How about an open-source converter that reads.doc filese and produces a (complete) XML file conforming to these published reference schemas? I say complete because as I recall, their XML export in the latest version of office was half-hearted at best, and lost a LOT of formatting. So by "complete" I mean lossless.
Now THAT would be a real contribution from Microsoft! It would be a kind-of one-sided open source project however. They would have to provide all the core code since only they know what precisely is in the.doc files. Other people could add glitz, GUI wrappers, OpenOffice integration, etc. And unfortunately none of these things actually benefits Microsoft directly.
I agree that a lot of *new* hardware works well with Windows XP because the manufacturers built drivers for it. As Linux gains recognition, more and more manufacturers will realize the need to make Linux drivers, or at least release the underlying protocol specs. This is already starting to happen.
As for how OpenOffice can't read 100% of MSOffice files, your logic leads to an impasse. If the only thing MS did was to continue to introduce bugs, raise the price of Office, degrade the already-poor XML export facility, and continually change the proprietary format, then your users would just pay more and more money and suffer for it, unable to switch to any other, superior office software. OpenOffice does a darn good job of cracking the secret MSOffice formats, but it can't be perfect because MSOffice does not use any standard file formats. You have to see this for what it is -- an open-and-shut case against using Office, and against proliferating Office documents. You have no guarantee that these documents will be readable 5 years from now except with legacy apps on legacy hardware.
Okay, that said, there's still a lot of Windows-only commercial software (besides games) that would be nice for Linux. Software such as OpenInventor (nothing open about it) for drafting, Reason and Pro Tools for audio, Quark Express for graphic design, etc. These are well-developed software suites which I don't see any competition for in the OS community. These kinds of applications are particularly difficult for open-source to tackle, partly because the users of these software are so far removed from the developers working on them, and partly because they are typically used professionally so there is both a money source as well as the need for some accountability/support.
So for Windows one has the standard fare such as Office, Quicken, ssh, acrobat, CD burning, registered winzip and gsview, etc. But you also have most of the open-source software via Win32 ports or cygwin. And in addition you have commercial games and "professional software" apps.
It's hard to argue that Linux, at present, has more software available for it. But what it does have is quite enough for most people, and often better quailty than the Win32/cygwin ports. Plus one has to add a LOT of additional, expensive software to a Windows system to bring it up to snuff. So if you don't need the games or specialty software, use Linux and you'll save several hundred $!
As for hardware setup, I've found it rather difficult to upgrade an old Windows 95,98,ME system to XP and get all the hardware working. It takes a lot of downloading, lots of mucking around and rebooting, and several iterations of updates before you're done.
And even newer computers have problems. A specific example is my 2-year-old Dell laptop with an nvidia video card. It came with XP even, but when reinstalling XP or updating the video driver, NVidia's driver installer doesn't recognize the video card because the ID string was modified by Dell. I had a difficult fight to get that working, partly because my Windows expertise is vanishing and being replaced with Linux expertise. But Nvidia's Linux driver installs itself on this same computer flawlessly. It must be using the numeric device ID instead of the vendor string.
I was surprised to hear that FreeBSD was switching to Xorg. Wasn't the main reason that most Linux distros switched because of a (minor) GPL licence conflict? I'd be surprised if there was any conflict with the new XFree86 license and BSD. My coarse understanding of the BSD licence is: "I'm licensing this code in the most minimal and least obtrusive way because I don't care about licenses, only writing good code. Do whatever you want with it. Enhance it a bit, add your own logos, claim it as your own work and sell it back to me even, I just don't care."
You're billing the "Windows API simulator" as a bad thing, which seems to me like faulty reasoning. It would be impossible to switch to a non-windows operating system and run Windows programs without some kind of emulation, no matter how good that other OS was. If someone isn't going to switch if they have to use an emulator, then they will never switch. That goes for any OS at any time.
AFAIK, wine is the only Windows emulator out there. So if a user is married to any particular Windows-only software, then Linux (or FreeBDS, Solaris, etc.) presents them with the only way out. This, together with LOTS of tangible benefits (security, cost, maintainability, ease of use, availabilty of free high-quality software), is a very GOOD reason to switch.
I guess I'm in the position now that I run Linux on everything that I can, and get very annoyed when I can't. Windows is becoming more and more frustrating for me to use.
BUT, there are still a few Windows-only applications that are important to me, and I'm not convinced that WineX covers all these. Here's an incomplete list for me at present. Note that none of these are games: * Corel Draw: Far superior to oodraw, kontour, xfig, etc. * Pinnacle Studio: I'm not sure this is the best of the quick-n-easy home video editors, and it is the buggiest program I've seen in 25 years, but it can do the job. I need to investigate Cinelerra more, but they never even claim that it's easy to use or fast. * Quicken, Turbo-Tax: Probably quicken can be replaced by gnucash, but I don't think there's any Linux-compatable tax prep software. * Epson R800 printer driver: I'm waiting. But I also want decent color correction. * Cakewalk, and other good audio editing programs out there such as Pro tools. Is there anything that runs on Linux which comes close?
So as much as I want to go 100% Linux, I just can't quite do it. I have to keep one computer at home running Windows. I'm a Linux zealot at heart, but am still intensely grateful for both cygwin and wine, and wish these projects the very best.
Actually, strange-anticharm mesons have been observed and extensively studied for years. I'm too young to know just when the first Ds was discovered, but my guess is in the early 80's. I know these were being carefully studies by the late 80's.
But SELEX's particle is a new excitation of this bound state that hasn't been seen before.
Their preprint (which may change before being accepted for publication) is at: http://xxx.lanl.gov/abs/hep-ex/0406045
Looking at the paper, it appears to be a good result, but I'm surprised that there doesn't seem to be any discussion of systematic errors. More specifically, they don't say anything about how the signal evolves with different event selection cuts. I am very curious to see if the signal survives clean-up cuts on the eta.
I'll believe it when I see independant confirmation. I expect that within the year this can either be confirmed or denied by another experiment such as BELLE, BaBar, CDF, D0, Focus, or CLEO.
Well, it is great to see some headlilnes like this reach Slashdot!
I've been dual-booting linux/WinXP for a while now, too. Even though I only use XP about 5% of the time, after a few months it gets bogged down with registry clutter and such, and begins running much slower than a fully-loaded Linux system. I recently lost my hard drive and re-installed both OSes from scratch. I've installed very little on WinXP now, and it boots faster than Linux but still runs a bit slower. It feels to me like Windows systems are entropic and degrade over time, while Linux systems do not.
BTW, I've been using the Gentoo distro for over a year now, and absolutely love it. I think that with the same window manager (I use KDE on fast systems, and gnome or XFCE on slower ones) Gentoo is generally faster.
