The spec has to have a rock-solid security model required for implementors, and a good security test suite must be freely available. Without these, the database will turn out to be a major hack vector. With a great security model, we only have to worry about bugs. As it stands, the spec covers security very lightly.
The spec has these sections that mean people are at least thinking about security. I hope there are actual security experts involved:
Suggestions, if you care about the code you will write in your career:
1) Figure out whether you're learning deep CS fundamentals in your undergrad program. A good Masters program will teach you CS fundamentals that make you a better software engineer in a real job. Some undergrad programs don't do that, and knowing the difference should help you decide.
2) Research (really research) the curricula at some graduate programs. They differ. You can probably tell if any of them sound like you'd learn a lot.
3) If you target the Masters degree, GET IT FAST. Make it a goal to finish the degree and start using it. Do as much programming as possible during the degree program, so you're ready to code for real. Use it as a powerup.
I'm in the gene group that gets a headache when I go 24+ hours without caffeine.
But I can painlessly quit whenever I feel like it. I use a modified binary exponential backoff algorithm. For example, if I'm drinking 1 cup a day:
Take it down to.75 dose: Drink less than a full cup for 1-2 days.
Then drink a half cup for 2-3 days.
Then drink half that, == a soda or half a tea for 2-3 days.
Then don't drink any.
Backing the dosage off slowly completely avoids headaches. For me. YMMV, but give it a shot. If you usually drink more, you might want to take a few more days (2 days per binary step).
"Such extortionate practices should not be allowed."
Such extortionate practices should be punished. Disallowing future misdeeds isn't good enough. The RIAA should be (legally) punished for bad behavior. Otherwise there's no real incentive for organizations like that to stop.
I would not say this:
"2) It's hard to do stealthy reentry."
We don't know how hard it is, or whether it's already easy enough. That's the beauty of stealthiness. For example, if a military were doing stealthy re-entries all the time, you might not know.
Before the F-117 and B-2, stealth aircraft were considered too hard to do. But there they are.
Survival of the human species. This is a social & political application, not scientific.
When you read the newspaper for the past 100 years, there's evidence the species might destroy itself. Often, wars happen because people don't understand their place in the universe. For example, a country can be run by people who believe a deity will save the faithful, so world turmoil and war is ok. Other times, wars are a symptom of tyranny (somebody wants "power").
Humans are very visually oriented. Pretty pictures pull the eye in.
As more people gaze at pretty pictures of the gi-normous universe full of stars, more people realize we live at a single star, and there are many many other stars similar to ours. That may reduce destructive effects of belief the deity who will "save" you wants you to kill people.
Gazing at pretty pictures can also inspire people who would otherwise permit tyranny (humans are lazy). A strong enough desire to find out about the pretty pictures can make people impatient to know truth. (e.g. "Why are there more stars in the Antennae galaxies than people on earth?", and "Why should these people be governed like this?")
The above possibilities are weak if you want an application you can apply science or technology to. But they're still important. When your species goes away, it reduces your ability to do science.
Is 3 hours long enough for a security check for ballistic style flights? Maybe if you are stripped, x-rayed, evacuated, and enclosed in very tough fabric or metal.
Compared to a measly 500-MPH airplane, a terrorist who takes over or blows up a piloted ballistic missle can attack many new targets with new methods.
I want to fly that flight, and I want to float around and have fun. But until the world returns to a politically stable state, I want passengers in the spaceship locked down and/or checked out for 3 days, not 3 hours.
Based on projects I've done in a similar area of tech, here's what I expect:
SBC's stock price will rise as people believe they're doing this big moneymaker, HDTV, wow.
After the top 10% of SBC management makes at least $20 million each from the stock spike ($100 mil for top 1%), their attention will turn to the next bait. The marketing will continue. Woo hoo.
Meanwhile, SBC will run into a combination of legal or financial trouble: anti-trust, bad accounting, misleading investors, and/or new legislation from Orrin Hatch.
