The letter MS Peru wrote (If I get a chance, I'll post a translation later) regarding this bill uses all the tricks we've seen them use in the USA. The letter intentionally misunderstands provisions, disregards inconvenient legal precident when useful and adopts a hardline legalist (v. moralist) attitude when useful, and makes bombastic claims about the dire consequences of even considering OSS/Free Software.
Hopefully Microsoft's rather weaker hold on the Peruvian government will allow them to get some reasonable guidelines in place so they don't get screwed like the US government.
I agree that 3ware makes excellent products. One slight nitpick though. The Adaptec 2400A beats 3ware's 7450 in some RAID level and test combinations, though it is limited to 2 channels whereas 3ware offers the 7850, with 8 channels.
That depends what RAID level you're talking about. Striping will double both read and write speeds but halve reliability. Mirroring does nothing for write speeds but doubles read speed (if done right), and squares your MTBF. IE, if 1 drive has 100,000hr MTBF, two mirrored drives have 100,000 * 100,000 = 10 billion hours MTBF. The RAID controller however will likely fail *long* before both drives do simultaneously, and systems to handle controller failure without interruption are *expensive*.
I don't mean to jump on you in particular, but/. posters in general seem pretty clueless about storage issues. If you're honestly interested, check out storagereview.
Same with LCDs. LCD pixels don't die off like CRT pixels do - they stay at full brightness until it's time to change. However, the screen *is* refreshed at a certain rate. LCDs are poorer at fast motion than CRTs because the minimum rise/fall time on an LCD pixel is slower than the die-off time on CRTs so a pixel can't go from white to black quickly as something moves across the screen. This is important to the refresh rate question because a refresh rate that exceeds the ability of the LCD to register the changes is wasted. Most LCDs quote in the 60-80Hz range for refresh, but the pixels can only respond to ~40 refreshes per second.
So, to recap: High refresh rates aren't necessary on an LCD to maintain a stable picture, but LCDs *DO* refresh and this needs to be taken into account.
Depends what you mean. There are physically 1600 pixels by 1200 pixels, so if you want 1280x1024, some of the logical pixels will be mapped to 1 physical pixel, some to 2. Now, there's various blurring algorithms to consider, but it still looks bad. You could do 800x600 perfectly (each logical pixel = 4 physical pixels), but why would you want to?
You bitch at me and *I* get the -1 troll? Well fuck you too.
The low-end powerbook used to be $2299 and is now $2499. The midrange one also went up $200, and the high-end is $100 more. Sure, you get a better product - that's what "refresh" means. Still, the prices went up - you can't get a powerbook for $2299 anymore. That's all I was mentioning. I don't know whether it's a less-public reaction to rising LCD prices or whether Apple's selling Tibooks as fast as they can make them and feel confident asking for more profit, or what.
The old midrange product, which you're referring to as "high end", is not the new low-end - it had half a gig before and is now down to 256M and Apple wants $200 to boost it up to 512 again.
The URL Google gave me has an IP number rather than xxxx.google.com so it looks a little suspicious. If you're worried I'm sending you off to goat sex, do a Google search for "apple store powerbook" and take the second result.
Alright. The DoJ clearly isn't doing a good job. The states' case is just going to be appealed anyway and likely won't do a very good job either. Why don't we just apply the same standards you and I would be held to? Let's get together a jury of Microsoft's peers. Let's see here:
1) Microsoft is an OS vendor. Sun, Be (what's left of 'em), and Apple ought to be there. 2) Microsoft is an office apps vendor. Lotus might like a seat. 3) Microsoft is a video game console vendor. I'm sure Sony and Nintendo have some choice words. 4) Microsoft provides internet service. Let's add AOL/TW. 5) Microsoft provides a web server, a database, a mail server, and other such apps. Let's get someone from the Apache foundation, Oracle, Sendmail, and what the hell, the Samba team too. 6) Microsoft writes a lot of buggy code, so let's get an old Netscape exec in too to round out our dozen.
I'll bet we'd see some substantive remedies then!
Before you complain that Be is hardly a peer of Microsoft, consider how 12 upper-middle-class white folks can be considered peers of a poor black woman.
Lesse here - hard drive capacity is growing at what, 60% per year? The largest consumer hard drive right now is 160G. So, that gives us: 2002: 160G 2003: 256G 2004: 410G 2005: 655G 2006: 1.0T 2007: 1.6T 2008: 2.6T 2009: 4.3T 2010: 6.8T 2011: 11.0T 2012: 17.5T 2013: 28.0T
So, if history is any indication, it will be somewhat more than 10 years (just under 15) before 100TB drives become available at the consumer level.
