PoliceAbuse.org investigates illegal behavior by cops. They have quite a track record of collecting video and audio coverage, they've worked with TV stations, and they're having some real effect.
One big problem with SETI work is that it's looking for obsolete forms of radio. SETI@Home's algorithms should be able to detect an AM or FM signal. Maybe TDMA. CDMA, no way. OBIC, probably not. Digital HDTV, probably not. So we're looking for an advanced civilization that uses 1940s radio technology.
Older radio technologies (AM audio, FM audio, analog TV (which is AM video, FM audio)) had a strong "carrier", a big sine wave component. Most of the RF power wasn't really carrying any information. But it was easy to detect the signal. Newer technologies look like noise unless you know what to look for. It's like listening to telephone modems; the data from modern modems just sounds like a hiss. It has the statistics of pure noise unless you know what to look for. Early, low-speed, modems sounded like beeps and warbles, and were easy to identify as modem signals.
Remember, SETI@Home is looking for signals against a very noisy background. You could pick out an AM or FM carrier easily, because you can see it over a large number of cycles. There's a dumb, obvious redundancy in the carrier. But a modern noise-like RF signal against a noisy background is really hard to detect unless you know what you're looking for. If there's redundancy to get through the noise, it's probably more subtle, like data for forward error correction. To even detect that is tough.
It's worth reading the entire order from Judge Wells. However, for the benefit of those who don't enjoy reading legal documents, here's are the highlights. These are the Judge's words:
As repeatedly noted by IBM, concurrent with SCO's court
filed allegations has been SCO's siren song sounding the strength
of its case to the public. At a trade show in 2003 SCO shared
with the public a presentation outlining SCO's claims against
IBM. SCO identified four categories of alleged
misappropriation:... Finally, in the presentation SCO also gave
"one example of many" of line by line copying between the System
V Code and Linux kernel code.14
SCO... was ordered.... to provide and identify with specificity all lines of code in Linux that it claims rights to.
In December 2003, near the beginning of this case, the court
ordered SCO to, "identify and state with specificity the source
code(s) that SCO is claiming form the basis of their action
against IBM." Even if SCO lacked the code behind methods and
concepts at this early stage, SCO could have and should have, at
least articulated which methods and concepts formed "the basis of
their action against IBM." At a minimum, SCO should have
identified the code behind their method and concepts in the final
submission pursuant to this original order entered in December
2003 and Judge Kimball's order entered in July 2005.
SCO was ordered on multiple occasions to answer IBM's interrogatories
which in this court's view covered methods and concepts and a
request for the code behind them. Thus, SCO's failure to provide
code for the methods and concepts it claims were misappropriated
is also a violation of Rule 26(e) in addition to a violation of
this court's orders.
Based on the foregoing, the court finds that SCO has had
ample opportunity to articulate, identify and substantiate its
claims against SCO. The court further finds that such failure
was intentional and therefore willful based on SCO's disregard of
the court's orders and failure to seek clarification. In the 118
view of the court it is almost like SCO sought to hide its case
until the ninth inning in hopes of gaining an unfair advantage
despite being repeatedly told to put "all evidence . . . on the
table."
Accordingly, the court finds that SCO willfully failed to
comply with the court's orders.
Based on the foregoing, the court GRANTS in PART IBM's
Motion to Limit SCO's Claims.
Essentially, the claims of copyright infringment in Linux based on UNIX source code just got thrown out of court. There are a few minor claims remaining, but they're minor and mostly related to old contractual issues that can only involve IBM, not third parties using Linux.
This is all still pretrial manuvering, during which the case becomes better defined.
In the next phase, we have "dispositive motions", which will probably include a motion by IBM for summary judgement against SCO. Some more SCO claims will probably be thrown out at that phase.
"Web 2.0" sites tend to constrained by server load, not outgoing bandwidth. With all that extra server-side work, servers are busier than ever. Notice how many more sites load slowly today. Sometimes you'll even see a page load stall because some JavaScript is waiting for an ad server. (Watch the bottom bar in Firefox to see what you're waiting for.)
For overloaded database-driven sites, page load stalls make the Web look like it's 1997 again. Craigslist is really hurting during busy periods.
If you're in Ohio, write to your state legislator and complain about the "arrogant state employees at Kent State" who think they have the right to muzzle their students.
If you're at Kent State, step one is to register to vote, and get as many other students as you can to register. You know what to do after that.