WinXP definately does not multitask as smoothly as Linux (2.6 kernel) does. I can watch DVDs in Linux without any hiccups in the framerate, but in Windows the video freezes up every minute or so. (Shutting down wireless networking and anti-virus helps marginally but doesn't eliminate the problem.) When running Neverwinter Nights, the framerate is better in Linux with the same resolution settings, but the mouse cursor is not hardware-accellerated so in the end it feels more responsive in Windows. but that's Bioware's fault, not Linux.
You can also optimize your Linux system in lots of ways to increase performance if it feels too draggy to you. You can show only window outlines when you drag them, for example. Also antialiased fonts, while nice, really slow things down. On very old computers (I have a 266 MHz dinosaur behind me) I turn off AA completely. I think that Windows renders AA fonts faster.
Gentoo Linux also deals with bloat nicely. You decide what packages you want to install, and can even decide what dependencies you want to have to a very high degree. On my toy 266MHz computer I am gradually playing with other Linux distros. I could barely get Fedore Core 2 to install! I had to do it in text mode otherwise the package selection process required too much memory. Running Fedora I had a *very* slow system whereas with a lean Gentoo installation it's quite tollerable. I also had more hardware compatability problems with Fedora than with Gentoo. I didn't expect that to happen!
Also, with the Gentoo live CD, I was able to move my existing Gentoo installation over to an LVM2-managed partition. Whether or not you use Gentoo, the live CD is a must-have item in the CD wallet.
I thought that zero-day refered to a guarded vault of exploits used by script kiddies. Maybe it means both? Googling for the term comes up with a mix of interpretations. If the exploit is discovered before the security teams know about the hole, it's also zero-day.
I was confused by this since the original post made me think that Jelmer authored the exploit. Reading the article made it clear that Jelmer found (was sent) the exploit first and studied it before understanding how to patch it. So in this case the exploit had been in the wild before it was "discovered."
The article also suggests (is this just spin?) that the exploit was not "so simple" at all.
This reminds me of a physics qualifying exam problem. If you fire your rockets on the nuclear-waste cargo ship straight toward the Sun and then coast, from Earth it would look like the ship heads toward the Sun, starts to veer to the right (if North is up), and follows a circular path that leads it right back to the Earth 1 year later. It's just like that boomerang that comes around and hits Wile E. Coyote in the back of the head. (This is ignoring Earth's gravity.)
As others pointed out, you have to cancel the orbital motion. I wonder if solar power or solar sails could be used to do this?
A google search turned up this discussion.
Here people point out that it is cheaper to send the waste anywhere _but_ the Sun. I like the idea of putting it in orbit. Crashing it into Venus of Jupiter seems like it might impact future science missions to those planets by introducing foreign substances to some very minor degree. They also point out that politics is the main problem, not cost. Another significant problem is that of disasters during launch. It's hard to imagine a launch system that is both cheap and has an acceptable failure probability. Would 1 in 1 billion be good enough for you? I don't know, probably not for me.
Sorry, slightly OT here. I always wondered why a sysadmin ran nightly defragmentation jobs on home disks on an OSF1 system. Every night from about 1am to 4am, the system would become completely useless. By "useless" I mean it might take 30 seconds to respond to "ls" in a directory with 20 files. Productive work was impossible, and overnight jobs weren't getting much time. Logging onto the file server, I would see about 10 defrag proceses running. I never talked to him much about it, but from your post I'm surprised it was necessary at all, particularly on a DAILY basis!
I don't know what filesystem he was using -- I know virtually nothing about OSF1 from a sysadmin perspective. Well, this was 5 years ago. Now I think he's still running OSF1 for a fileserver that you can't log into, because the OSF1 filesystem is "so much more advanced" than any Linux offerings. Now with LVM2 and the various mature journaling filesystems in Linux, I very much doubt that's true anymore. He's always been of the mindset that you get what you pay for. Our compute servers are all running Linux these days, and that's what I really care about.
I just read it as a gender-neutral statement. Switching gender mid-sentence is sometimes less obtrusive then saying "his/her" all the time, and more personal than using "their" and "they".
It seems clear that digital information has an increasing need for copy protection. If Palladium fails (and I hope it does), what are we left with?
Aside from proprietary software, music, books, and videos, I suppose in a future with molecular nanotechnology manufacturing, almost everything could be purchased as digital iniformation. Just download the design file and software for that new palmtop or whatever, send it to your home nanofactory, and voila! I can see two possibilities. One is that all nano-engineering work is to be licensed with some equivalent to a GPL. Since this would presumably apply to almost all commodoties we have today, this requires a new economic model beyond the free market or capitalism! Sounds cool to me, but I have no idea what it would be. The other alternative is less revolutionary, but absolutely requires an effective DRM technology.
I've been reinstalling gentoo over the past few days on my laptop (old HD crashed), and am using distcc this time around. There is some speedup, but according to the cache hits stats, it's not much -- maybe 10%. I wouldn't expect *any* speedup on first-time compilation of anything. I wonder how that works?
That was off-topic. Here's my on-topic post then: My top-ten Linux apps are:
emacs kde firefox gimp gqview acroread xmm s openoffice ogle mplayer tetex wine
Oops, that was a top-twelve. And I'm not including things like gcc, samba, cups, reiserfs, tar, and other such system-level utilities. I guess I could bump wine into the low-level group and only have a top-eleven.
I hear that a lot of high-optical-zoom digicams are sacrificing wide-angle capability for further zoom. You can easily find stats like "6x optical zoom" on some of the zoomier cameras these days, but you never see the whole range, such as "1.5x - 6x". Some of the nicer cameras (not just DSLR's) allow you to change lenses, which is really what you want if you need good telephoto.
I think the most important characteristics of a digicam are: * true resolution (distinct horizontal and vertical lines across the picture) * zoom range (both min. and max. optical zoom) * effective f-stop (light gathering) at a given zoom * picture-to-picture delay
Unfortunately, none of these stats are easy to find on a box or placcard. But some websites such as www.imaging-resource.com work these out for you.
I'm reading this article late, and now I've been trying to find an intelligent discussion of the issues. I feel that the article was way off the mark, and your post comes the closest to a good discussion that I've found. Thanks!
The article argues that a smaller CCD chip will have more noise problems than a bigger one. It's not obvious to me that this is true. What one needs is more light-gathering power -- a wider aperture. Whether a lens with a given aperture focuses it's light down to a smaller or larger area is irrelevant. Each pixel would receive the same amount of light in either case. But it's partially true that a higher-resolution camera needs a larger lens to achieve the same S/N noise ratio per pixel. Well that's exactly true for photon counting-statistics noise, but there's also thermal noise, etc. I don't know which noise source is more prominent. Certainly in astrophotography, where one is dealing with much longer exposure times, thermal noise is the killer if one doesn't cool the CCD a lot.