Despite being able to surmount what were once thought to be intractable magnetic effects limiting hard drive density, I don't believe hard drives will make it past a terabyte or so. We are quickly approaching the point where the energy involved in changing a 1 to a 0 (and yes, I realize that with common data encoding schemes, it's not that simple) is less than the thermal energy present in the system. Just as fickle electrons are being replaced by photons in data transport, I think they will replace electrons in data storage as well. Photons leave each other alone so you can pack them more densely; they're low-power; they're resistant to external forces such as electric and magnetic fields and various forms of radiation.
I'm confident there will be 100TB storage devices commercially available within 15-20 years (to give myself a little wiggle room), but if it's based on spinning disks of any kind, and one of you can find me, I'll eat my shirt.
I just might get a first post:) Probably not, since I'm going to put some content here, but it's a nice thought.
I must say I'm a little torn on the issue. It's great if the technology can grow livers and hearts and kidneys and bone marrow and what have you, but I'm not sure there's any good reason to clone entire people. It'll be interesting to see how this all plays out, in any event.
Why don't we just skip all this inbetween crap and go straight to that [single-celled beasties]?
Nasty little buggers have DNA in them that has to be filtered out. Of course, muscle tissue has DNA in it too, but the ratio of DNA to usable protein and other nutrients is much lower, and one can get an actual meal without stones (I think that's what it is you're trying to avoid here...) if it's made from meat rather than single-celled organisms. I assume that's what you meant by "single-celled proteins" which is obviously wrong. One way to tackle the problem is to take the excess DNA out (something about a gram per day max rings a bell), and the other way is to make something without that much in it to start with. A logical way to go about it would be to grow bigger cells (have you seen how much bigger animal tissue cells are than your average bacteria? That's why macrophages can eat 'em up by the hundreds.), and as it turns out meat is something people want and it has nice big cells too.
The real trouble comes when trying to determine which of the spam that says I can opt-out actually means it, and which of the spam is just harvesting/validating my address.
For the final output image that is fine. The human eye can (on average) distinguish just north of 9 bits per color component. The problem is rounding and other systematic errors pushing past the 9 bit mark and becoming visible. This is not useful for your typical computer monitor which can't display 24bit color right anyway, but other output formats have enough quality that the error (=lack of precision) would be noticeable.
I can't comment about software limitations in Maya et al.
Katz is once again missing the point. So what if most people throw their tech out when it breaks? When my pants break, I send them to good will and buy new ones. A friend of mine is a seamstress and fixes broken clothes and frequently makes them in the first place. She's not so good with the computer though. And I don't see any reason why she should be. There is too much information in the world for one person to be knowledgeable about everything. The best you can hope for is to know a little about alot and a lot about a little and try to surround yourself with people whose lots correspond to your littles.
The same patters that Katz is bemoaning in the tech industry are true of clothes (I can probably sew a button on, but anything more complicated than that and I'm lost) and cars (I can change a tire, but that's probably about it) and books (if it's not in English or Spanish, I'm lost) and bridges (I drive over them. They don't break) and . . . you get the idea.
Why download files of suspicious origin and quality from someone who might go offline in the middle of your download, when you can get them from friends who know what they're doing? I used a P2P client whose name I can't even remember anymore once but it sucked for those very reasons. I have a friend who runs a fileserver with about 50,000 tracks on it. They're all well-labeled, have ID3 tags, are encoded at good bit rates with good encoders, and he's not going offline without warning people first. Only friends have accounts on the machine, and he accepts logins only through SSH and file transfers only through SCP. There's no comparison between the level of service he provides and what a P2P client provides.
P2P tools are just that. Tools. Like FTP, SCP, ICQ file transfer, AOL file transfer, &c. Their existence does not create piracy - it is just another way to do it. Resnet here experiences massively more traffic due to kazaa and audiogalaxy than FTP and SCP and I expect this is generally true. Combined with the fact that there's no money behind them, they are easy targets for the huge media companies. If AOL/TW and thee RIAA members were really serious, they'd sue AOL/TW and Microsoft too.