OK, these "distro makers" are downloading vast amounts of material covered by the GPL for free and then redistributing it for money or advertising. (MEPIS sticks in an Earthlink signup icon, for example.) And then they whine that they have to provide the source for the free stuff they're reselling.
Even worse, some of these distro makers want you to sign up for a "support contract". If they don't have a repository of the source, their support probably isn't worth much.
This ad is so awful that it looks like an attempt to sabotage OpenOffice. The design is terrible, the message is confusing, and it will look even worse on newsprint. Did Microsoft have something to do with this? Anybody competent could do better. The first line ought to read something like this.
Free office software. No purchase necessary. No fees. No subscriptions. Nothing to pay, ever. No ads. No spyware. No limits on use. No limits on copying. No charge for upgrades. No kidding.
How is this possible? OpenOffice is free software, developed by hundreds of users and companies worldwide....
There are some big problems with Blu-Ray. Getting the compression right is hard. I was watching a Blu-Ray demo at the Sony Style store at the Metreon in San Francisco. Now this is Sony gear in a Sony retail store set up by Sony employees playing a Sony demo disk in an environment intended to show the technology at its best. And I'm seeing blocky areas of bright light jumping in the background in a concert video. It looks like the compression algorithm has trouble with camera rotation.
Some of the content looks great; some looks terrible. It's painfully clear that you can't just dump the content into the compressor and expect good results; it's going to become another labor-intensive step in post production, at least for a while.
QNX is no longer pursuing the small-user market, so they don't do that any more.
But the QNX distribution disk is bootable from CD only, without using any hard drive space. This gets you a little memoryless system that will connect to the Internet via Ethernet or PPPoE, and offers a web browser.
You can build a useful mini-distro for QNX today, if you have a development seat with redistribution rights. Make your own Internet appliance, with everything in flash memory or even ROM. The i-opener was such a device.
Bringing back the stateless i-Opener concept could be useful, if it had an Ethernet port instead of a dial-up modem. Good for kiosk systems, public-access Internet terminals, hotel room systems, and other places where you don't want to bother with system administration.
Cell phone assembly tends to be heavily automated, because the devices are designed for it. Here's a cell phone assembly line in Michigan. Cell phones are typically a stack of flat subassemblies sandwiched between the halves of the case, which makes for good automated assembly.
Surface mount components are almost always placed and soldered by machine, but it's quite feasible to insert most through-hole components by machine, too. Wiring harnesses remain a headache. You can design computers without them; the Macintosh IIci, with a power supply that slipped directly into tabs on the motherboard, was a nice example. But you have to integrate case and board design for that to work.
Here's are pictures from a US manufacturer of PC boards. Notice how it's done. No long row of women putting in components; it's one guy standing around watching the machines do the work. Automated insertion machines put in the components, and transfer conveyors connect the machines.
That's the way it should be.
Only the really low wages of China make labor-intensive manual assembly feasible. Even in Mexico, you'd use automated assembly. Assembly in Japan has been automated for decades. If the US imposed import duties on very-low-wage countries that equalized wage costs to even $1/hour, this excessive "offshoring" would stop.
The trend now is towards L1B visas. These were something the US originally lobbied for, to allow Americans working for US multinationals to work abroad. But it works both ways. If a multinational company has operations in the US, they can move employees to their US locations under an L1B visa as an intracompany transfer.
L1B visas offer many advantages for the employer:
There's no "cap"; any number can be issued.
There's no need to advertise the job in the US.
There's no "prevailing wage" requirement.
The employee is totally controlled by the employer; they can't change jobs on an L1B visa, and if they quit, they must leave the US immediately.
Big companies (1000 US employees or more) can apply for a "blanket L1 visa", and then move employees into the US freely.
There's a requirement that the employee had to be employed by the company for a year before coming to the US, and they're supposed to have "specialized knowledge", but that's about it.
So this has become a favorite tool of body shops; hire in India, make them work in a call center for a year, then send them to the US to work cheap.
Well, let's see what they commit to contractually:
Ibackup
Pro Softnet makes no representation or warranty... (C) that the data and files you store in your account will not be lost or damaged.
XDriveXdrive, in its sole discretion, may terminate your password, account or use of the Service and remove and discard any Data within the Service if you fail to comply with this Agreement....