So what are the image-sensor-size issues really? You mention chromatic abberation near the boundaries of the projected image circle. Putting, for example, a lens designed for a Canon 35mm camera onto a D300 with it's APS-sized CMOS sensor will give you less chromatic abberation because the lens is essentially over-designed for that camera. For regular CCD cameras, the lenses are specifically designed for the CCD and so chromatic abberation is entirely a question of how well-made the lenses are. It's not fundamentally a matter of sensor size.
There is also the issue, for CCD's anyway, of charge bleeding from one site to another. That effect will increase as the pixel density and light-gathering increases. But again I don't know if it's an important effect in practice.
So I don't think the article's claims are well-supported yet. But I certainly agree that the MP count of a camera does not equate with it's resolution. It obviously puts an upper bound on the resolution, nothing more. Dave at http://www.imaging-resource.com has done methodic tests of digital camera resolution and many other camera aspects in a very nice way. It turns out that although there is a lot of variation between the models, generally the higher-MP cameras have a better resolution. That doesn't happen if you just swap CCD's and keep the optics the same. But most companies realize the trade-offs involved in camera manufacturing, and will usually put better optics (including aperture) in a higher-MP camera.
I'm not an expert in digital photography, but I know my physics and I've been enjoying the Canon Digital Rebel (D300) for the past month now. I like it more and more as I use it. I've taken about 2400 pictures so far, so I feel that I've already saved a lot of money over a film camera. I hear pros say that they take about 100 pictures for every really good one. I'm not a pro so I'll have to take a lot more! With this attitude, and knowing that practice makes perfect, film development costs alone would bankrupt me before I become a good photographer with a film camera. So I like the economics of the D300 over a film SLR. The only question now is whether or not it's as good as a film SLR. Right now I think it's better.
As for resolution, I don't really know what the grain count is for ISO100 35mm film. I see one report that it's about 13-15 million grains. (Is this only relevant to B&W photography though?) But from what I've seen so far, my Digital Rebel (6 MP) photos show as much detail as a film camera. The prints on an Epson R800 are just like film prints. So as far as I can tell, the only difference with the digital SLR is that I can choose my "film speed" with the press of a button, and I don't have to pay anything for all the bad photos I take.
What I found fascinating is that choosing a faster "film speed" on the D300 gives you nearly the same effect as using a real high-speed film, but for rather different reasons. Yes, fundamentally less light give
This reminds me of a problem I now have with my turntable. Yes, I still have a few vinyl records which I would happily replace with CD's given the opportunity. Anyway there appears to be a direct DC coupling from the record player through my tuner/amp to my speakers. When I play records, the little wobbles in them translate directly into rather severe visible motion of the speaker's woofers at around 2-10 Hz. I dare not play these recordings too loudly, since I don't really want to be probing the limits of my speaker cones. I obviously need to add some kind of passive highpass filter in this system, but I haven't yet found a good (and cheap) way to do this. I'm sure it's not hard to build yourself. I have some experience with audio electronics, but not quite enough yet to know the details of a turntable signal (what's the impedance?) and how to design the circuit accordingly.
Partly I'm looking for casual advice on the subject, and partly I'm warning that DC coupling, especially for vinyl, is a bad thing. One of the main features of professional audio equipment is good protection circuitry, and this includes AC coupling (for example balanced audio lines). Since we can't really hear below 20 Hz, anything lower than that is electronic noise which is best rejected.
I know this site doesn't go into as much depth as a book could, but it's a good start. I honestly think that this material should be the basis of a required course in high school, or at least interwoven into english and/or history courses. The widespread use of propaganda is perhaps the greatest threat to democracy today.
As long as people are running with X11R6 servers, they will want new apps compiled against Y-windows or whatever to run. Does the planned X11R6 compatability layer work both ways?
I think the optional extensions are great -- this allows for multiple implementations of X without an unbearable burden of supporting *everything*. If the process of natural selection is going to work, you want a protocol that's not overly demanding.
By separating the widgets and toolkits from the X server, you again reduce complexity and allow for multiple implementations at these different layers. I guess the widgets and themes could somehow be linked into the X server, but now I'm rapidly entering the realm of speculation.
I keep hearing the argument that X is bad because it's old. Unix is even older, and look what happened with it! I'm not saying you're wrong, I just want to hear about the details. Is it necessarily true that the entire protocol is insufficient and cannot be satisfactorily extended? Why not?
I guess efficiency could be the main issue. In part it's a matter of how the clients are designed. I've seen some relatively simple (just stock widgets) Java programs that, with Sun's VM, absolutely crawl over a 10 MBit connection. I mean that it takes half a minute to redraw a quarter of a window or pop up a menu. Compression seems not to help much because it's all limited by latency. (I can imagine the communique: "Can I draw this pixel here?" "What color?" "Red." "Okay." "You sure?" "pretty sure."...) On the other hand, I've seen glitzier programs (emacs, even Mozilla) perform just fine through the same network. But I can see room for improvement. VNC seems to be faster in some ways. And Microsoft has some technology that does a similar thing fairly quickly, but that's relying on a higher-level protocol passing Windows API calls or something. So that's what you mean about integrating the widgets and toolkits somehow.
I totally agree here. Basically, the present voting system is such that the two-party system is a stable equilibrium. I'm not entirely sure why this is true, but it's not hard to see some of the reasons and history offers the proof. If Nader got any significant votes (and last time even 1% was significant enough), he would ensure Bush's reelection. So to the party he is more closely aligned with (Democratic) he's enemy #1. The only people who should properly be voting for him are those who hate the two leading candidates absolutely equally. Since that's a small number of people, the two front-running parties remain unchallengable. If I end up disliking both front-runners equally this time around, I'll either go with Nader or "Bufgoo" (the google hit is correct).
I herad in Ireland they can vote for a 2nd and 3rd choice (etc?), and if their 1st choice is loosing the vote goes to the 2nd. I think this upsets the equillibrium and would truly allow the public to represent themselves with their votes. Unfortunately I don't see any hope for a system like this one to be established in the US in my lifetime. Another possible way to upset the 2-party system is if someone new suddenly comes on the scene with unbelievable charisma so that they immediately gain a large fraction of votes. They would also have to be a middle-of-the-road type of candidate. Nader is neither immensely popular nor middle-of-the-road. And anyway this kind of singular event would not really upset the equillibrium, it would simply establish a new party/parties and there would still eventually be two. 3 -> 1 -> 2 I think.