I'm torn between wanting them to cut it out because it's just silly and wanting them to win and teach people to be a little careful and use encryption. Spreading packets all over the internet with your IP and the names of the copyrighted works you're downloading is just stupid. People are paying attention. My ISP told me flat-out that they've sold their souls (isn't that a good Slashdot phrase?) to Sony (among others, though only Sony was mentioned by name) who analyzes every packet they handle searching for copyrighted works.
100-120 IO/sec? These SCSI drives you're talking about are several years old.
Storagereview reports 120 IO/sec for Western Digital's top IDE drive, the WD1200BB. See Storage Review. A top-end SCSI drive such as the Seagate X15-36lp performs on the order of 360 IO/sec. See Storage Review.
For the interested, the X15 runs about $14/G, and you would need about 100 drives to equal the IO/sec of the RAM drive. That's $1400/G; minimum $60,000, and about 700W power consumption and about 2kW total when you add cooling to that. High quality PC2100 DDR is in the $550-600 range per gig, and about 10 watts after cooling.
A Windows mailer? That looks good? Is fully RFC821/822 comlpiant?
Mulberry! http://www.cyrusoft.com/mulberry. Yup, it even costs money so you know it's good. OR, download a demo and try it yourself. OFFICIALLY supports Windows, Linux, Solaris, and MacOS (incl. OSX).
Was I really that droll? My post was intended to be funny. I can watch (and copy) DVDs from any region, with or without CSS. I can strip macrovision, compress to MPEG4, and archive any video DVD I've come across. I can do all this on commodity hardware I assembled myself, under a free OS, using free software in a lonely shack in North Dakota powered with solar panels I bought as government surplus with cash in Missouri.*
* That too, was an attempt at humor. Please laugh, or at least chuckle or grin.
To clarify, my objection is not "getting less than what I paid for," it's getting a headache for my money. It's having a 100G hard drive that can't be split into two 50G partitions. It's having a 20G partition that's larger than a 20G hard drive. It's moving data from one partition to another, and the partition that's at the end of the drive is too small because the drive is ~7% smaller than it should be.
I will admit to having high standards for usability, but is it really too much to ask that a megabyte of memory = a megabyte of floppy disc space = a megabyte of memory stick / CF / smartmedia / IBM Microdrive = a megabyte of hard drive space = a megabyte of network bandwidth = a megabyte on a CD, DVD, tape, etc.? This is what standards are supposed to be for, damnit!
Actually, what I propose has the following benefits:
1) Hard drive manufacturers get to keep using MB =1EE9.
2) People who know why it sucks that 2^20 is so close to a million will be happy with real disc capacities that can be divided into partitions logically.
3) People with no clue will still be able to compare hard disc sizes without conversion.
The idea is simple: keep reporting the "fake" size but make the hard drives "real" sizes. For example, rather than advertise a 120G hard drive that's really ~111.75G, advertise a 130G hard drive that is really 120G and can be divided into nice-sized partitions. Or, better yet, let's have the full 128G that LBA allows, and if they want to call it 137G, so what?
Lable the box with an asterisk that says "1GB = 2^30 bytes" instead of "1GB = 1 billion bytes" so we know for sure, and presto, the best of both worlds. This can't be such a novel idea that nobody has thought of it before...
And the bitch of it is, he's right. Euler's method works fine if the steps are small compared to the second derivative but when the function you are integrating varies wildly, it becomes horribly inaccurate for reasonable-sized steps.
Example: a sine wave. The derivative of a sine is known, but for this example, it isn't. Instead, we're running through Euler's method. Steps are taken at fixed intervals and the function evaluated. Put those steps 2*pi radians apart, and you'll get an answer that has no bearing on reality. Put your steps pi radians apart, and you'll get zero.
This is why hard collisions are difficult - the forces are near zero while the object is careening towards something, and then they spike up very high very fast and go back to nearly zero very fast. The zero part is uninteresting, but the spike has to be very carefully calculated. If time steps are too coarse, the game will be erratic - most objects will collide very poorly and the few percent of collisions where the sample takes place near the peak of the force curve will accelerate wildly away.
If it weren't finals week, I'd make a graphic showing an original function and that function reconstructed from the derivative found by Euler's method with varying step sizes. It would clearly show the worst thing about Euler's method, namely that error accumulates instead of cancelling.
Slashdot Mirror
The letter MS Peru wrote (If I get a chance, I'll post a translation later) regarding this bill uses all the tricks we've seen them use in the USA. The letter intentionally misunderstands provisions, disregards inconvenient legal precident when useful and adopts a hardline legalist (v. moralist) attitude when useful, and makes bombastic claims about the dire consequences of even considering OSS/Free Software.