Microsoft will deliver software as good as it always is. This will slow the deployment and provide bad publicity as the paying beta testers whine about outages, lack of programming, and cost.
SBC is a phone company:
end-user cost will start at twice the initial marketing estimates
the billing model will charge extra for every little thing (10 cents per skipped commercial? $1/hour over 3 hours use per day, outside primetime hours excluding sporting events in overtime?)
the project will take 2 years extra due to new legislation or evil competitors.
they have expertise in aggressively mass-marketing a low-margin service and maintaining communication infrastructure, not selling a sparkly cool new combination of technology
Everybody driving the project will act surprised when there's not enough bandwidth to deliver all the HDTV.
Competitive projects will develop standards, and work better, and run on Linux boxes, and provide a better UI designed by a hungry teenager in Banaglore, and will be used by < 1% of the HDTV market since there is no marketing budget.
SBC will pull the plug when the project turns into a cost center that no longer spikes the stock price up.
HDTV over IP will ever-after sound like a bad idea, since people will remember how this project went.
Of course there's a chance the project will succeed if somebody at the top of SBC is a total freak for HDTV and is a good enough salesman to persuade SBC's board.
Here's a link to the table of contents of the Journal of Psychopharmacology, the journal that is publishing the actual scientific study.
I don't see the study in the online version right now (2004-10-03 14:15 PDT) but it might be there "now" depending on when you read this. The Journal appears to post PDFs of published articles, but it's a subscription-based journal so maybe not.
I would bet that most people would only have "withdrawal symptoms" for a day at most. I'm guessing (very unscientifically) that for most people the need for coffee is a routine.
I realize people have to post anecdotes about themselves and make conclusions about billions of humans in our species. After all, I believe Sturgeon's Law, and not everybody knows what science actually is.
But does such idiocy have to get modded up and not get modded back down? Ackkk.
I recently discovered this biological basis for the additive/subtractive gospel, and I was shocked by the ignorance I had been taught.
View any of the plots of cone absorption, and it is obvious that there are cones detecting light that is not a color you can call red, or green, or blue. Your eyes are surely not narrowband-filtering RGB into your brain. Yet our major media - video, film, print - all show us color using technologies based on this very old compression technique (RGB is a 3-band DCT at every viewed pixel).
Not to mention the randomness thrown into the mixture of RGB sensors and emitters (digicams, video cams, film layers -> printers, televisions, LCDs). If they don't use the same wavelengths of R, G, and B throughout the chain, they cannot stimulate cones the way the photons would have that first went through the camera's lens. The reason this is relevant should be obvious: the cone system is biological. Hint: no two people are alike. The curves are surely different for everyone. If you don't trick the cones well, your brain very likely does notice the compression and it mumbles. "The sunset over the Grand Canyon was better than that; I know it was."
Paintings can be made to look more realistic in places than photo, film, or video systems, because over the years, people have found paints that reflect accurate colors. The painted color you see is reflected/filtered/chosen by microscopic features in paint material; e.g. some materials can actually reflect "yellow" photons instead of reflecting greenish and reddish photons that are also in the stream, so that yellow isn't put into your brain by tricking your cones.
Tickling/tricking the cone system so harshly seems really lame now that we can do better - in the sensors, the compression, and display. Yeah, we have production lines dedicated to RGB spectral compression. Does that mean the world cannot change?
In 100 years, people will view antique RGB photos and laugh. "Look how washed out it is! Couldn't they see?" Future imaging systems might not just be RGB + CMY, but they will use more than RGB.
That is why you're all basically right, and Tweakers are right, and SR and AT are right.
People use computers for different things. It's stupid to care about people who say "RAID 0 is [good,bad] for all users" because nobody is all-users. RAID 0 is useless if all you do is surf and check mail. RAID 0 is great for rotating 100 MB images in Photoshop.
I have Photoshop, video editing, audio editing apps, and a development environment. Those apps can use the boost from RAID 0, specifically for data i/o. Other apps, including the OS, might get a boost, but reliability is much more important there, so I wouldn''t use the RAID 0 drives for those.