If you need 15 examples in order to solve 15 problems on some general topic, I might suggest that you haven't learned anything.
If solving a problem becomes a monkey-see, monkey-do type of excersize where you've been trained to use specific techniques on certain homework problems, then the student is practicing a technique but not understanding the subject. If the homework problems make the student think a bit and extend those "examples" in new ways, then they might be learning. Hopefully calc students expect that at the end of the course they can solve real-world problems that haven't been solved before, and apply the tools of calculus in ways that they haven't been explicitly taught. If they can't, then the entire course was a waste of time.
A good professor should be able to help any student gain this kind of working understanding of their subject, provided the student is also willing to work as hard as necessary. But there are a lot of professors out there that aren't that good. Since students don't often have much choice in the matter, they might have to look for help elsewhere.
Another problem is that students who have spent more time rehearsing techniques (recipies) and less time actually learning math tend to do better on timed, standardized tests. So to some extent the system punishes good students and teachers.
Thanks, I was searching hard for an on-topic post and am glad to see yours. The question is: How can people make money while writing open-source code?
I think the short answer is that programmers can be (and are being) paid for writing open source that satisfies the immediate and particular needs of the company they work for.
The long answer is to compare two models of software and its place in our economy. One model is the proprietary code one, where there are two types of companies: one which produces software, and one which uses the software. In this model there is a strong distinction also between computer users and programmers.
The other model is one where software is free, but programmers will be hired by companies to customize and extend software to meet their peculiar needs. This model tends to blur the distinction between programmers and users. Computers run programs, and so to use a computer is in some sense to program it.
The second model is world where software tools are more effectively leveraged and more valuable. So libraries become more complete, languages more powerful, and programming becomes easier for everyone. Good documentation is also valued.
The first model encourages building a higher barrier between users and programmers, so that the trade secrets needed to program are a commodity by themselves. That is the transparent mindset of the author of the letter. The attitude is that the knowledge of programming is one's net worth, and that giving it away by writing free code somehow lessens your worth.
I think that the proprietary software model does not work as well in a free market economy, where competition is the driving force behind innovation. In this model, competition necessarily leads to duplication of effort. Also having multiple competing proprietary OSes or software suites will multiply the need for specialized knowledge, and thus divide the value if a programmer's training. For example, if a DB programmer was trained in using Microsoft Access, but not trained in the equally-popular Nanosoft Gain (tm), they would only qualify for half of the current job openings. This explains Microsoft's delusions of benevolence. By dominating the market, they think they are reducing duplicated efforts to compete aginst them, and increasing the value of MS-trained programmers. In the world where software is a commodity, software companies naturally gravitate toward a monopoly.
Competition still thrives in the OS world, but software is not one of the trade commodities therein. Companies use software to help them be competetive with other goods or services.
So for these reasons and many more, companies are beginning to realize the value in the open software model. This model requires a critical mass of free software for it to work well, but I think that in many areas we've reached that critical mass.
Well, at least that's they way I see it. But I'm working in acedamia as a post-doc researcher, and don't have any comp-sci degrees. So I'm not experiencing any of this first-hand. But a lot of what I do is write (and use heavily) open-source code. My job requires that I be a good programmer, but that's not my profession. So naturally I support the OS model since it allows "amateur" programmers like me do my job. But my guesses about OS programming jobs in the industry is just from hearsay, so it's good to see other posts confirming it.
You could make a hat out of mu-metal. It's usually heavy stuff, and a bit expensive.
Seriously, this article reminds me of a presentation of magnetotherapy (I think David Letterman aired the clip) where the demonstrator was explaining that "you treat the dog along the diagonal." Absolutely hysterical!
I read a tiny bit of the paper just now, and it seems at least partially legit. But I'm no expert so it's hard for me to tell. Many of these kinds of results are based on 2-sigma confidence levels, which means there's about a 5% chance (rate) of a bogus report. So multiply 5% by the number of studies being done (maybe tens of thousands per year), and you get the rate of BS that we all hear about. Of course we laymen never hear about the experiments with negative results, so we prefferentially tune in to the statistical fluctuations. But this paper does cite a lot of earlier work demonstrating an effect.
Assuming that the article has some legitimacy, I wonder how many hours of shaving corresponds to 48 hours at.01 mT? I would think a shaver is much weaker than that, and scale in the distance factors too. But I'm just guessing.
Another hole in the analysis is that it was done on rat brains, and the study showed a number of ways the effect could be negated by various vitamins and such. Do humans have any of these or other ways of blocking the effects? Unfortunately you can't just buy a few dozen laboratory-grade human beings and run tests on them...
What I'm most disturbed by is the lack of discussion about science and technology in the US election coverage. I had to probe very deeply on the web in order to find any kind of statements about the candidate's positions on funding basic science in the 2000 election. I am finding a similar lack of coverage this year as well. Does the public really not care about these issues?
What I finally dug up in 2000 was that Bush thought that basic science should all be privately funded even though there's no short-term motivation for a company to fund a branch of science whos discoveries are public domain. So that statement was ludicrous. Gore's statement was similarly hard to find, and not quite so insane. Something like "funding levels should be on a par with applied science." Now, whether or not they have or would have made good on their campaign platform is mostly beside the point. Since there was NO coverage in the media on these issues, it doesn't matter what the candidates say on page 127, 2nd paragraph, 3rd sentence of some party manifesto. And for all I know they could think that a megabyte is an awfully big sandwich at McD's!
What can we do to encourage more discussion of these issues in the press?
As for tax cuts... I'm more concerned with the wise spending of my taxes than whether or not they go up or down 1%.
I like an idea posted earlier of using a single port to knock on, that looks for a certain key signature in the first packet. This can be sniffed more easily than a port-knocking sequence, but...
If you don't have too many users on a system, you could use a key sequence. Each user would be given a key generator, which generates a near-infinite number of keys in a certain sequence. The host system would be looking for a particular key, and its matching key generator would then advance once it's been used.
This system would have to be used with a single-port type of access because a client requires a simple (trinary) response from the server. The response can be either "used already", "wrong", or "right". The client then tries keys in turn until it receives the "unused" signal, at which point it connects on the desired port. If there were a lot of active users of the system, this process might take some time. If the only server responses were "used already" or "unused" (wrong or right), then a client could get permanently off track if packets were lost because it would not know that it went beyond the sequence.
This system might still have problems with simultaneous logins. And it is "opening up" an additional port. But I can't think of any way that the knocking port alone could be used for a remote exploit. One could also abuse this system if one could selectively block transmissions both ways. I don't know if that's possible or not.