XDrive makes no representation, warranty, or guaranty as to the reliablity, timeliness,
quality, suitability, truth, availability, accuracy or completeness of the service or any content.
FirstBackupSubscriber does not desire this Agreement to provide liability for loss or damage due directly or indirectly to occurrences, or consequences there from, which the service is designed to deter or avert. If subscriber desires additional liability coverage, it shall be the subscriber's sole responsibility to secure it from an insurance carrier or other agency of subscriber's choice, at subscriber's own expense. Subscriber shall bring no suit against FB as a result of any loss arising from this Agreement.
Acpana
The software product and all services are provided "as is" without any warranty or condition
of any kind, either expressed or implied. Use of the software product and all services is at end user's own risk
ElephantDrive
The company's aggregate liability, and the aggregate liability of its licensors, to you or
any third parties in any circumstances is limited to $100.
Mozy
You understand and agree that... you will be solely responsible for any damages to your computer system or loss of data
So, even though some of these outfits make advertising claims like "IT NEVER FORGETS
ElephantDrive uses military-grade encryption and large scale disaster recovery techniques so your data is stored safely for as long as you keep your account.", they don't stand behind those claims. It would thus be inappropriate to trust any of them with important data.
Over at SpecialHam, the forum for bottom-feeder spammers, it seems to be business as usual today. No mention of any crackdown in China. Typical message: "Please give me ICQ UINs of poeple who make installations at trojaned computers. I need to install some software." There's some gloating over the collapse of BlueSecurity. Some new ways to spam Myspace. But no real concerns about enforcement today.
Starting out such conversations with: "This conversation is being recorded for quality control purposes. Anything you say may be used against you in a court of law. Identify yourself, please" is often helpful.
From reading the available documentation and tutorials, it all seems very primitive, sort of like Lego Mindstorms meets ".NET". You can issue commands to actuators and get events from input devices. That's about all you get in robotics functionality. No sign of the hard stuff, like vision processing, map building, GPS/INS integration, motion planning, forward or reverse kinematics, or any serious robotics stuff.
The target (on the robot) apparently has to run at least Windows Server 2003 with ".NET". That's a lot of baggage for something that does so little. You'd expect that the target would run Windows CE or something, but apparently not.
The whole thing is very event-oriented, like Windows, rather than real-time cycle-oriented, like most real robotics applications. It seems to be an extension of Microsoft Web Services. It even uses SOAP.
Their simulator is so weak it can't even handle something with joints. (They bought into Ageia, the "fast but dumb" approach to physics). That's disappointing for 2006. We were doing better a decade ago, and by now, everybody else serious has joints working. Simulation for robotics has to be considerably better than that for games, and most of the game simulators cheat quite a bit to get the speed up. That kills you in a robot simulator.
What all these systems do is quite similar. The basic idea is this:
There's some way to find out what other programs are running that are willing to talk, a "resource location" protocol.
Once you find out who's willing to talk, there's a way to find out what messages they understand and their formats. That's the key to all this.
Then, hopefully, there's some security mechanism which decides who's allowed to make which calls to whom.
Then, the caller and callee can communicate in a subroutine-like way, with some assurance that they're talking the same language.
This all originated as an internal mechanism within Visual Basic and was responsible for its success. Alan Cooper had finally implemented a reasonably idiot-proof way to make A talk to B. That's why Visual Basic controls were so powerful. Then Microsoft made it a general-purpose mechanism. The Unix/Linux world didn't pick up on this for a long time. The Gnome and OpenOffice people get it, and their stuff plays well together. Most other extensibility is a master-slave relationship, usually involving "plug-ins", "toolbars", or "scripting".
Part of the problem is that the main languages in the Unix/Linux world are C and C++, which have no introspection. If you have a structure, and you need to go through all its fields and get their name and type info so you can format and marshall them for transmission, there's no way to do that from within C or C++, other than by writing glue code by hand for each interface. Preprocessors have been built to deal with this problem (ones exist for OpenRPC and SOAP), but they're clunky and complicate the build process. CORBA requires that you define each interface in its own special language, IDL. The Java world has a solution to this problem, but it's Java-only. Microsoft has put extensions into their various dialects of C and C++ to deal with this, but they were too Microsoft-specific to go into the standard.
So making A talk to B without extra work remains hard in the Unix/Linux world.