Hopefully Microsoft's rather weaker hold on the Peruvian government will allow them to get some reasonable guidelines in place so they don't get screwed like the US government.
I agree that 3ware makes excellent products. One slight nitpick though. The Adaptec 2400A beats 3ware's 7450 in some RAID level and test combinations, though it is limited to 2 channels whereas 3ware offers the 7850, with 8 channels.
That depends what RAID level you're talking about. Striping will double both read and write speeds but halve reliability. Mirroring does nothing for write speeds but doubles read speed (if done right), and squares your MTBF. IE, if 1 drive has 100,000hr MTBF, two mirrored drives have 100,000 * 100,000 = 10 billion hours MTBF. The RAID controller however will likely fail *long* before both drives do simultaneously, and systems to handle controller failure without interruption are *expensive*.
/. posters in general seem pretty clueless about storage issues. If you're honestly interested, check out storagereview.
I don't mean to jump on you in particular, but
Same with LCDs. LCD pixels don't die off like CRT pixels do - they stay at full brightness until it's time to change. However, the screen *is* refreshed at a certain rate. LCDs are poorer at fast motion than CRTs because the minimum rise/fall time on an LCD pixel is slower than the die-off time on CRTs so a pixel can't go from white to black quickly as something moves across the screen. This is important to the refresh rate question because a refresh rate that exceeds the ability of the LCD to register the changes is wasted. Most LCDs quote in the 60-80Hz range for refresh, but the pixels can only respond to ~40 refreshes per second.
So, to recap: High refresh rates aren't necessary on an LCD to maintain a stable picture, but LCDs *DO* refresh and this needs to be taken into account.
Depends what you mean. There are physically 1600 pixels by 1200 pixels, so if you want 1280x1024, some of the logical pixels will be mapped to 1 physical pixel, some to 2. Now, there's various blurring algorithms to consider, but it still looks bad. You could do 800x600 perfectly (each logical pixel = 4 physical pixels), but why would you want to?
You bitch at me and *I* get the -1 troll? Well fuck you too.
The low-end powerbook used to be $2299 and is now $2499. The midrange one also went up $200, and the high-end is $100 more. Sure, you get a better product - that's what "refresh" means. Still, the prices went up - you can't get a powerbook for $2299 anymore. That's all I was mentioning. I don't know whether it's a less-public reaction to rising LCD prices or whether Apple's selling Tibooks as fast as they can make them and feel confident asking for more profit, or what.
The old midrange product, which you're referring to as "high end", is not the new low-end - it had half a gig before and is now down to 256M and Apple wants $200 to boost it up to 512 again.
Compare Google's cache of Apple's old Powerbook page and Apple's new Powerbook page. They've raised the prices, along with refreshing their product line.
The URL Google gave me has an IP number rather than xxxx.google.com so it looks a little suspicious. If you're worried I'm sending you off to goat sex, do a Google search for "apple store powerbook" and take the second result.
Alright. The DoJ clearly isn't doing a good job. The states' case is just going to be appealed anyway and likely won't do a very good job either. Why don't we just apply the same standards you and I would be held to? Let's get together a jury of Microsoft's peers. Let's see here:
1) Microsoft is an OS vendor. Sun, Be (what's left of 'em), and Apple ought to be there.
2) Microsoft is an office apps vendor. Lotus might like a seat.
3) Microsoft is a video game console vendor. I'm sure Sony and Nintendo have some choice words.
4) Microsoft provides internet service. Let's add AOL/TW.
5) Microsoft provides a web server, a database, a mail server, and other such apps. Let's get someone from the Apache foundation, Oracle, Sendmail, and what the hell, the Samba team too.
6) Microsoft writes a lot of buggy code, so let's get an old Netscape exec in too to round out our dozen.
I'll bet we'd see some substantive remedies then!
Before you complain that Be is hardly a peer of Microsoft, consider how 12 upper-middle-class white folks can be considered peers of a poor black woman.
IE in Mac OS X follows the standards a lot better than IE in Windows.
Including, as I hear it, the Microsoft-standard crash factor.
Lesse here - hard drive capacity is growing at what, 60% per year? The largest consumer hard drive right now is 160G. So, that gives us:
2002: 160G
2003: 256G
2004: 410G
2005: 655G
2006: 1.0T
2007: 1.6T
2008: 2.6T
2009: 4.3T
2010: 6.8T
2011: 11.0T
2012: 17.5T
2013: 28.0T
So, if history is any indication, it will be somewhat more than 10 years (just under 15) before 100TB drives become available at the consumer level.