Slashdot Mirror
How about an open-source converter that reads .doc filese and produces a (complete) XML file conforming to these published reference schemas? I say complete because as I recall, their XML export in the latest version of office was half-hearted at best, and lost a LOT of formatting. So by "complete" I mean lossless.
.doc files. Other people could add glitz, GUI wrappers, OpenOffice integration, etc. And unfortunately none of these things actually benefits Microsoft directly.
Now THAT would be a real contribution from Microsoft! It would be a kind-of one-sided open source project however. They would have to provide all the core code since only they know what precisely is in the
I agree that a lot of *new* hardware works well with Windows XP because the manufacturers built drivers for it. As Linux gains recognition, more and more manufacturers will realize the need to make Linux drivers, or at least release the underlying protocol specs. This is already starting to happen.
As for how OpenOffice can't read 100% of MSOffice files, your logic leads to an impasse. If the only thing MS did was to continue to introduce bugs, raise the price of Office, degrade the already-poor XML export facility, and continually change the proprietary format, then your users would just pay more and more money and suffer for it, unable to switch to any other, superior office software. OpenOffice does a darn good job of cracking the secret MSOffice formats, but it can't be perfect because MSOffice does not use any standard file formats. You have to see this for what it is -- an open-and-shut case against using Office, and against proliferating Office documents. You have no guarantee that these documents will be readable 5 years from now except with legacy apps on legacy hardware.
Okay, that said, there's still a lot of Windows-only commercial software (besides games) that would be nice for Linux. Software such as OpenInventor (nothing open about it) for drafting, Reason and Pro Tools for audio, Quark Express for graphic design, etc. These are well-developed software suites which I don't see any competition for in the OS community. These kinds of applications are particularly difficult for open-source to tackle, partly because the users of these software are so far removed from the developers working on them, and partly because they are typically used professionally so there is both a money source as well as the need for some accountability/support.
So for Windows one has the standard fare such as Office, Quicken, ssh, acrobat, CD burning, registered winzip and gsview, etc. But you also have most of the open-source software via Win32 ports or cygwin. And in addition you have commercial games and "professional software" apps.
It's hard to argue that Linux, at present, has more software available for it. But what it does have is quite enough for most people, and often better quailty than the Win32/cygwin ports. Plus one has to add a LOT of additional, expensive software to a Windows system to bring it up to snuff. So if you don't need the games or specialty software, use Linux and you'll save several hundred $!
As for hardware setup, I've found it rather difficult to upgrade an old Windows 95,98,ME system to XP and get all the hardware working. It takes a lot of downloading, lots of mucking around and rebooting, and several iterations of updates before you're done.
And even newer computers have problems. A specific example is my 2-year-old Dell laptop with an nvidia video card. It came with XP even, but when reinstalling XP or updating the video driver, NVidia's driver installer doesn't recognize the video card because the ID string was modified by Dell. I had a difficult fight to get that working, partly because my Windows expertise is vanishing and being replaced with Linux expertise. But Nvidia's Linux driver installs itself on this same computer flawlessly. It must be using the numeric device ID instead of the vendor string.
I was surprised to hear that FreeBSD was switching to Xorg. Wasn't the main reason that most Linux distros switched because of a (minor) GPL licence conflict? I'd be surprised if there was any conflict with the new XFree86 license and BSD. My coarse understanding of the BSD licence is: "I'm licensing this code in the most minimal and least obtrusive way because I don't care about licenses, only writing good code. Do whatever you want with it. Enhance it a bit, add your own logos, claim it as your own work and sell it back to me even, I just don't care."
You're billing the "Windows API simulator" as a bad thing, which seems to me like faulty reasoning. It would be impossible to switch to a non-windows operating system and run Windows programs without some kind of emulation, no matter how good that other OS was. If someone isn't going to switch if they have to use an emulator, then they will never switch. That goes for any OS at any time.
AFAIK, wine is the only Windows emulator out there. So if a user is married to any particular Windows-only software, then Linux (or FreeBDS, Solaris, etc.) presents them with the only way out. This, together with LOTS of tangible benefits (security, cost, maintainability, ease of use, availabilty of free high-quality software), is a very GOOD reason to switch.
Agreed.
I guess I'm in the position now that I run Linux on everything that I can, and get very annoyed when I can't. Windows is becoming more and more frustrating for me to use.
BUT, there are still a few Windows-only applications that are important to me, and I'm not convinced that WineX covers all these. Here's an incomplete list for me at present. Note that none of these are games:
* Corel Draw: Far superior to oodraw, kontour, xfig, etc.
* Pinnacle Studio: I'm not sure this is the best of the quick-n-easy home video editors, and it is the buggiest program I've seen in 25 years, but it can do the job. I need to investigate Cinelerra more, but they never even claim that it's easy to use or fast.
* Quicken, Turbo-Tax: Probably quicken can be replaced by gnucash, but I don't think there's any Linux-compatable tax prep software.
* Epson R800 printer driver: I'm waiting. But I also want decent color correction.
* Cakewalk, and other good audio editing programs out there such as Pro tools. Is there anything that runs on Linux which comes close?
So as much as I want to go 100% Linux, I just can't quite do it. I have to keep one computer at home running Windows. I'm a Linux zealot at heart, but am still intensely grateful for both cygwin and wine, and wish these projects the very best.
Actually, strange-anticharm mesons have been observed and extensively studied for years. I'm too young to know just when the first Ds was discovered, but my guess is in the early 80's. I know these were being carefully studies by the late 80's.
But SELEX's particle is a new excitation of this bound state that hasn't been seen before.
Their preprint (which may change before being accepted for publication) is at:
http://xxx.lanl.gov/abs/hep-ex/0406045
Looking at the paper, it appears to be a good result, but I'm surprised that there doesn't seem to be any discussion of systematic errors. More specifically, they don't say anything about how the signal evolves with different event selection cuts. I am very curious to see if the signal survives clean-up cuts on the eta.
I'll believe it when I see independant confirmation. I expect that within the year this can either be confirmed or denied by another experiment such as BELLE, BaBar, CDF, D0, Focus, or CLEO.
Well, it is great to see some headlilnes like this reach Slashdot!
So, 1+0 is better than 0+1? Well there goes algebra. My education is already obsolete!
(For some reason non-commutativity scares me more than 1+0 = 10.)
I've been dual-booting linux/WinXP for a while now, too. Even though I only use XP about 5% of the time, after a few months it gets bogged down with registry clutter and such, and begins running much slower than a fully-loaded Linux system. I recently lost my hard drive and re-installed both OSes from scratch. I've installed very little on WinXP now, and it boots faster than Linux but still runs a bit slower. It feels to me like Windows systems are entropic and degrade over time, while Linux systems do not.