All the benchmarks I've read are basically useless, since I have to make many educated guesses about whether my mix of apps AND tasks will run like the benchmark. You and I might spend the hours to make those guesses, but if 95% of users cannot use the benchmark, why do benchmarking?
Here's an app that would be more useful than any single benchmark:
Maintain a database of how an application's performance is affected by [RAID 0, RAM size, P3 vs P4 vs HT vs G5, whatever].
As time goes by, the database should characterize an app's performance in higher detail: people care about running a filter or a transition on a video stream of N GB. Each app might have 2-20 tasks to characterize.
Curious User CU either enters a list of applications they have and a system description, OR this app scans their drive & figures out their system.
CU has to say how "important" each given app is. For example, Photoshop filters are really important and everything else is not.
The app computes a weighted sum of how much each factor affects apps & tasks the CU cares about.
The app says, "There is a 63% chance that using [RAID0, or whatever] will improve speed for you."
If they're small enough for me to put on many, many things I own, and if I control the database that maps an ID to a thing, I can walk through my house with an RFID scanner looking for the keys, or the OpenGL book, or my apartment's rental contract, or whatever. Instant pseudo-organization!
There are practical questions:
How do we make it quick and easy to tag lots of our things? I want to go tap, tap, tap with a pencil-sized dispenser and tag 3 things.
How do we make it easy to record the ID & name of a thing we just tagged? An iPod would hold a lot of data... maybe metadata should be audio, or should be input as audio and converted to text.
How can we make it easy to say "I want to find [thing-name]"? Should all data and searches be verbal instead of text? With some speech-to-text, you could tie a lot of your tagged objects to other data: say the name of the song you want and your scanner leads you to the CD.
What's a good way to represent the data of tagged stuff? If names of things are audio, is there a best simple fuzzy representation of words? We should be able to say "contract" and find the apartment contract, not my contact lens.
What privacy is really required? If I put randomly numbered tags on stuff, nobody can scan my house for the UPC of my solid gold Aibo. But I should be the only one who can make its tag respond. And I cannot go resetting the password on every tag in the house and the storage .
Can the tags be made thin and small enough to tag lots of pieces of paper? Hunting for a specific piece of paper in your filing cabinet (the pile on the floor) could become obsolete, or much easier.
These ideas are now published here on Slashdot, may have bee mentioned before; we haven't seen them patented (yet). Are there any RFID ideas you want add to this thread, before some evil SOB patents them?
("Method of doing another obvious thing with RFID", patent # 10,438,109,852)
If all you want is to keep ideas free, you shouldn't need a patent on them. Just publish them in an easy-to-find location as "anti-patents".
Set up an anti-patent database, findable on the web through obvious domains (unpatentable.org, anti-patent.org, etc). Accept all submissions of ideas, completely keyworded and timestamped, just as searchable as the USPTO's patent database. It doesn't matter too much if the same idea is submitted twice; advise people to search and avoid that, if you want to save disk space. When you find out an idea in the database is patented, leave it in the database but add references to the patents. Maintain security of the database and accuracy of timestamps, because without credibility the database is worthless.
A centralized site like this makes it easy for anybody who wants to use these ideas to say, "Look - you can't sue me for using your patent; this idea was published to the world two months before you filed for patent." (IANAL)
As a side effect, patent examiners could use the database to find prior art, but they don't have to use this system for it to help.
Don't expect the USPTO to do the job you want them to do. You know they're broken, and they'll drag us through thousands more bad patents before they change, if they ever do. Here: the new Director of the USPTO as of 2004-01-12 touts his help getting the DMCA passed: Jon Dudas. What more do you need to know?
Fixing the broken USPTO will take time, organization, money, and expertise. Publishing anti-patents outside of any government might be the simplest way. Maybe the EFF could organize the effort.
People outside the US have different laws and jurisdictions to worry about.
But this database might help in those jurisdictions too (e.g. Microsoft patenting obvious XML usage in Europe).
