SpamBayes is a free open-source plug-in for Outlook. It's tiny. No reboot. It gets things wrong a little for the first week or so as you start to train it, but it very rapidly gets much better. Just remember to occasionally check your Junk Mail and Junk Suspects folders for any proper mails that slipped through, and that you do have to cut it some slack while you're training it. After a few weeks, it's extremely accurate. Coupled with AVG Anti-Virus, and your Outlook experience gets rather better.
There has to be a statistical reason why your car insurance is so absurdly high when you're a late teen, with a steady decrease before a significant reduction at the age of 35. Certainly your appetite for risk behind the wheel doesn't completely reflect your all round maturity in life, but I'd suggest a strong correlation.
I'm all intrigued. Tell us about your morning coffee ritual. I love Slashdot threads that involve coffee because I'm always in awe of the real "out there" coffee drinkers.
Seriously. I'd love to hear what you do. Part voyeurism, part looking for tips to improve my own humble brew.
Why do I need a Tivo or a Media Centre PC in order to play this content if I already have a PC and broadband?
Aegilops
Best Racing Simulator would be Grand Prix Legends
on
Genre-Defining Games?
·
· Score: 1
If you want arcade fun, then GT4 certainly sounds the business. But for serious racing, even 7 years after its release there's still nothing like it for accuracy and immersion - although LFS and our own favourite open source project Motorsport are getting close.
Don't make the mistake of thinking this is a nostalgic sympathy vote for a game that would look dismal to fresh eyes. See for yourself. If you're seriously into driving, you owe it to yourself to spend the $10 or so to check it out. If you're up for the challenge, start with this.
Anandtech do some pretty reasonable guides. They assume you've been following industry trends but I'm in the same boat as you - I went from a Celeron 400 to a P4 2.6 overclock - that took quite a bit of homework to research.
They have four main guides that they update every month or so: cheap, medium ( = reasonably high end), gamer-money-no-object (admittedly rather old now, from Nov 04), and overclock-city (very old - Sep 04). The theory is that they periodically review one of their guides and make sure it's up to date. Clearly some guides get reviewed more often than others.
I also used to go to Dans Data but I think he's too busy reviewing toys, moving house and having a girlfriend nowadays.
I was there, hugely underqualified amongst a room packed with full-on Java devs. There were about 600 attendees. Where I get this completely wrong, please be gentle and help out, don't skin me alive, please.
My recollections:
Eclipse and other Java IDEs
I recall James commenting that the presence of other Java IDEs was healthy as it promotes competition and encourages the best dev tools. He used a sports analogy about needing more than one team to have a match. Two of the day's briefings were about the new Sun IDEs, hence the reporter's focus on this. I recall he had tried Eclipse and was being gracious but non-committal about it.
DVD Technology
James did say that it was a very sloppy bit of crypto, and wasn't intended as a rock solid algorithm but being sufficient to force companies to license the appropriate patents and other agreements.
.NET Security
Again, underqualified here, but he commented that the.NET support of C and C++ was... oh, I wish I could remember the colourful phrase... something like: "the most brain-dead design decision they could have made". I think it was to do with pointer manipulation and arithmetic, which Microsoft allows in the CLR and causes major problems, rather than in Java which keeps you away from all that.
Embedded Software
He was asked what still gets him really excited. He mentioned about the new tiny chips they're building with a micro OS on them, with integrated sensors - I think he called the micro-OS "squawk". He made reference to people using them to plaster all over the sides of bridges to monitor stress patterns. He said that right now they're about an inch square but hope to get them down to the size of your thumbnail. He said that it was a tragedy that you had to have a battery to power these things, as the chip is dwarfed by the battery itself. He mentioned the ubiquity of Java in mobile phones a few times. A few people ribbed him about toasters. One guy pressed him on nanotechnology - driving at an NBIC point of view - James didn't really take the bait which is a shame as I thought that was an interesting question.
He mentioned on a couple of occasions about doing some work with real time applications of Java. And that this is an interesting area as people wouldn't normally think of Java as a good language for real-time, but he said that when you have conversations with these guys, that what they are doing in real-time applications is really scary and "out there", i.e. the current real-time software approaches are really complex, convoluted and much more fragile than you would either think or would like.
James also talked about support for OS X (full production support for Java dev tools, just maybe not the daily builds), and the difficulties of working with Apple (no secret previews or insider knowledge).