BTW, I've been using the Gentoo distro for over a year now, and absolutely love it. I think that with the same window manager (I use KDE on fast systems, and gnome or XFCE on slower ones) Gentoo is generally faster.
WinXP definately does not multitask as smoothly as Linux (2.6 kernel) does. I can watch DVDs in Linux without any hiccups in the framerate, but in Windows the video freezes up every minute or so. (Shutting down wireless networking and anti-virus helps marginally but doesn't eliminate the problem.) When running Neverwinter Nights, the framerate is better in Linux with the same resolution settings, but the mouse cursor is not hardware-accellerated so in the end it feels more responsive in Windows. but that's Bioware's fault, not Linux.
You can also optimize your Linux system in lots of ways to increase performance if it feels too draggy to you. You can show only window outlines when you drag them, for example. Also antialiased fonts, while nice, really slow things down. On very old computers (I have a 266 MHz dinosaur behind me) I turn off AA completely. I think that Windows renders AA fonts faster.
Gentoo Linux also deals with bloat nicely. You decide what packages you want to install, and can even decide what dependencies you want to have to a very high degree. On my toy 266MHz computer I am gradually playing with other Linux distros. I could barely get Fedore Core 2 to install! I had to do it in text mode otherwise the package selection process required too much memory. Running Fedora I had a *very* slow system whereas with a lean Gentoo installation it's quite tollerable. I also had more hardware compatability problems with Fedora than with Gentoo. I didn't expect that to happen!
Also, with the Gentoo live CD, I was able to move my existing Gentoo installation over to an LVM2-managed partition. Whether or not you use Gentoo, the live CD is a must-have item in the CD wallet.
I thought that zero-day refered to a guarded vault of exploits used by script kiddies. Maybe it means both? Googling for the term comes up with a mix of interpretations. If the exploit is discovered before the security teams know about the hole, it's also zero-day.
I was confused by this since the original post made me think that Jelmer authored the exploit. Reading the article made it clear that Jelmer found (was sent) the exploit first and studied it before understanding how to patch it. So in this case the exploit had been in the wild before it was "discovered."
The article also suggests (is this just spin?) that the exploit was not "so simple" at all.
This reminds me of a physics qualifying exam problem. If you fire your rockets on the nuclear-waste cargo ship straight toward the Sun and then coast, from Earth it would look like the ship heads toward the Sun, starts to veer to the right (if North is up), and follows a circular path that leads it right back to the Earth 1 year later. It's just like that boomerang that comes around and hits Wile E. Coyote in the back of the head. (This is ignoring Earth's gravity.)
As others pointed out, you have to cancel the orbital motion. I wonder if solar power or solar sails could be used to do this?
A google search turned up this discussion. Here people point out that it is cheaper to send the waste anywhere _but_ the Sun. I like the idea of putting it in orbit. Crashing it into Venus of Jupiter seems like it might impact future science missions to those planets by introducing foreign substances to some very minor degree. They also point out that politics is the main problem, not cost. Another significant problem is that of disasters during launch. It's hard to imagine a launch system that is both cheap and has an acceptable failure probability. Would 1 in 1 billion be good enough for you? I don't know, probably not for me.
Sorry, slightly OT here. I always wondered why a sysadmin ran nightly defragmentation jobs on home disks on an OSF1 system. Every night from about 1am to 4am, the system would become completely useless. By "useless" I mean it might take 30 seconds to respond to "ls" in a directory with 20 files. Productive work was impossible, and overnight jobs weren't getting much time. Logging onto the file server, I would see about 10 defrag proceses running. I never talked to him much about it, but from your post I'm surprised it was necessary at all, particularly on a DAILY basis!
I don't know what filesystem he was using -- I know virtually nothing about OSF1 from a sysadmin perspective. Well, this was 5 years ago. Now I think he's still running OSF1 for a fileserver that you can't log into, because the OSF1 filesystem is "so much more advanced" than any Linux offerings. Now with LVM2 and the various mature journaling filesystems in Linux, I very much doubt that's true anymore. He's always been of the mindset that you get what you pay for. Our compute servers are all running Linux these days, and that's what I really care about.
I just read it as a gender-neutral statement. Switching gender mid-sentence is sometimes less obtrusive then saying "his/her" all the time, and more personal than using "their" and "they".
It seems clear that digital information has an increasing need for copy protection. If Palladium fails (and I hope it does), what are we left with?
Aside from proprietary software, music, books, and videos, I suppose in a future with molecular nanotechnology manufacturing, almost everything could be purchased as digital iniformation. Just download the design file and software for that new palmtop or whatever, send it to your home nanofactory, and voila! I can see two possibilities. One is that all nano-engineering work is to be licensed with some equivalent to a GPL. Since this would presumably apply to almost all commodoties we have today, this requires a new economic model beyond the free market or capitalism! Sounds cool to me, but I have no idea what it would be. The other alternative is less revolutionary, but absolutely requires an effective DRM technology.
I've been reinstalling gentoo over the past few days on my laptop (old HD crashed), and am using distcc this time around. There is some speedup, but according to the cache hits stats, it's not much -- maybe 10%. I wouldn't expect *any* speedup on first-time compilation of anything. I wonder how that works?
m s
That was off-topic. Here's my on-topic post then:
My top-ten Linux apps are:
emacs
kde
firefox
gimp
gqview
acroread
xm
openoffice
ogle
mplayer
tetex
wine
Oops, that was a top-twelve. And I'm not including things like gcc, samba, cups, reiserfs, tar, and other such system-level utilities. I guess I could bump wine into the low-level group and only have a top-eleven.
I hear that a lot of high-optical-zoom digicams are sacrificing wide-angle capability for further zoom. You can easily find stats like "6x optical zoom" on some of the zoomier cameras these days, but you never see the whole range, such as "1.5x - 6x". Some of the nicer cameras (not just DSLR's) allow you to change lenses, which is really what you want if you need good telephoto.
I think the most important characteristics of a digicam are:
* true resolution (distinct horizontal and vertical lines across the picture)
* zoom range (both min. and max. optical zoom)
* effective f-stop (light gathering) at a given zoom
* picture-to-picture delay
Unfortunately, none of these stats are easy to find on a box or placcard. But some websites such as www.imaging-resource.com work these out for you.
I'm reading this article late, and now I've been trying to find an intelligent discussion of the issues. I feel that the article was way off the mark, and your post comes the closest to a good discussion that I've found. Thanks!