Despite being able to surmount what were once thought to be intractable magnetic effects limiting hard drive density, I don't believe hard drives will make it past a terabyte or so. We are quickly approaching the point where the energy involved in changing a 1 to a 0 (and yes, I realize that with common data encoding schemes, it's not that simple) is less than the thermal energy present in the system. Just as fickle electrons are being replaced by photons in data transport, I think they will replace electrons in data storage as well. Photons leave each other alone so you can pack them more densely; they're low-power; they're resistant to external forces such as electric and magnetic fields and various forms of radiation.
I'm confident there will be 100TB storage devices commercially available within 15-20 years (to give myself a little wiggle room), but if it's based on spinning disks of any kind, and one of you can find me, I'll eat my shirt.
Nah. If Sony forced you to buy 7 games with the PS2, we'd all be screaming. If Microsoft does it . . .
Did it really take me 12 minutes to type that?
I just might get a first post:) Probably not, since I'm going to put some content here, but it's a nice thought.
I must say I'm a little torn on the issue. It's great if the technology can grow livers and hearts and kidneys and bone marrow and what have you, but I'm not sure there's any good reason to clone entire people. It'll be interesting to see how this all plays out, in any event.
Why don't we just skip all this inbetween crap and go straight to that [single-celled beasties]?
Nasty little buggers have DNA in them that has to be filtered out. Of course, muscle tissue has DNA in it too, but the ratio of DNA to usable protein and other nutrients is much lower, and one can get an actual meal without stones (I think that's what it is you're trying to avoid here...) if it's made from meat rather than single-celled organisms. I assume that's what you meant by "single-celled proteins" which is obviously wrong. One way to tackle the problem is to take the excess DNA out (something about a gram per day max rings a bell), and the other way is to make something without that much in it to start with. A logical way to go about it would be to grow bigger cells (have you seen how much bigger animal tissue cells are than your average bacteria? That's why macrophages can eat 'em up by the hundreds.), and as it turns out meat is something people want and it has nice big cells too.
The real trouble comes when trying to determine which of the spam that says I can opt-out actually means it, and which of the spam is just harvesting/validating my address.
.procmailrc.
Use your own opt-out list. Call it
For the final output image that is fine. The human eye can (on average) distinguish just north of 9 bits per color component. The problem is rounding and other systematic errors pushing past the 9 bit mark and becoming visible. This is not useful for your typical computer monitor which can't display 24bit color right anyway, but other output formats have enough quality that the error (=lack of precision) would be noticeable.
I can't comment about software limitations in Maya et al.
Katz is once again missing the point. So what if most people throw their tech out when it breaks? When my pants break, I send them to good will and buy new ones. A friend of mine is a seamstress and fixes broken clothes and frequently makes them in the first place. She's not so good with the computer though. And I don't see any reason why she should be. There is too much information in the world for one person to be knowledgeable about everything. The best you can hope for is to know a little about alot and a lot about a little and try to surround yourself with people whose lots correspond to your littles.
The same patters that Katz is bemoaning in the tech industry are true of clothes (I can probably sew a button on, but anything more complicated than that and I'm lost) and cars (I can change a tire, but that's probably about it) and books (if it's not in English or Spanish, I'm lost) and bridges (I drive over them. They don't break) and . . . you get the idea.
Why download files of suspicious origin and quality from someone who might go offline in the middle of your download, when you can get them from friends who know what they're doing? I used a P2P client whose name I can't even remember anymore once but it sucked for those very reasons. I have a friend who runs a fileserver with about 50,000 tracks on it. They're all well-labeled, have ID3 tags, are encoded at good bit rates with good encoders, and he's not going offline without warning people first. Only friends have accounts on the machine, and he accepts logins only through SSH and file transfers only through SCP. There's no comparison between the level of service he provides and what a P2P client provides.
P2P tools are just that. Tools. Like FTP, SCP, ICQ file transfer, AOL file transfer, &c. Their existence does not create piracy - it is just another way to do it. Resnet here experiences massively more traffic due to kazaa and audiogalaxy than FTP and SCP and I expect this is generally true. Combined with the fact that there's no money behind them, they are easy targets for the huge media companies. If AOL/TW and thee RIAA members were really serious, they'd sue AOL/TW and Microsoft too.