There was some talk about scripting languages for Java, with reference to (I'm struggling here) Jython or Ruby. He said people get really religious about it, and he likes the fact that there is not "a single one choice" - the variety is healthy and good.
He spoke encouragingly over the Java Community Process - people ask him "where is Java going" and he observed that it was healthy that it is not down to one single individual. He talked about the Mustang development process and encouraged people to contribute feedback / ideas or participate in the development process.
There were some questions about greatest regrets over the language but sadly I am too underqualified to answer. Whatever he said was really insightful (something about operator casting? I can't recall... I remember him alluding to people accosting him over the left shift operator not being an I/O operator.. but I could have my wires crossed here) - he said "If I had taken out feature XXX of the language, 5% of the audience would say - Yes! that's exactly right, 40% would shrug, and the remainder would come after me with knives drawn".
He was open, genial, down to Earth, and was a pleasure to go hear speak. I'd recommend it if you get the chance to see something similar near you.
Firstly, I have to agree totally with it being a poor show about a lack of Linux or UNIX clients. My lowly OpenBSD box could be chugging away 24x7: other posters have made similar references to non-Windows kit they could use that is actually of significant value (unlike mine). No argument there.
I just downloaded the agent myself on a WinXP box, and the default install is *NOT* to run as a screensaver. The NT kernel (and 2K, XP etc) supports thread prioritisation, so this process is chugging away right now with the default priority set to Low. You can see this from Task Manager by selecting View, Select Columns, Base Priority. While composing this post I have chugged through 4% of a batch, for what it's worth.
Incidentally, the client is a re-skin and tarting up of the old UD client, as used a few years back for the University of Oxford cancer research project. What's interesting is that back then, UD was getting paid for building the massively distributed client. That in itself is not inherently bad as such, but was worth knowing at the time. I'd expect the same this time 'round, too.
I've been running both for a while now. Some observations:
Google Desktop supports Unicode which is invaluable if you use non-ASCII languages, like my wife does. YMMV. However the Google Desktop search is not integrated into the Outlook shell (understandably) nor the Google Deskbar, which I think is an obvious oversight - and suggested as much to Google.
Lookout allows you to index mapped drive letters or network locations, which Google Desktop doesn't. This is great for me where I have documents on a laptop's local hard drive as well as on network shares. I can't quantify it, but I think it has slowed down my Outlook 2003 a little, particularly on start-up. Most hits are returned in less than a tenth of a second. My major gripe about Lookout is that when I move items from my inbox to my PST it stuffs up the index - I know it rebuilds the full index once a month, but more often than not I look for something that has been indexed as being in my Inbox, yet I have since moved it to this month's PST folder. Nonetheless, it gives me a clue what to look for in my PST. I predominantly use Lookout for Outlook at work, and can't really comment on how this compares to using Google Desktop search on a "busy" Outlook mailbox.
Both systems use the CPU power of your workstation to build the indexes when idle. I think this is poor. E.g. you disconnect from the network and go roaming. When you return to the office, you want to find all documents and mails containing 'squeamish ossifrage'. Why should you have to wait for your PC to do the indexing? And does your indexing process "touch" each file? If it does, it could seriously screw up any attempts to archive all old data - everything would look current as every file was being touched by all PCs' indexing programs.
Surely it's feasible to get a master indexing catalogue built from a number of indexing sources. What I would like to see is an indexer for Exchange that indexes each individual mailbox but returns user-specific queries. So when I dock back at the office, I can immediately search for new documents that have been delivered to the Exchange server while I've been disconnected, and indexed on my behalf. Of course, what hits I get returned are unique to me as only my mailbox index is visible to me - as your mailbox index is unique to you. Meanwhile, common areas, such as shared file servers / public folders / web content etc can have their index shared across both of us.
Nonetheless, do not underestimate the joy of being able to use either of these tools and have an instantaneous method for locating a buried document that you know is somewhere on your PC, yet cannot remember precisely where.
I work on the principle that a user, if motivated enough, will get a copy of the file off site. USB key, CD-R, email, laptop, take out hard drive, print out a hex dump of file then fax and OCR it, carrier pigeon... they'll find a way.
With IRM, my understanding (untested at this time) is that you can flag certain document types so that they have to be verified against the rights management server, which would be a box on your internal network.