The article argues that a smaller CCD chip will have more noise problems than a bigger one. It's not obvious to me that this is true. What one needs is more light-gathering power -- a wider aperture. Whether a lens with a given aperture focuses it's light down to a smaller or larger area is irrelevant. Each pixel would receive the same amount of light in either case. But it's partially true that a higher-resolution camera needs a larger lens to achieve the same S/N noise ratio per pixel. Well that's exactly true for photon counting-statistics noise, but there's also thermal noise, etc. I don't know which noise source is more prominent. Certainly in astrophotography, where one is dealing with much longer exposure times, thermal noise is the killer if one doesn't cool the CCD a lot.
So what are the image-sensor-size issues really? You mention chromatic abberation near the boundaries of the projected image circle. Putting, for example, a lens designed for a Canon 35mm camera onto a D300 with it's APS-sized CMOS sensor will give you less chromatic abberation because the lens is essentially over-designed for that camera. For regular CCD cameras, the lenses are specifically designed for the CCD and so chromatic abberation is entirely a question of how well-made the lenses are. It's not fundamentally a matter of sensor size.
There is also the issue, for CCD's anyway, of charge bleeding from one site to another. That effect will increase as the pixel density and light-gathering increases. But again I don't know if it's an important effect in practice.
So I don't think the article's claims are well-supported yet. But I certainly agree that the MP count of a camera does not equate with it's resolution. It obviously puts an upper bound on the resolution, nothing more. Dave at http://www.imaging-resource.com has done methodic tests of digital camera resolution and many other camera aspects in a very nice way. It turns out that although there is a lot of variation between the models, generally the higher-MP cameras have a better resolution. That doesn't happen if you just swap CCD's and keep the optics the same. But most companies realize the trade-offs involved in camera manufacturing, and will usually put better optics (including aperture) in a higher-MP camera.
I'm not an expert in digital photography, but I know my physics and I've been enjoying the Canon Digital Rebel (D300) for the past month now. I like it more and more as I use it. I've taken about 2400 pictures so far, so I feel that I've already saved a lot of money over a film camera. I hear pros say that they take about 100 pictures for every really good one. I'm not a pro so I'll have to take a lot more! With this attitude, and knowing that practice makes perfect, film development costs alone would bankrupt me before I become a good photographer with a film camera. So I like the economics of the D300 over a film SLR. The only question now is whether or not it's as good as a film SLR. Right now I think it's better.
As for resolution, I don't really know what the grain count is for ISO100 35mm film. I see one report that it's about 13-15 million grains. (Is this only relevant to B&W photography though?) But from what I've seen so far, my Digital Rebel (6 MP) photos show as much detail as a film camera. The prints on an Epson R800 are just like film prints. So as far as I can tell, the only difference with the digital SLR is that I can choose my "film speed" with the press of a button, and I don't have to pay anything for all the bad photos I take.
What I found fascinating is that choosing a faster "film speed" on the D300 gives you nearly the same effect as using a real high-speed film, but for rather different reasons. Yes, fundamentally less light give
This reminds me of a problem I now have with my turntable. Yes, I still have a few vinyl records which I would happily replace with CD's given the opportunity. Anyway there appears to be a direct DC coupling from the record player through my tuner/amp to my speakers. When I play records, the little wobbles in them translate directly into rather severe visible motion of the speaker's woofers at around 2-10 Hz. I dare not play these recordings too loudly, since I don't really want to be probing the limits of my speaker cones. I obviously need to add some kind of passive highpass filter in this system, but I haven't yet found a good (and cheap) way to do this. I'm sure it's not hard to build yourself. I have some experience with audio electronics, but not quite enough yet to know the details of a turntable signal (what's the impedance?) and how to design the circuit accordingly.
Partly I'm looking for casual advice on the subject, and partly I'm warning that DC coupling, especially for vinyl, is a bad thing. One of the main features of professional audio equipment is good protection circuitry, and this includes AC coupling (for example balanced audio lines). Since we can't really hear below 20 Hz, anything lower than that is electronic noise which is best rejected.
I know this site doesn't go into as much depth as a book could, but it's a good start. I honestly think that this material should be the basis of a required course in high school, or at least interwoven into english and/or history courses. The widespread use of propaganda is perhaps the greatest threat to democracy today.
As long as people are running with X11R6 servers, they will want new apps compiled against Y-windows or whatever to run. Does the planned X11R6 compatability layer work both ways?
...) On the other hand, I've seen glitzier programs (emacs, even Mozilla) perform just fine through the same network. But I can see room for improvement. VNC seems to be faster in some ways. And Microsoft has some technology that does a similar thing fairly quickly, but that's relying on a higher-level protocol passing Windows API calls or something. So that's what you mean about integrating the widgets and toolkits somehow.
I think the optional extensions are great -- this allows for multiple implementations of X without an unbearable burden of supporting *everything*. If the process of natural selection is going to work, you want a protocol that's not overly demanding.
By separating the widgets and toolkits from the X server, you again reduce complexity and allow for multiple implementations at these different layers. I guess the widgets and themes could somehow be linked into the X server, but now I'm rapidly entering the realm of speculation.
I keep hearing the argument that X is bad because it's old. Unix is even older, and look what happened with it! I'm not saying you're wrong, I just want to hear about the details. Is it necessarily true that the entire protocol is insufficient and cannot be satisfactorily extended? Why not?
I guess efficiency could be the main issue. In part it's a matter of how the clients are designed. I've seen some relatively simple (just stock widgets) Java programs that, with Sun's VM, absolutely crawl over a 10 MBit connection. I mean that it takes half a minute to redraw a quarter of a window or pop up a menu. Compression seems not to help much because it's all limited by latency. (I can imagine the communique: "Can I draw this pixel here?" "What color?" "Red." "Okay." "You sure?" "pretty sure."
I totally agree here. Basically, the present voting system is such that the two-party system is a stable equilibrium. I'm not entirely sure why this is true, but it's not hard to see some of the reasons and history offers the proof. If Nader got any significant votes (and last time even 1% was significant enough), he would ensure Bush's reelection. So to the party he is more closely aligned with (Democratic) he's enemy #1. The only people who should properly be voting for him are those who hate the two leading candidates absolutely equally. Since that's a small number of people, the two front-running parties remain unchallengable. If I end up disliking both front-runners equally this time around, I'll either go with Nader or "Bufgoo" (the google hit is correct).