Using a higher sampling rate has to do more with counteracting clock jitter and the error introduced by non ideal equipment.
Also, a higher sampling rate means music signals are more likely to retain phase information at high frequencies.
I'm still surprised after all these years that people appear to think frequency is the only signal in audio. Phase is pretty important in music, and I think sampling at 2x the sampled frequency makes a hash of phase info near the hi end (think of your favorite band's ride cymbal, or most instruments in an orchestra, or electronic music chorusing effects). I believe I hear this effect, but I've not read nor conducted any double-blind test addressing this, so - grain of salt.
For lots of audio, like telephony, phase is less important, but Intel's new spec will certainly be used to present music. Music may sound better when recorded with high sample rates and played back using this spec.
As other comments point out, it takes more than a monopoly or a standard to make a vulnerable software monoculture. It takes a lot of installations & executions of the same software with the same flaw.
If compilers randomize the physical layout of memory allocation and of machine code, it will be harder for exploits to succeed on the same software. Two compiles of the same software will not have the same vulnerabilities.
This is effective when different users run different compiles of your code. (Linux, anyone?)
If compilers generate code that randomizes executable structure at program start time, then you only have to release one compile - all installations can be exactly the same, but two running programs (or OSes) will not have the same vulnerabilities.
This is genetics-in-compilers. Learn from biology and make software diverse when it runs, harder to attack. Evolve compilers.
Why not encrypt your source code, encrypt your executables (doesn't even have to be well encrypted), and "don't use" the patented algorithms. "I have discovered a novel algorithm that does the same thing a new way. Trade secret."
Then if patent lawyers file a claim against you, sic some DMCA lawyers on them and let them eat each other.
Slashdot Mirror
The spec has to have a rock-solid security model required for implementors, and a good security test suite must be freely available. Without these, the database will turn out to be a major hack vector. With a great security model, we only have to worry about bugs. As it stands, the spec covers security very lightly.
The spec has these sections that mean people are at least thinking about security. I hope there are actual security experts involved:
"Thus, strictly following the origin model described in this specification is important for user security."
If you want this thing to succeed, you have an interest in the security model.
Suggestions, if you care about the code you will write in your career:
1) Figure out whether you're learning deep CS fundamentals in your undergrad program. A good Masters program will teach you CS fundamentals that make you a better software engineer in a real job. Some undergrad programs don't do that, and knowing the difference should help you decide.
2) Research (really research) the curricula at some graduate programs. They differ. You can probably tell if any of them sound like you'd learn a lot.
3) If you target the Masters degree, GET IT FAST. Make it a goal to finish the degree and start using it. Do as much programming as possible during the degree program, so you're ready to code for real. Use it as a powerup.
But I can painlessly quit whenever I feel like it. I use a modified binary exponential backoff algorithm. For example, if I'm drinking 1 cup a day:
Backing the dosage off slowly completely avoids headaches. For me. YMMV, but give it a shot. If you usually drink more, you might want to take a few more days (2 days per binary step).
Such extortionate practices should be punished. Disallowing future misdeeds isn't good enough. The RIAA should be (legally) punished for bad behavior. Otherwise there's no real incentive for organizations like that to stop.
I would not say this:
"2) It's hard to do stealthy reentry."
We don't know how hard it is, or whether it's already easy enough. That's the beauty of stealthiness. For example, if a military were doing stealthy re-entries all the time, you might not know.
Before the F-117 and B-2, stealth aircraft were considered too hard to do. But there they are.
Survival of the human species. This is a social & political application, not scientific.
When you read the newspaper for the past 100 years, there's evidence the species might destroy itself. Often, wars happen because people don't understand their place in the universe. For example, a country can be run by people who believe a deity will save the faithful, so world turmoil and war is ok. Other times, wars are a symptom of tyranny (somebody wants "power").
Humans are very visually oriented. Pretty pictures pull the eye in.
As more people gaze at pretty pictures of the gi-normous universe full of stars, more people realize we live at a single star, and there are many many other stars similar to ours. That may reduce destructive effects of belief the deity who will "save" you wants you to kill people.