So - good for you - you've snuck the file off-site. So when you go to open it up, the file (only openable using Office 2003) recognises that it has to be verified against the central rights server - and if that can't be contacted, then you don't get to open it up. This means that the data on your CD-Rs / USB keys / whatever are no use when off-site. It's the document itself that becomes protected, rather than stopping users from taking advantage of any number of methods to get a file off-site.
Of course, perhaps the technology doesn't work that way in practice, but it's on my to-do list of technologies to evaluate for precisely this purpose.
Just to expand upon some of your examples a little:
Software package distribution to end-users (a la SMS or Group Policy)
Desktop lockdown policies, e.g. very restricted access for, say, a call centre, "normal" access for the general users, maybe a more elevated level of access for the odd rogue punk
Desktop roaming and profiles, i.e. a user should be able to log on to any desktop and receive all of his/her applications and data
Expanding the above point - if a PC fails, it should be trivial to either re-image or swap out the hardware and have the user back up and running almost immediately. I.e. no local data / no local installs
Strong method of validating integrity of the desktop, particularly in regulated industries (banking, pharmaceuticals etc) - i.e. how can you "prove" that the machine has not been tampered with, and so is operating correctly. Sounds daft? Try working in a regulated industry...
Hardware inventory / monitoring toolkits (in an ideal world, you'd have a single machine image for the whole company to make support of your desktop image easier, but life frequently isn't that simple)
Software inventory / monitoring toolkits (not all software will be freely licensed, you may be distributing some proprietary software that runs on your free systems
Remote control software to enable support staff to assist users remotely
Your examples of automatically distributing patches (and forcing, and preventing logon from un-patched machines) for both OS and applications is exactly right, along with having the control to test and select what patches are distributed to end users. No doubt many of my examples above are already addressed, and this is after all what you'd be paying a Linux expert to help you with (read: commercial support organisation and consultancy - IBM would likely be a good fit, along with many others). Remember, a corporation could well take the view that if it's worth doing, it's worth doing right - i.e. choose Linux for the right reasons, but don't automatically assume that they will want to do it for zero cost - both in the initial purchase price as well as the ongoing maintenance.
The greatest gotcha is almost certainly application remediation and migration, not the desktop end (i.e. end-user trailing, rebuilding PCs).
They will likely have a few thousand applications to worry about, although that amount of diversity could easily be reduced as part of the migration to Linux. E.g. they may have 20 different graphical FTP clients - when they move to Linux, they only need one.
Fair point, but by the same token, your argument is equally valid for the OS X cluster at Virginia Tech. Not having eyeballed it personally, and basing my information purely off Steve Jobs keynote speech at MacWorld 2004, they made a point of saying "hey, this is great, we have the same GUI that runs on the cluster as the one that runs on my personal workstation".
It's quite likely that your suggestion of a GUI-less edition of Windows is feasible, and you'd hope that with the BSD core of OS X that it's even easier for Apple to do the same. You'd have to address the question to what extent is it possible to perform all functions through a non-GUI interface.
Sorry for the "me too" but I would totally endorse this recommendation. We were advised to get one from our sister company. Although I was a bit skeptical at first, it soon became apparent that it was a tremendous time saver, particularly compared to the laborious manual alternative. We got the 50 page sheet feeder (would consider that a 'must') model and it was great. Same size as a small fax machine, dead simple to use. Integrated with our Exchange address book too.
We never bothered pushing the model to explore further functionality (e.g. I proposed we looked at programming it to scan documents to save output TIFFs into a central folder, which we could then use best-of-breed OCR software to convert to text) but the potential was clear.
Has anyone compared GT to Grand Prix Legends for driving realism / physics? I've never played / seen GT so have no way of comparing, but it seems pretty much unanimous wisdom that GPL is the pinnacle of driving physics, despite being launched over 5 years ago.
You could do worse than to have a look at Retriever from 80-20 software. One of their top sales guys demoed the product to me about 9 months ago - it does integrated searching and categorisation (without having to pre-load metadata tags), similar to the search engine used by Telstra in Oz. Try searching for 'broadband' - it categorises the search results automatically (look at the left hand margin) without requiring document authors filling in masses of mandatory property sheets.