The intercommunication system within OpenOffice, the mechanism that allows embedding spreadsheets and drawings in other documents, is CORBA-based. Sort of. Actually, it's something called UNO, which started life as CORBA but went off in an XML direction.
GNOME also uses CORBA internally. But its CORBA isn't compatible with the one from OpenOffice.
The UNIX/Linux world has never really had a good way for applications on the same machine to intercommunicate in a subroutine-like way. Microsoft has OLE/COM/DCOM/ActiveX, which is clunky but always available. In the Linux/Unix world, there's nothing you can assume is always there. There's OpenRPC, there's CORBA (in about five incompatible forms), there's Java RMI, and there are various kludges built out of pipes. But there's been no convergence in two decades.
That's Craigslist, making it impossible to compete because of their low, low prices. And it's working. The newspaper industry is furious at being underpriced. Tough.
There may be a big problem looming for Hollywood. If you transfer 35MM film to 1080p, the film grain often shows up. Compressing all that useless film grain noise, which has no frame to frame coherence, will use up a big fraction of the data capacity. It will also mess up the motion compression, which usually results in annoying jaggies. So it's probably necessary to filter out at least some of the film grain. But if you filter out the film grain, you lose resolution.
The reviews of the new Blu-ray disks ("the picture looks too soft and flat") indicates that there's probably too much filtering.
Somewhere in LA, there are probably members of SMTPE struggling with this, trying to figure out the right tradeoffs between resolution loss and compression overload when converting existing films.
Slashdot Mirror
PoliceAbuse.org investigates illegal behavior by cops. They have quite a track record of collecting video and audio coverage, they've worked with TV stations, and they're having some real effect.
They could use some web site design help.
One big problem with SETI work is that it's looking for obsolete forms of radio. SETI@Home's algorithms should be able to detect an AM or FM signal. Maybe TDMA. CDMA, no way. OBIC, probably not. Digital HDTV, probably not. So we're looking for an advanced civilization that uses 1940s radio technology.
Older radio technologies (AM audio, FM audio, analog TV (which is AM video, FM audio)) had a strong "carrier", a big sine wave component. Most of the RF power wasn't really carrying any information. But it was easy to detect the signal. Newer technologies look like noise unless you know what to look for. It's like listening to telephone modems; the data from modern modems just sounds like a hiss. It has the statistics of pure noise unless you know what to look for. Early, low-speed, modems sounded like beeps and warbles, and were easy to identify as modem signals.
Remember, SETI@Home is looking for signals against a very noisy background. You could pick out an AM or FM carrier easily, because you can see it over a large number of cycles. There's a dumb, obvious redundancy in the carrier. But a modern noise-like RF signal against a noisy background is really hard to detect unless you know what you're looking for. If there's redundancy to get through the noise, it's probably more subtle, like data for forward error correction. To even detect that is tough.
There's discussion of that typo on Groklaw. The judge will probably issue a correction. It's not a big deal.
It's worth reading the entire order from Judge Wells. However, for the benefit of those who don't enjoy reading legal documents, here's are the highlights. These are the Judge's words:
Essentially, the claims of copyright infringment in Linux based on UNIX source code just got thrown out of court. There are a few minor claims remaining, but they're minor and mostly related to old contractual issues that can only involve IBM, not third parties using Linux.
This is all still pretrial manuvering, during which the case becomes better defined. In the next phase, we have "dispositive motions", which will probably include a motion by IBM for summary judgement against SCO. Some more SCO claims will probably be thrown out at that phase.
"Web 2.0" sites tend to constrained by server load, not outgoing bandwidth. With all that extra server-side work, servers are busier than ever. Notice how many more sites load slowly today. Sometimes you'll even see a page load stall because some JavaScript is waiting for an ad server. (Watch the bottom bar in Firefox to see what you're waiting for.)
For overloaded database-driven sites, page load stalls make the Web look like it's 1997 again. Craigslist is really hurting during busy periods.
OK. So don't donate to Kent State's athletic scholarship program. Put a note on your website and blog encouraging others not to do so. Write to Tim Hall, Associate Athletic Director for Development at Kent State and tell him what you did. Starve the beast.
If you're in Ohio, write to your state legislator and complain about the "arrogant state employees at Kent State" who think they have the right to muzzle their students.
If you're at Kent State, step one is to register to vote, and get as many other students as you can to register. You know what to do after that.