Gazing at pretty pictures can also inspire people who would otherwise permit tyranny (humans are lazy). A strong enough desire to find out about the pretty pictures can make people impatient to know truth. (e.g. "Why are there more stars in the Antennae galaxies than people on earth?", and "Why should these people be governed like this?")
The above possibilities are weak if you want an application you can apply science or technology to. But they're still important. When your species goes away, it reduces your ability to do science.
Compared to a measly 500-MPH airplane, a terrorist who takes over or blows up a piloted ballistic missle can attack many new targets with new methods.
I want to fly that flight, and I want to float around and have fun. But until the world returns to a politically stable state, I want passengers in the spaceship locked down and/or checked out for 3 days, not 3 hours.
- SBC's stock price will rise as people believe they're doing this big moneymaker, HDTV, wow.
- After the top 10% of SBC management makes at least $20 million each from the stock spike ($100 mil for top 1%), their attention will turn to the next bait. The marketing will continue. Woo hoo.
- Meanwhile, SBC will run into a combination of legal or financial trouble: anti-trust, bad accounting, misleading investors, and/or new legislation from Orrin Hatch.
- Microsoft will deliver software as good as it always is. This will slow the deployment and provide bad publicity as the paying beta testers whine about outages, lack of programming, and cost.
- SBC is a phone company:
-
end-user cost will start at twice the initial marketing estimates
-
the billing model will charge extra for every little thing (10 cents per skipped commercial? $1/hour over 3 hours use per day, outside primetime hours excluding sporting events in overtime?)
-
the project will take 2 years extra due to new legislation or evil competitors.
-
they have expertise in aggressively mass-marketing a low-margin service and maintaining communication infrastructure, not selling a sparkly cool new combination of technology
-
Everybody driving the project will act surprised when there's not enough bandwidth to deliver all the HDTV.
-
Competitive projects will develop standards, and work better, and run on Linux boxes, and provide a better UI designed by a hungry teenager in Banaglore, and will be used by < 1% of the HDTV market since there is no marketing budget.
-
SBC will pull the plug when the project turns into a cost center that no longer spikes the stock price up.
-
HDTV over IP will ever-after sound like a bad idea, since people will remember how this project went.
Of course there's a chance the project will succeed if somebody at the top of SBC is a total freak for HDTV and is a good enough salesman to persuade SBC's board.I don't see the study in the online version right now (2004-10-03 14:15 PDT) but it might be there "now" depending on when you read this. The Journal appears to post PDFs of published articles, but it's a subscription-based journal so maybe not.
I realize people have to post anecdotes about themselves and make conclusions about billions of humans in our species. After all, I believe Sturgeon's Law, and not everybody knows what science actually is.
But does such idiocy have to get modded up and not get modded back down? Ackkk.
View any of the plots of cone absorption, and it is obvious that there are cones detecting light that is not a color you can call red, or green, or blue. Your eyes are surely not narrowband-filtering RGB into your brain. Yet our major media - video, film, print - all show us color using technologies based on this very old compression technique (RGB is a 3-band DCT at every viewed pixel).
Not to mention the randomness thrown into the mixture of RGB sensors and emitters (digicams, video cams, film layers -> printers, televisions, LCDs). If they don't use the same wavelengths of R, G, and B throughout the chain, they cannot stimulate cones the way the photons would have that first went through the camera's lens. The reason this is relevant should be obvious: the cone system is biological. Hint: no two people are alike. The curves are surely different for everyone. If you don't trick the cones well, your brain very likely does notice the compression and it mumbles. "The sunset over the Grand Canyon was better than that; I know it was."
Paintings can be made to look more realistic in places than photo, film, or video systems, because over the years, people have found paints that reflect accurate colors. The painted color you see is reflected/filtered/chosen by microscopic features in paint material; e.g. some materials can actually reflect "yellow" photons instead of reflecting greenish and reddish photons that are also in the stream, so that yellow isn't put into your brain by tricking your cones.