Having had a quick scan of their web page ("Check those URLs!"), it appears they are repositioning Retriever away from the individual use towards an enterprise-wide solution (e.g. point it at your file servers and Exchange box), but there might be some mileage in a single-user copy. I recall there was an evaluation version at one point, however all that's there now is a video you could look at.
When reading the article I also wondered what would prevent someone from taping over the camera lens. While it would nevertheless make for some awkward questioning if the user were challenged as to what they were actually shooting at (particularly if the GPS system indicated you were a looong way away from any shooting range), I figured they could work on a similar method as used in SLR cameras.
With usual disclaimer of not being a camera buff (IANACB), an SLR is better than your standard compact camera in that what you see through the viewfinder is exactly the same as what is eventually recorded on film. OK, so with a camera they use mirrors to redirect the line of sight which clearly wouldn't work with bullets!
However they could have a small lens or mirror pop into the barrel (is this feasible?) and retract a fraction of a second before the bullet is fired. After all, the gun is designed to only fire a maximum of 3 times per second, I'm sure they could coordinate it as part of the internals.
If they did that then the only way of circumventing it would be to put a bit of tape over the muzzle exit. I'm sure that wouldn't go down too well with the gun's owner.
Finally, your point about being at a shooting range - if they programmed in the locations of licensed shooting ranges into the GPS system then it could presumably be configured not to record at those locations. Which neatly suggests that THAT is the place you should perform all of your gangland killings.
Very interesting point you make regarding Genetic Programming paradigm. I'm guessing (not recognising the 'Koza') that this is a genetic algorithm approach. If my assumption is correct, then in theory you could map a string of attributes to the character (e.g. propensity to attack, strength, range of eyesight, peripheral vision, preparedness to sustain damage before running away etc) and let them loose. The combinations of attributes that don't work out (e.g. poor eyesight, poor strength, never run away) will get wiped out soon enough, whereas other combinations (obvious) survive for the next fight.
What your point refers to is that the agents' parameters may change throughout a battle (e.g. getting 'dirtier'). This interested me. Traditionally with a GA you would use crossover and mutation operators to mix up the characteristics. Without descending into juvenile humour, clearly crossover would not be implemented through a mating concept, but rather a 'watch and learn' approach, where segments of a characters attributes are transferred to another.
E.g. Orc 1 watches Orc 2, a bloodthirsty maniac with a strong dirty streak coupled with self-preservation, and decides to copy in those characteristics into his own (Orc 1's) characteristic set. I'm sure the odd randomization / mutation operator could be simply implemented to adjust a random characteristic to freshen up the gene pool.
This works fine for behavioural characteristics (e.g. "I'm being a chicken! Those brave guys are doing much better than me!") where a character can 'choose' to adjust an attribute. The concept doesn't seem so plausible when it concerns physical attributes - e.g. "hey I'm not that strong! I'll just pump up these muscles a bit and do some serious damage". You'd be unimpressed if the two armies entered a race condition based on height, with warriors sprouting up like trees (obvious Ent joke in the making here).
In terms of the GA, what might be missing in all this is the ability to resurrect excellent segments of the overall parameter set and reintroduce them into the population. After all, the watch and learn approach should in theory only be limited to other characters in the immediate vicinity of the character (unless you throw the realistic approach totally away, and are prepared to copy string segmentss arbitrarily around and see what happens). What happens to someone with a brilliant tradeoff between aggression and self-preservation, but was killed through having chronic shortsightedness?
In any event it sounds like it would be an interesting exercise if the parameter sets could be configured or tuned by an armies 'leader' (e.g. Saruman, Sauron...) and let loose in battle to see how they fared against another army. Similar to other "hands off" simulations where you primed your agent / army / character with characteristics and then let them loose to fare for themselves.
In conclusion, sadly I suspect that if MASSIVE is commercially released it will be comprehensively out of the reach of mere mortals such as us.
Slashdot Mirror
SpamBayes is a free open-source plug-in for Outlook. It's tiny. No reboot. It gets things wrong a little for the first week or so as you start to train it, but it very rapidly gets much better. Just remember to occasionally check your Junk Mail and Junk Suspects folders for any proper mails that slipped through, and that you do have to cut it some slack while you're training it. After a few weeks, it's extremely accurate. Coupled with AVG Anti-Virus, and your Outlook experience gets rather better.
Recommended.
