up
The developers at Playdead, the team behind the hauntingly brilliant Xbox Live Arcade hit Limbo really wanted to get their hands on a human skull, and after the surprisingly lengthy reviews process required to secure a human skull in Denmark, the devs finally got one. Why a skull, you ask? To test sound for their new game, of course.
Playdead CEO and co-founder Dino Patti told me this weekend during IndieCade 2010 that not only are the devs hard at work on a follow up to their downloadable masterpiece, which Patti suggested will most assuredly appeal to Limbo fans, but they are doing some rather elaborate experiments with sound for it as well.
Read more: http://g4tv.com/thefeed/blog/post/708002/Playdead-Using-Human-Skull-To-Test-Audio-For-Limbo--Follow-Up.html#readmore#ixzz124NrvlQG
Playdead CEO and co-founder Dino Patti told me this weekend during IndieCade 2010 that not only are the devs hard at work on a follow up to their downloadable masterpiece, which Patti suggested will most assuredly appeal to Limbo fans, but they are doing some rather elaborate experiments with sound for it as well.
Read more: http://g4tv.com/thefeed/blog/post/708002/Playdead-Using-Human-Skull-To-Test-Audio-For-Limbo--Follow-Up.html#readmore#ixzz124NrvlQG
Sunday, October 10, 2010
Friday, October 8, 2010
Starcraft 2 1v1 changes
Since the release of Wings of Liberty, the StarCraft II balance team has been diligently studying how the game is being played, playing it ourselves, and seeing how evenly the three races match up on the battlefield. While we regularly examine Battle.net player data and statistics from all regions, the information we're examining in this blog was pulled exclusively from the North American region for simplicity's sake.
Protoss are played 38.5% of the time.
Terran are played 38.0% of the time.
Zerg are played 23.5% of the time.
These are overall percentages, but they're mirrored in nearly the same exact separation through each of the leagues. This clearly shows that zerg are played less often than the other races. When we look at things like the Top 200, we like to have this data in-hand so we can ensure that it's proportionate to the amount of each race actually being played. We don't want to have a huge chunk of zerg players sitting somewhere further down, unable to rise through the ranks.
Let's take a look at win percentages on a race-versus-race basis. This is something else we look at to see how matchups are faring over many games. These numbers take individual player skill into account, which helps to avoid the 50% win/loss percentage effect that the matchmaking system can impart on straight win/loss ratios.
Win % in Diamond (accounting for player skill)
49.6% win rate for Protoss when fighting Terran.
52.8% win rate for Protoss when fighting Zerg.
49.6% win rate for Terran when fighting Zerg.
Win % in Platinum (accounting for player skill)
56.3% win rate for Protoss when fighting Terran.
47.3% win rate for Protoss when fighting Zerg.
44.5% win for Terran when fighting Zerg.
Win % in Gold (accounting for player skill)
61.0% win rate for Protoss when fighting Terran.
61.1% win rate for Protoss when fighting Zerg.
49.5% win rate for Terran when fighting Zerg.
Win % in Silver (accounting for player skill)
63.6% win rate for Protoss when fighting Terran.
50.7% win rate for Protoss when fighting Zerg.
51.6% win rate for Terran when fighting Zerg.
Win % in Bronze (accounting for player skill)
59.0% win rate for Protoss when fighting Terran.
55.1% win rate for Protoss when fighting Zerg.
45.4% win rate for Terran when fighting Zerg.
As you can see there are some issues with protoss vs. terran in many of the leagues. From our own play experience, as well as feedback from the community, this matches pretty closely with what we're already aware of. We're working on solutions. What we're also aware of is that, while the numbers don't necessarily support the need for zerg changes across all leagues, the feedback from the community as well as our own play experience tells us that improvements are necessary to make zerg matchups feel and play better.
The balance changes in our next patch will primarily focus on improving the zerg.
Here are a few of the changes we currently have planned:
• We're increasing roach range. This will allow roaches to be more effective in large groups, giving the zerg more options in the mid to end game.