Tickling/tricking the cone system so harshly seems really lame now that we can do better - in the sensors, the compression, and display. Yeah, we have production lines dedicated to RGB spectral compression. Does that mean the world cannot change?
In 100 years, people will view antique RGB photos and laugh. "Look how washed out it is! Couldn't they see?" Future imaging systems might not just be RGB + CMY, but they will use more than RGB.
People use computers for different things. It's stupid to care about people who say "RAID 0 is [good,bad] for all users" because nobody is all-users. RAID 0 is useless if all you do is surf and check mail. RAID 0 is great for rotating 100 MB images in Photoshop.
I have Photoshop, video editing, audio editing apps, and a development environment. Those apps can use the boost from RAID 0, specifically for data i/o. Other apps, including the OS, might get a boost, but reliability is much more important there, so I wouldn''t use the RAID 0 drives for those.
All the benchmarks I've read are basically useless, since I have to make many educated guesses about whether my mix of apps AND tasks will run like the benchmark. You and I might spend the hours to make those guesses, but if 95% of users cannot use the benchmark, why do benchmarking?
Here's an app that would be more useful than any single benchmark:
If they're small enough for me to put on many, many things I own, and if I control the database that maps an ID to a thing, I can walk through my house with an RFID scanner looking for the keys, or the OpenGL book, or my apartment's rental contract, or whatever. Instant pseudo-organization!
There are practical questions:
Set up an anti-patent database, findable on the web through obvious domains (unpatentable.org, anti-patent.org, etc). Accept all submissions of ideas, completely keyworded and timestamped, just as searchable as the USPTO's patent database. It doesn't matter too much if the same idea is submitted twice; advise people to search and avoid that, if you want to save disk space. When you find out an idea in the database is patented, leave it in the database but add references to the patents. Maintain security of the database and accuracy of timestamps, because without credibility the database is worthless.
A centralized site like this makes it easy for anybody who wants to use these ideas to say, "Look - you can't sue me for using your patent; this idea was published to the world two months before you filed for patent." (IANAL)
As a side effect, patent examiners could use the database to find prior art, but they don't have to use this system for it to help.
Don't expect the USPTO to do the job you want them to do. You know they're broken, and they'll drag us through thousands more bad patents before they change, if they ever do. Here: the new Director of the USPTO as of 2004-01-12 touts his help getting the DMCA passed: Jon Dudas. What more do you need to know?
Fixing the broken USPTO will take time, organization, money, and expertise. Publishing anti-patents outside of any government might be the simplest way. Maybe the EFF could organize the effort.
People outside the US have different laws and jurisdictions to worry about. But this database might help in those jurisdictions too (e.g. Microsoft patenting obvious XML usage in Europe).
Also, a higher sampling rate means music signals are more likely to retain phase information at high frequencies.
I'm still surprised after all these years that people appear to think frequency is the only signal in audio. Phase is pretty important in music, and I think sampling at 2x the sampled frequency makes a hash of phase info near the hi end (think of your favorite band's ride cymbal, or most instruments in an orchestra, or electronic music chorusing effects). I believe I hear this effect, but I've not read nor conducted any double-blind test addressing this, so - grain of salt.
For lots of audio, like telephony, phase is less important, but Intel's new spec will certainly be used to present music. Music may sound better when recorded with high sample rates and played back using this spec.
If compilers randomize the physical layout of memory allocation and of machine code, it will be harder for exploits to succeed on the same software. Two compiles of the same software will not have the same vulnerabilities. This is effective when different users run different compiles of your code. (Linux, anyone?)
If compilers generate code that randomizes executable structure at program start time, then you only have to release one compile - all installations can be exactly the same, but two running programs (or OSes) will not have the same vulnerabilities.
This is genetics-in-compilers. Learn from biology and make software diverse when it runs, harder to attack. Evolve compilers.
Then if patent lawyers file a claim against you, sic some DMCA lawyers on them and let them eat each other.
Rough justice for stupid governance.