Aegilops
There has to be a statistical reason why your car insurance is so absurdly high when you're a late teen, with a steady decrease before a significant reduction at the age of 35. Certainly your appetite for risk behind the wheel doesn't completely reflect your all round maturity in life, but I'd suggest a strong correlation.
Aegilops
I'm all intrigued. Tell us about your morning coffee ritual. I love Slashdot threads that involve coffee because I'm always in awe of the real "out there" coffee drinkers.
Seriously. I'd love to hear what you do. Part voyeurism, part looking for tips to improve my own humble brew.
Aegilops
Holy Moley. All this time and I was aiming at that damn cross...
Aegilops
Q: What do you get if you cross a sheep with a goat?
A: mod sheep mod goat n hat sine theta
Q: What do you get if you cross a sheep with a mountain climber?
A: Nothing, because you can't cross a vector with a scalar.
Aegilops
Why do I need a Tivo or a Media Centre PC in order to play this content if I already have a PC and broadband?
Aegilops
If you want arcade fun, then GT4 certainly sounds the business. But for serious racing, even 7 years after its release there's still nothing like it for accuracy and immersion - although LFS and our own favourite open source project Motorsport are getting close.
Don't make the mistake of thinking this is a nostalgic sympathy vote for a game that would look dismal to fresh eyes. See for yourself. If you're seriously into driving, you owe it to yourself to spend the $10 or so to check it out. If you're up for the challenge, start with this.
Aegilops
Anandtech do some pretty reasonable guides. They assume you've been following industry trends but I'm in the same boat as you - I went from a Celeron 400 to a P4 2.6 overclock - that took quite a bit of homework to research.
They have four main guides that they update every month or so: cheap, medium ( = reasonably high end), gamer-money-no-object (admittedly rather old now, from Nov 04), and overclock-city (very old - Sep 04). The theory is that they periodically review one of their guides and make sure it's up to date. Clearly some guides get reviewed more often than others.
I also used to go to Dans Data but I think he's too busy reviewing toys, moving house and having a girlfriend nowadays.
Hope this helps.
Aegilops
My recollections:
Eclipse and other Java IDEs
I recall James commenting that the presence of other Java IDEs was healthy as it promotes competition and encourages the best dev tools. He used a sports analogy about needing more than one team to have a match. Two of the day's briefings were about the new Sun IDEs, hence the reporter's focus on this. I recall he had tried Eclipse and was being gracious but non-committal about it.
DVD Technology
James did say that it was a very sloppy bit of crypto, and wasn't intended as a rock solid algorithm but being sufficient to force companies to license the appropriate patents and other agreements.
Again, underqualified here, but he commented that the .NET support of C and C++ was ... oh, I wish I could remember the colourful phrase ... something like: "the most brain-dead design decision they could have made". I think it was to do with pointer manipulation and arithmetic, which Microsoft allows in the CLR and causes major problems, rather than in Java which keeps you away from all that.
Embedded Software
He was asked what still gets him really excited. He mentioned about the new tiny chips they're building with a micro OS on them, with integrated sensors - I think he called the micro-OS "squawk". He made reference to people using them to plaster all over the sides of bridges to monitor stress patterns. He said that right now they're about an inch square but hope to get them down to the size of your thumbnail. He said that it was a tragedy that you had to have a battery to power these things, as the chip is dwarfed by the battery itself. He mentioned the ubiquity of Java in mobile phones a few times. A few people ribbed him about toasters. One guy pressed him on nanotechnology - driving at an NBIC point of view - James didn't really take the bait which is a shame as I thought that was an interesting question.
He mentioned on a couple of occasions about doing some work with real time applications of Java. And that this is an interesting area as people wouldn't normally think of Java as a good language for real-time, but he said that when you have conversations with these guys, that what they are doing in real-time applications is really scary and "out there", i.e. the current real-time software approaches are really complex, convoluted and much more fragile than you would either think or would like.
James also talked about support for OS X (full production support for Java dev tools, just maybe not the daily builds), and the difficulties of working with Apple (no secret previews or insider knowledge).
There was some talk about scripting languages for Java, with reference to (I'm struggling here) Jython or Ruby. He said people get really religious about it, and he likes the fact that there is not "a single one choice" - the variety is healthy and good.