I'm torn between wanting them to cut it out because it's just silly and wanting them to win and teach people to be a little careful and use encryption. Spreading packets all over the internet with your IP and the names of the copyrighted works you're downloading is just stupid. People are paying attention. My ISP told me flat-out that they've sold their souls (isn't that a good Slashdot phrase?) to Sony (among others, though only Sony was mentioned by name) who analyzes every packet they handle searching for copyrighted works.
100-120 IO/sec? These SCSI drives you're talking about are several years old.
Storagereview reports 120 IO/sec for Western Digital's top IDE drive, the WD1200BB. See Storage Review. A top-end SCSI drive such as the Seagate X15-36lp performs on the order of 360 IO/sec. See Storage Review.
For the interested, the X15 runs about $14/G, and you would need about 100 drives to equal the IO/sec of the RAM drive. That's $1400/G; minimum $60,000, and about 700W power consumption and about 2kW total when you add cooling to that. High quality PC2100 DDR is in the $550-600 range per gig, and about 10 watts after cooling.
A Windows mailer? That looks good? Is fully RFC821/822 comlpiant?
Mulberry! http://www.cyrusoft.com/mulberry. Yup, it even costs money so you know it's good. OR, download a demo and try it yourself. OFFICIALLY supports Windows, Linux, Solaris, and MacOS (incl. OSX).
Was I really that droll? My post was intended to be funny. I can watch (and copy) DVDs from any region, with or without CSS. I can strip macrovision, compress to MPEG4, and archive any video DVD I've come across. I can do all this on commodity hardware I assembled myself, under a free OS, using free software in a lonely shack in North Dakota powered with solar panels I bought as government surplus with cash in Missouri.*
* That too, was an attempt at humor. Please laugh, or at least chuckle or grin.
Gosh, I guess that means I just won't be able to play it here in the US, with my US-bought DVD-ROM drive, huh?
To clarify, my objection is not "getting less than what I paid for," it's getting a headache for my money. It's having a 100G hard drive that can't be split into two 50G partitions. It's having a 20G partition that's larger than a 20G hard drive. It's moving data from one partition to another, and the partition that's at the end of the drive is too small because the drive is ~7% smaller than it should be.
I will admit to having high standards for usability, but is it really too much to ask that a megabyte of memory = a megabyte of floppy disc space = a megabyte of memory stick / CF / smartmedia / IBM Microdrive = a megabyte of hard drive space = a megabyte of network bandwidth = a megabyte on a CD, DVD, tape, etc.? This is what standards are supposed to be for, damnit!
Actually, what I propose has the following benefits:
1) Hard drive manufacturers get to keep using MB =1EE9.
2) People who know why it sucks that 2^20 is so close to a million will be happy with real disc capacities that can be divided into partitions logically.
3) People with no clue will still be able to compare hard disc sizes without conversion.
The idea is simple: keep reporting the "fake" size but make the hard drives "real" sizes. For example, rather than advertise a 120G hard drive that's really ~111.75G, advertise a 130G hard drive that is really 120G and can be divided into nice-sized partitions. Or, better yet, let's have the full 128G that LBA allows, and if they want to call it 137G, so what?
Lable the box with an asterisk that says "1GB = 2^30 bytes" instead of "1GB = 1 billion bytes" so we know for sure, and presto, the best of both worlds. This can't be such a novel idea that nobody has thought of it before...
And the bitch of it is, he's right. Euler's method works fine if the steps are small compared to the second derivative but when the function you are integrating varies wildly, it becomes horribly inaccurate for reasonable-sized steps.
Example: a sine wave. The derivative of a sine is known, but for this example, it isn't. Instead, we're running through Euler's method. Steps are taken at fixed intervals and the function evaluated. Put those steps 2*pi radians apart, and you'll get an answer that has no bearing on reality. Put your steps pi radians apart, and you'll get zero.
This is why hard collisions are difficult - the forces are near zero while the object is careening towards something, and then they spike up very high very fast and go back to nearly zero very fast. The zero part is uninteresting, but the spike has to be very carefully calculated. If time steps are too coarse, the game will be erratic - most objects will collide very poorly and the few percent of collisions where the sample takes place near the peak of the force curve will accelerate wildly away.
If it weren't finals week, I'd make a graphic showing an original function and that function reconstructed from the derivative found by Euler's method with varying step sizes. It would clearly show the worst thing about Euler's method, namely that error accumulates instead of cancelling.