OK, these "distro makers" are downloading vast amounts of material covered by the GPL for free and then redistributing it for money or advertising. (MEPIS sticks in an Earthlink signup icon, for example.) And then they whine that they have to provide the source for the free stuff they're reselling.
Even worse, some of these distro makers want you to sign up for a "support contract". If they don't have a repository of the source, their support probably isn't worth much.
This ad is so awful that it looks like an attempt to sabotage OpenOffice. The design is terrible, the message is confusing, and it will look even worse on newsprint. Did Microsoft have something to do with this? Anybody competent could do better. The first line ought to read something like this.
Free office software. No purchase necessary. No fees. No subscriptions. Nothing to pay, ever. No ads. No spyware. No limits on use. No limits on copying. No charge for upgrades. No kidding.
How is this possible? OpenOffice is free software, developed by hundreds of users and companies worldwide. ...
There are some big problems with Blu-Ray. Getting the compression right is hard. I was watching a Blu-Ray demo at the Sony Style store at the Metreon in San Francisco. Now this is Sony gear in a Sony retail store set up by Sony employees playing a Sony demo disk in an environment intended to show the technology at its best. And I'm seeing blocky areas of bright light jumping in the background in a concert video. It looks like the compression algorithm has trouble with camera rotation.
Some of the content looks great; some looks terrible. It's painfully clear that you can't just dump the content into the compressor and expect good results; it's going to become another labor-intensive step in post production, at least for a while.
People in politics who don't matter any more:
QNX is no longer pursuing the small-user market, so they don't do that any more. But the QNX distribution disk is bootable from CD only, without using any hard drive space. This gets you a little memoryless system that will connect to the Internet via Ethernet or PPPoE, and offers a web browser.
You can build a useful mini-distro for QNX today, if you have a development seat with redistribution rights. Make your own Internet appliance, with everything in flash memory or even ROM. The i-opener was such a device.
Bringing back the stateless i-Opener concept could be useful, if it had an Ethernet port instead of a dial-up modem. Good for kiosk systems, public-access Internet terminals, hotel room systems, and other places where you don't want to bother with system administration.
Cell phone assembly tends to be heavily automated, because the devices are designed for it. Here's a cell phone assembly line in Michigan. Cell phones are typically a stack of flat subassemblies sandwiched between the halves of the case, which makes for good automated assembly.
Surface mount components are almost always placed and soldered by machine, but it's quite feasible to insert most through-hole components by machine, too. Wiring harnesses remain a headache. You can design computers without them; the Macintosh IIci, with a power supply that slipped directly into tabs on the motherboard, was a nice example. But you have to integrate case and board design for that to work.
Here's are pictures from a US manufacturer of PC boards. Notice how it's done. No long row of women putting in components; it's one guy standing around watching the machines do the work. Automated insertion machines put in the components, and transfer conveyors connect the machines. That's the way it should be.
Only the really low wages of China make labor-intensive manual assembly feasible. Even in Mexico, you'd use automated assembly. Assembly in Japan has been automated for decades. If the US imposed import duties on very-low-wage countries that equalized wage costs to even $1/hour, this excessive "offshoring" would stop.
Right; it started as a reciprocal thing. The US offered L1 visas and encouraged other countries to offer something equivalent.
The trend now is towards L1B visas. These were something the US originally lobbied for, to allow Americans working for US multinationals to work abroad. But it works both ways. If a multinational company has operations in the US, they can move employees to their US locations under an L1B visa as an intracompany transfer.
L1B visas offer many advantages for the employer:
There's a requirement that the employee had to be employed by the company for a year before coming to the US, and they're supposed to have "specialized knowledge", but that's about it.
So this has become a favorite tool of body shops; hire in India, make them work in a call center for a year, then send them to the US to work cheap.
Like any consumer grade service it comes with nothing more than a gentleman's guarantee.
Compare the warranty on your car.
Well, let's see what they commit to contractually:
So, even though some of these outfits make advertising claims like "IT NEVER FORGETS ElephantDrive uses military-grade encryption and large scale disaster recovery techniques so your data is stored safely for as long as you keep your account.", they don't stand behind those claims. It would thus be inappropriate to trust any of them with important data.
Over at SpecialHam, the forum for bottom-feeder spammers, it seems to be business as usual today. No mention of any crackdown in China. Typical message: "Please give me ICQ UINs of poeple who make installations at trojaned computers. I need to install some software." There's some gloating over the collapse of BlueSecurity. Some new ways to spam Myspace. But no real concerns about enforcement today.