He spoke encouragingly over the Java Community Process - people ask him "where is Java going" and he observed that it was healthy that it is not down to one single individual. He talked about the Mustang development process and encouraged people to contribute feedback / ideas or participate in the development process.
There were some questions about greatest regrets over the language but sadly I am too underqualified to answer. Whatever he said was really insightful (something about operator casting? I can't recall ... I remember him alluding to people accosting him over the left shift operator not being an I/O operator.. but I could have my wires crossed here) - he said "If I had taken out feature XXX of the language, 5% of the audience would say - Yes! that's exactly right, 40% would shrug, and the remainder would come after me with knives drawn".
He was open, genial, down to Earth, and was a pleasure to go hear speak. I'd recommend it if you get the chance to see something similar near you.
Aegilops
The library of the University of Oxford, i.e. the Bodleian Library, was the first "copyright" library in the UK - one of only three - which means that it automatically gets a copy of every book published in the UK.
Aegilops
Firstly, I have to agree totally with it being a poor show about a lack of Linux or UNIX clients. My lowly OpenBSD box could be chugging away 24x7: other posters have made similar references to non-Windows kit they could use that is actually of significant value (unlike mine). No argument there.
I just downloaded the agent myself on a WinXP box, and the default install is *NOT* to run as a screensaver. The NT kernel (and 2K, XP etc) supports thread prioritisation, so this process is chugging away right now with the default priority set to Low. You can see this from Task Manager by selecting View, Select Columns, Base Priority. While composing this post I have chugged through 4% of a batch, for what it's worth.
Incidentally, the client is a re-skin and tarting up of the old UD client, as used a few years back for the University of Oxford cancer research project. What's interesting is that back then, UD was getting paid for building the massively distributed client. That in itself is not inherently bad as such, but was worth knowing at the time. I'd expect the same this time 'round, too.
Cheers
Aegilops
I've been running both for a while now. Some observations:
Google Desktop supports Unicode which is invaluable if you use non-ASCII languages, like my wife does. YMMV. However the Google Desktop search is not integrated into the Outlook shell (understandably) nor the Google Deskbar, which I think is an obvious oversight - and suggested as much to Google.
Lookout allows you to index mapped drive letters or network locations, which Google Desktop doesn't. This is great for me where I have documents on a laptop's local hard drive as well as on network shares. I can't quantify it, but I think it has slowed down my Outlook 2003 a little, particularly on start-up. Most hits are returned in less than a tenth of a second. My major gripe about Lookout is that when I move items from my inbox to my PST it stuffs up the index - I know it rebuilds the full index once a month, but more often than not I look for something that has been indexed as being in my Inbox, yet I have since moved it to this month's PST folder. Nonetheless, it gives me a clue what to look for in my PST. I predominantly use Lookout for Outlook at work, and can't really comment on how this compares to using Google Desktop search on a "busy" Outlook mailbox.
Both systems use the CPU power of your workstation to build the indexes when idle. I think this is poor. E.g. you disconnect from the network and go roaming. When you return to the office, you want to find all documents and mails containing 'squeamish ossifrage'. Why should you have to wait for your PC to do the indexing? And does your indexing process "touch" each file? If it does, it could seriously screw up any attempts to archive all old data - everything would look current as every file was being touched by all PCs' indexing programs.
Surely it's feasible to get a master indexing catalogue built from a number of indexing sources. What I would like to see is an indexer for Exchange that indexes each individual mailbox but returns user-specific queries. So when I dock back at the office, I can immediately search for new documents that have been delivered to the Exchange server while I've been disconnected, and indexed on my behalf. Of course, what hits I get returned are unique to me as only my mailbox index is visible to me - as your mailbox index is unique to you. Meanwhile, common areas, such as shared file servers / public folders / web content etc can have their index shared across both of us.
Nonetheless, do not underestimate the joy of being able to use either of these tools and have an instantaneous method for locating a buried document that you know is somewhere on your PC, yet cannot remember precisely where.
Aegilops
Curious. When I'm thinking of the name Spiro Agnew my mind often turns to thoughts that it is an anagram of grow a penis.
It's a very new technology from the company everyone loves to hate, but the new Information Rights Management (IRM) capability in Office 2003 looks like a potential solution to this problem.