I herad in Ireland they can vote for a 2nd and 3rd choice (etc?), and if their 1st choice is loosing the vote goes to the 2nd. I think this upsets the equillibrium and would truly allow the public to represent themselves with their votes. Unfortunately I don't see any hope for a system like this one to be established in the US in my lifetime. Another possible way to upset the 2-party system is if someone new suddenly comes on the scene with unbelievable charisma so that they immediately gain a large fraction of votes. They would also have to be a middle-of-the-road type of candidate. Nader is neither immensely popular nor middle-of-the-road. And anyway this kind of singular event would not really upset the equillibrium, it would simply establish a new party/parties and there would still eventually be two. 3 -> 1 -> 2 I think.
If you need 15 examples in order to solve 15 problems on some general topic, I might suggest that you haven't learned anything.
If solving a problem becomes a monkey-see, monkey-do type of excersize where you've been trained to use specific techniques on certain homework problems, then the student is practicing a technique but not understanding the subject. If the homework problems make the student think a bit and extend those "examples" in new ways, then they might be learning. Hopefully calc students expect that at the end of the course they can solve real-world problems that haven't been solved before, and apply the tools of calculus in ways that they haven't been explicitly taught. If they can't, then the entire course was a waste of time.
A good professor should be able to help any student gain this kind of working understanding of their subject, provided the student is also willing to work as hard as necessary. But there are a lot of professors out there that aren't that good. Since students don't often have much choice in the matter, they might have to look for help elsewhere.
Another problem is that students who have spent more time rehearsing techniques (recipies) and less time actually learning math tend to do better on timed, standardized tests. So to some extent the system punishes good students and teachers.
Thanks, I was searching hard for an on-topic post and am glad to see yours. The question is: How can people make money while writing open-source code?
I think the short answer is that programmers can be (and are being) paid for writing open source that satisfies the immediate and particular needs of the company they work for.
The long answer is to compare two models of software and its place in our economy. One model is the proprietary code one, where there are two types of companies: one which produces software, and one which uses the software. In this model there is a strong distinction also between computer users and programmers.
The other model is one where software is free, but programmers will be hired by companies to customize and extend software to meet their peculiar needs. This model tends to blur the distinction between programmers and users. Computers run programs, and so to use a computer is in some sense to program it.
The second model is world where software tools are more effectively leveraged and more valuable. So libraries become more complete, languages more powerful, and programming becomes easier for everyone. Good documentation is also valued.
The first model encourages building a higher barrier between users and programmers, so that the trade secrets needed to program are a commodity by themselves. That is the transparent mindset of the author of the letter. The attitude is that the knowledge of programming is one's net worth, and that giving it away by writing free code somehow lessens your worth.
I think that the proprietary software model does not work as well in a free market economy, where competition is the driving force behind innovation. In this model, competition necessarily leads to duplication of effort. Also having multiple competing proprietary OSes or software suites will multiply the need for specialized knowledge, and thus divide the value if a programmer's training. For example, if a DB programmer was trained in using Microsoft Access, but not trained in the equally-popular Nanosoft Gain (tm), they would only qualify for half of the current job openings. This explains Microsoft's delusions of benevolence. By dominating the market, they think they are reducing duplicated efforts to compete aginst them, and increasing the value of MS-trained programmers. In the world where software is a commodity, software companies naturally gravitate toward a monopoly.
Competition still thrives in the OS world, but software is not one of the trade commodities therein. Companies use software to help them be competetive with other goods or services.
So for these reasons and many more, companies are beginning to realize the value in the open software model. This model requires a critical mass of free software for it to work well, but I think that in many areas we've reached that critical mass.
Well, at least that's they way I see it. But I'm working in acedamia as a post-doc researcher, and don't have any comp-sci degrees. So I'm not experiencing any of this first-hand. But a lot of what I do is write (and use heavily) open-source code. My job requires that I be a good programmer, but that's not my profession. So naturally I support the OS model since it allows "amateur" programmers like me do my job. But my guesses about OS programming jobs in the industry is just from hearsay, so it's good to see other posts confirming it.
Seriously, this article reminds me of a presentation of magnetotherapy (I think David Letterman aired the clip) where the demonstrator was explaining that "you treat the dog along the diagonal." Absolutely hysterical!
I read a tiny bit of the paper just now, and it seems at least partially legit. But I'm no expert so it's hard for me to tell. Many of these kinds of results are based on 2-sigma confidence levels, which means there's about a 5% chance (rate) of a bogus report. So multiply 5% by the number of studies being done (maybe tens of thousands per year), and you get the rate of BS that we all hear about. Of course we laymen never hear about the experiments with negative results, so we prefferentially tune in to the statistical fluctuations. But this paper does cite a lot of earlier work demonstrating an effect.
Assuming that the article has some legitimacy, I wonder how many hours of shaving corresponds to 48 hours at .01 mT? I would think a shaver is much weaker than that, and scale in the distance factors too. But I'm just guessing.
Another hole in the analysis is that it was done on rat brains, and the study showed a number of ways the effect could be negated by various vitamins and such. Do humans have any of these or other ways of blocking the effects? Unfortunately you can't just buy a few dozen laboratory-grade human beings and run tests on them...
What I finally dug up in 2000 was that Bush thought that basic science should all be privately funded even though there's no short-term motivation for a company to fund a branch of science whos discoveries are public domain. So that statement was ludicrous. Gore's statement was similarly hard to find, and not quite so insane. Something like "funding levels should be on a par with applied science." Now, whether or not they have or would have made good on their campaign platform is mostly beside the point. Since there was NO coverage in the media on these issues, it doesn't matter what the candidates say on page 127, 2nd paragraph, 3rd sentence of some party manifesto. And for all I know they could think that a megabyte is an awfully big sandwich at McD's!
What can we do to encourage more discussion of these issues in the press?
As for tax cuts... I'm more concerned with the wise spending of my taxes than whether or not they go up or down 1%.
I like an idea posted earlier of using a single port to knock on, that looks for a certain key signature in the first packet. This can be sniffed more easily than a port-knocking sequence, but...
If you don't have too many users on a system, you could use a key sequence. Each user would be given a key generator, which generates a near-infinite number of keys in a certain sequence. The host system would be looking for a particular key, and its matching key generator would then advance once it's been used.
This system would have to be used with a single-port type of access because a client requires a simple (trinary) response from the server. The response can be either "used already", "wrong", or "right". The client then tries keys in turn until it receives the "unused" signal, at which point it connects on the desired port. If there were a lot of active users of the system, this process might take some time. If the only server responses were "used already" or "unused" (wrong or right), then a client could get permanently off track if packets were lost because it would not know that it went beyond the sequence.
This system might still have problems with simultaneous logins. And it is "opening up" an additional port. But I can't think of any way that the knocking port alone could be used for a remote exploit. One could also abuse this system if one could selectively block transmissions both ways. I don't know if that's possible or not.