Starting out such conversations with: "This conversation is being recorded for quality control purposes. Anything you say may be used against you in a court of law. Identify yourself, please" is often helpful.
Microsoft has a web site for Robotics Studio. There's a free download version.
From reading the available documentation and tutorials, it all seems very primitive, sort of like Lego Mindstorms meets ".NET". You can issue commands to actuators and get events from input devices. That's about all you get in robotics functionality. No sign of the hard stuff, like vision processing, map building, GPS/INS integration, motion planning, forward or reverse kinematics, or any serious robotics stuff.
The target (on the robot) apparently has to run at least Windows Server 2003 with ".NET". That's a lot of baggage for something that does so little. You'd expect that the target would run Windows CE or something, but apparently not.
The whole thing is very event-oriented, like Windows, rather than real-time cycle-oriented, like most real robotics applications. It seems to be an extension of Microsoft Web Services. It even uses SOAP.
Their simulator is so weak it can't even handle something with joints. (They bought into Ageia, the "fast but dumb" approach to physics). That's disappointing for 2006. We were doing better a decade ago, and by now, everybody else serious has joints working. Simulation for robotics has to be considerably better than that for games, and most of the game simulators cheat quite a bit to get the speed up. That kills you in a robot simulator.
For now, Yobotics and Player/Stage remain way ahead.
The U.S. Army also offers free laser eye surgery to soldiers. Preference is given to combat troops. "The bottom line is that if you're in the middle of a fight and you can't see the enemy before they see you, you're dead". The Army has been doing this since 2001. Combat troops with glasses are now considered obsolete.
What all these systems do is quite similar. The basic idea is this:
This all originated as an internal mechanism within Visual Basic and was responsible for its success. Alan Cooper had finally implemented a reasonably idiot-proof way to make A talk to B. That's why Visual Basic controls were so powerful. Then Microsoft made it a general-purpose mechanism. The Unix/Linux world didn't pick up on this for a long time. The Gnome and OpenOffice people get it, and their stuff plays well together. Most other extensibility is a master-slave relationship, usually involving "plug-ins", "toolbars", or "scripting".
Part of the problem is that the main languages in the Unix/Linux world are C and C++, which have no introspection. If you have a structure, and you need to go through all its fields and get their name and type info so you can format and marshall them for transmission, there's no way to do that from within C or C++, other than by writing glue code by hand for each interface. Preprocessors have been built to deal with this problem (ones exist for OpenRPC and SOAP), but they're clunky and complicate the build process. CORBA requires that you define each interface in its own special language, IDL. The Java world has a solution to this problem, but it's Java-only. Microsoft has put extensions into their various dialects of C and C++ to deal with this, but they were too Microsoft-specific to go into the standard. So making A talk to B without extra work remains hard in the Unix/Linux world.
The intercommunication system within OpenOffice, the mechanism that allows embedding spreadsheets and drawings in other documents, is CORBA-based. Sort of. Actually, it's something called UNO, which started life as CORBA but went off in an XML direction.
GNOME also uses CORBA internally. But its CORBA isn't compatible with the one from OpenOffice.
The UNIX/Linux world has never really had a good way for applications on the same machine to intercommunicate in a subroutine-like way. Microsoft has OLE/COM/DCOM/ActiveX, which is clunky but always available. In the Linux/Unix world, there's nothing you can assume is always there. There's OpenRPC, there's CORBA (in about five incompatible forms), there's Java RMI, and there are various kludges built out of pipes. But there's been no convergence in two decades.
That's Craigslist, making it impossible to compete because of their low, low prices. And it's working. The newspaper industry is furious at being underpriced. Tough.
There may be a big problem looming for Hollywood. If you transfer 35MM film to 1080p, the film grain often shows up. Compressing all that useless film grain noise, which has no frame to frame coherence, will use up a big fraction of the data capacity. It will also mess up the motion compression, which usually results in annoying jaggies. So it's probably necessary to filter out at least some of the film grain. But if you filter out the film grain, you lose resolution.
The reviews of the new Blu-ray disks ("the picture looks too soft and flat") indicates that there's probably too much filtering.
Somewhere in LA, there are probably members of SMTPE struggling with this, trying to figure out the right tradeoffs between resolution loss and compression overload when converting existing films.