• Fungal Growth will now prevent Blink, which will give zerg a way to stop endlessly Blinking stalkers which can be very challenging to deal with in large numbers.
• The Barracks are going to require a Supply Depot, which will impact a lot of early terran reaper pushes.
• The reaper speed upgrade will require the Factory, which is meant to weaken a lot of the early terran reaper attacks that dominate so many matches, especially in team games.
• We're making a number of increases to the health of zerg buildings, which will make the very vulnerable zerg technology structures more resistant to raids. We don't expect these hit point changes to have a super significant impact on the game, but the current numbers felt way too low.
We want our avid and talented StarCraft II players to know that we're here, we're listening, and that our intent is to continue making careful and measured approaches to balance based on community and fansite feedback, our numbers and data, watching pro players and tournaments, and our own time on Battle.net playing the game alongside you.
Protoss are played 38.5% of the time.
Terran are played 38.0% of the time.
Zerg are played 23.5% of the time.
These are overall percentages, but they're mirrored in nearly the same exact separation through each of the leagues. This clearly shows that zerg are played less often than the other races. When we look at things like the Top 200, we like to have this data in-hand so we can ensure that it's proportionate to the amount of each race actually being played. We don't want to have a huge chunk of zerg players sitting somewhere further down, unable to rise through the ranks.
Let's take a look at win percentages on a race-versus-race basis. This is something else we look at to see how matchups are faring over many games. These numbers take individual player skill into account, which helps to avoid the 50% win/loss percentage effect that the matchmaking system can impart on straight win/loss ratios.
Win % in Diamond (accounting for player skill)
49.6% win rate for Protoss when fighting Terran.
52.8% win rate for Protoss when fighting Zerg.
49.6% win rate for Terran when fighting Zerg.
Win % in Platinum (accounting for player skill)
56.3% win rate for Protoss when fighting Terran.
47.3% win rate for Protoss when fighting Zerg.
44.5% win for Terran when fighting Zerg.
Win % in Gold (accounting for player skill)
61.0% win rate for Protoss when fighting Terran.
61.1% win rate for Protoss when fighting Zerg.
49.5% win rate for Terran when fighting Zerg.
Win % in Silver (accounting for player skill)
63.6% win rate for Protoss when fighting Terran.
50.7% win rate for Protoss when fighting Zerg.
51.6% win rate for Terran when fighting Zerg.
Win % in Bronze (accounting for player skill)
59.0% win rate for Protoss when fighting Terran.
55.1% win rate for Protoss when fighting Zerg.
45.4% win rate for Terran when fighting Zerg.
As you can see there are some issues with protoss vs. terran in many of the leagues. From our own play experience, as well as feedback from the community, this matches pretty closely with what we're already aware of. We're working on solutions. What we're also aware of is that, while the numbers don't necessarily support the need for zerg changes across all leagues, the feedback from the community as well as our own play experience tells us that improvements are necessary to make zerg matchups feel and play better.
The balance changes in our next patch will primarily focus on improving the zerg.
Here are a few of the changes we currently have planned:
• We're increasing roach range. This will allow roaches to be more effective in large groups, giving the zerg more options in the mid to end game.
• Fungal Growth will now prevent Blink, which will give zerg a way to stop endlessly Blinking stalkers which can be very challenging to deal with in large numbers.
• The Barracks are going to require a Supply Depot, which will impact a lot of early terran reaper pushes.
• The reaper speed upgrade will require the Factory, which is meant to weaken a lot of the early terran reaper attacks that dominate so many matches, especially in team games.
• We're making a number of increases to the health of zerg buildings, which will make the very vulnerable zerg technology structures more resistant to raids. We don't expect these hit point changes to have a super significant impact on the game, but the current numbers felt way too low.
We want our avid and talented StarCraft II players to know that we're here, we're listening, and that our intent is to continue making careful and measured approaches to balance based on community and fansite feedback, our numbers and data, watching pro players and tournaments, and our own time on Battle.net playing the game alongside you.
Wednesday, October 6, 2010
Subscribe to:
Posts (Atom)