I work on the principle that a user, if motivated enough, will get a copy of the file off site. USB key, CD-R, email, laptop, take out hard drive, print out a hex dump of file then fax and OCR it, carrier pigeon
With IRM, my understanding (untested at this time) is that you can flag certain document types so that they have to be verified against the rights management server, which would be a box on your internal network.
So - good for you - you've snuck the file off-site. So when you go to open it up, the file (only openable using Office 2003) recognises that it has to be verified against the central rights server - and if that can't be contacted, then you don't get to open it up. This means that the data on your CD-Rs / USB keys / whatever are no use when off-site. It's the document itself that becomes protected, rather than stopping users from taking advantage of any number of methods to get a file off-site.
Of course, perhaps the technology doesn't work that way in practice, but it's on my to-do list of technologies to evaluate for precisely this purpose.
Aegilops
Just to expand upon some of your examples a little:
Software package distribution to end-users (a la SMS or Group Policy)
Desktop lockdown policies, e.g. very restricted access for, say, a call centre, "normal" access for the general users, maybe a more elevated level of access for the odd rogue punk
Desktop roaming and profiles, i.e. a user should be able to log on to any desktop and receive all of his/her applications and data
Expanding the above point - if a PC fails, it should be trivial to either re-image or swap out the hardware and have the user back up and running almost immediately. I.e. no local data / no local installs
Strong method of validating integrity of the desktop, particularly in regulated industries (banking, pharmaceuticals etc) - i.e. how can you "prove" that the machine has not been tampered with, and so is operating correctly. Sounds daft? Try working in a regulated industry...
Hardware inventory / monitoring toolkits (in an ideal world, you'd have a single machine image for the whole company to make support of your desktop image easier, but life frequently isn't that simple)
Software inventory / monitoring toolkits (not all software will be freely licensed, you may be distributing some proprietary software that runs on your free systems
Remote control software to enable support staff to assist users remotely
Your examples of automatically distributing patches (and forcing, and preventing logon from un-patched machines) for both OS and applications is exactly right, along with having the control to test and select what patches are distributed to end users. No doubt many of my examples above are already addressed, and this is after all what you'd be paying a Linux expert to help you with (read: commercial support organisation and consultancy - IBM would likely be a good fit, along with many others). Remember, a corporation could well take the view that if it's worth doing, it's worth doing right - i.e. choose Linux for the right reasons, but don't automatically assume that they will want to do it for zero cost - both in the initial purchase price as well as the ongoing maintenance.
Aegilops
The greatest gotcha is almost certainly application remediation and migration, not the desktop end (i.e. end-user trailing, rebuilding PCs).
They will likely have a few thousand applications to worry about, although that amount of diversity could easily be reduced as part of the migration to Linux. E.g. they may have 20 different graphical FTP clients - when they move to Linux, they only need one.
Aegilops
Fair point, but by the same token, your argument is equally valid for the OS X cluster at Virginia Tech. Not having eyeballed it personally, and basing my information purely off Steve Jobs keynote speech at MacWorld 2004, they made a point of saying "hey, this is great, we have the same GUI that runs on the cluster as the one that runs on my personal workstation".
It's quite likely that your suggestion of a GUI-less edition of Windows is feasible, and you'd hope that with the BSD core of OS X that it's even easier for Apple to do the same. You'd have to address the question to what extent is it possible to perform all functions through a non-GUI interface.
Sorry for the "me too" but I would totally endorse this recommendation. We were advised to get one from our sister company. Although I was a bit skeptical at first, it soon became apparent that it was a tremendous time saver, particularly compared to the laborious manual alternative. We got the 50 page sheet feeder (would consider that a 'must') model and it was great. Same size as a small fax machine, dead simple to use. Integrated with our Exchange address book too.
We never bothered pushing the model to explore further functionality (e.g. I proposed we looked at programming it to scan documents to save output TIFFs into a central folder, which we could then use best-of-breed OCR software to convert to text) but the potential was clear.
Aegilops
Has anyone compared GT to Grand Prix Legends for driving realism / physics? I've never played / seen GT so have no way of comparing, but it seems pretty much unanimous wisdom that GPL is the pinnacle of driving physics, despite being launched over 5 years ago.
Or finding out that an in-house developed and compiled application sent passwords in clear text.
From the Slashdot grammar police: credit where it's due. Your well crafted, balanced and insightful sentence was much appreciated.
Now, back to work educating posters about "it's" and "it was done wrongly" etc.
Aegilops
You could do worse than to have a look at Retriever from 80-20 software. One of their top sales guys demoed the product to me about 9 months ago - it does integrated searching and categorisation (without having to pre-load metadata tags), similar to the search engine used by Telstra in Oz. Try searching for 'broadband' - it categorises the search results automatically (look at the left hand margin) without requiring document authors filling in masses of mandatory property sheets.
Having had a quick scan of their web page ("Check those URLs!"), it appears they are repositioning Retriever away from the individual use towards an enterprise-wide solution (e.g. point it at your file servers and Exchange box), but there might be some mileage in a single-user copy. I recall there was an evaluation version at one point, however all that's there now is a video you could look at.
Disclaimer: I have no affiliation with 80-20.
Aegilops
28 Apr '01 - 64 bytes from 10.0.3.1: icmp_seq=0 ttl=255 time=6165731.1 ms
Aegilops
With usual disclaimer of not being a camera buff (IANACB), an SLR is better than your standard compact camera in that what you see through the viewfinder is exactly the same as what is eventually recorded on film. OK, so with a camera they use mirrors to redirect the line of sight which clearly wouldn't work with bullets!
However they could have a small lens or mirror pop into the barrel (is this feasible?) and retract a fraction of a second before the bullet is fired. After all, the gun is designed to only fire a maximum of 3 times per second, I'm sure they could coordinate it as part of the internals.
If they did that then the only way of circumventing it would be to put a bit of tape over the muzzle exit. I'm sure that wouldn't go down too well with the gun's owner.
Finally, your point about being at a shooting range - if they programmed in the locations of licensed shooting ranges into the GPS system then it could presumably be configured not to record at those locations. Which neatly suggests that THAT is the place you should perform all of your gangland killings.
Aegilops
Very interesting point you make regarding Genetic Programming paradigm. I'm guessing (not recognising the 'Koza') that this is a genetic algorithm approach. If my assumption is correct, then in theory you could map a string of attributes to the character (e.g. propensity to attack, strength, range of eyesight, peripheral vision, preparedness to sustain damage before running away etc) and let them loose. The combinations of attributes that don't work out (e.g. poor eyesight, poor strength, never run away) will get wiped out soon enough, whereas other combinations (obvious) survive for the next fight.
What your point refers to is that the agents' parameters may change throughout a battle (e.g. getting 'dirtier'). This interested me. Traditionally with a GA you would use crossover and mutation operators to mix up the characteristics. Without descending into juvenile humour, clearly crossover would not be implemented through a mating concept, but rather a 'watch and learn' approach, where segments of a characters attributes are transferred to another.
E.g. Orc 1 watches Orc 2, a bloodthirsty maniac with a strong dirty streak coupled with self-preservation, and decides to copy in those characteristics into his own (Orc 1's) characteristic set. I'm sure the odd randomization / mutation operator could be simply implemented to adjust a random characteristic to freshen up the gene pool.
This works fine for behavioural characteristics (e.g. "I'm being a chicken! Those brave guys are doing much better than me!") where a character can 'choose' to adjust an attribute. The concept doesn't seem so plausible when it concerns physical attributes - e.g. "hey I'm not that strong! I'll just pump up these muscles a bit and do some serious damage". You'd be unimpressed if the two armies entered a race condition based on height, with warriors sprouting up like trees (obvious Ent joke in the making here).
In terms of the GA, what might be missing in all this is the ability to resurrect excellent segments of the overall parameter set and reintroduce them into the population. After all, the watch and learn approach should in theory only be limited to other characters in the immediate vicinity of the character (unless you throw the realistic approach totally away, and are prepared to copy string segmentss arbitrarily around and see what happens). What happens to someone with a brilliant tradeoff between aggression and self-preservation, but was killed through having chronic shortsightedness?
In any event it sounds like it would be an interesting exercise if the parameter sets could be configured or tuned by an armies 'leader' (e.g. Saruman, Sauron...) and let loose in battle to see how they fared against another army. Similar to other "hands off" simulations where you primed your agent / army / character with characteristics and then let them loose to fare for themselves.
In conclusion, sadly I suspect that if MASSIVE is commercially released it will be comprehensively out of the reach of mere mortals such as us.
Aegilops