In my county they dumped the Diebold machines and went with another vendor using paper ballots which are scanned. They only have voting machines for disabled voters which are made by Sequoia. I didn't check, but believe they also require a paper trail on all machines.
The problem with the machines is you need to look at a lot of the people running the polling places. Machines should not be used unless they're so simple to set up that your grandmother could set it up.
I am also hoping that the democratic candidate wins as secretary of state for California. She has proven to be highly receptive to the Open Voting Consortium, which advocates a fully open-source voting system with a good paper trail. Remember, machines are also used to scan the paper ballots as well. The OVC solution looks pretty good as far as machines go. It prints a human and machine readable ballot and is designed to handle many different voting methods like instant runoff. Not only that, it can run on commodity hardware and so far one voting machine manufacturer has signed on.
By being open and available for experts to study, it will result in a much more transparency.
Paper ballots are not perfect either. Though they are easily humanly readable, they are also easily humanly fallable. I.e. if grandma doesn't fill in the oval completely or too lightly it might not be read, or if they accidently mark where it shouldn't be marked. The only solution for this is to have the ballots scanned at each precinct so the voter can verify their vote.
While I haven't played around much with Krita, I did see that it loads Nikon RAW NEF files correctly. I don't know about other cameras, but I would guess it uses the code from dcraw which handles many different raw formats.
You probably don't even need OCR. I read up on what Ironport does and if you can just detect things like the random dots and the line-type art (i.e. text) in the image one could probably catch a lot of it. It should be a lot less processor intensive as well.
I continue to have very good luck using Baysian filtering. I run DSpam, and even though it's an old version, it is highly effective. It seems to be resistant to poisoning that spammers do using random text since it also looks at how long it's been since a word has been used. If a word hasn't been used in several weeks then it forgets it. This allows it to adapt much more quickly.
I also found that blocking China, Russia, Nigeria and a few other countries also helps, as do RBLs.
Maybe on a bad day 1-2 spams get through, and in all the years I've run it I have only had 3 false positives, far less than I get with Thunderbird.
Part of the problem is a lot of spam filters need to catch up. I have been running DSpam which has been very effective, far more effective than Thunderbird. Unlike most baysian filters I've looked at, it forgets words if they are not used in a spam for a certain amount of time. This helps prevent poisoning. What seems to be more effective for the spammers is their use of image spam, with random dots and whatnot in it, though some filters now look for that (i.e. Ironport).
The problem I have with Thunderbird is I see a high percentage of false positives, even with a lot of training.
-Aaron
Re:Increase the CCD resolution of next telescope
on
The Hubble Lives On
·
· Score: 1
Some are less than that. The infra-red CCD is 64kilopixels, only 256x256 but has taken some gorgeous pictures. The number of pixels isn't everything, but more what you can do with those pixels and how sensitive they are. A coworker has an awesome picture on the wall that he took with the Hubble as a grad student that used about a week of time on the Hubble using the infrared camera to photograph the Orion nebula to see star formation.
Also, more pixels isn't always better. It's the quality of those pixels and the optics. Larger pixels tend to have less noise.
Another benefit of this is that it is possible to run plug-ins at nice CPU level and limit their resources. I.e. one can use ulimit to limit the amount of memory a plug-in can allocate.
The problem I think is that a lot of browser plug-ins won't work with 64-bit support, i.e. Flash. Konqueror solved this problem by making plugins run in a separate process context than the browser, so while the browser is 64-bit, it handles 32-bit binary plugins just fine. It has an added benefit that if a plugin goes berserk it doesn't take out the browser and I can kill the plugin task without affecting the browser.
This is another complaint I have about Firefox. They try and hide a lot of the configuration in about:config, rather than have the configuration options part of the GUI configuration dialog. This includes other things like cache settings and whatnot as well. Now granted, to keep from cluttering up the main configuration they might want to be put into an advanced configuration section.
For those who are interested in seeing a proper voting system put together, check out the Open Voting Consortium. They have a free, open-source voting platform that addresses all of the concerns. It has a verifiable paper trail as well as support for blind users and multiple languages.
I personally have donated money to this organization and believe they are doing the right thing in addressing the current mess we have now.
Their paper trail has a really nice feature in that it also prints a bar code for a quick machine recount of the ballots as well as a human readable output.
-Aaron
Re:For people who complain about GIMP
on
KOffice 1.6 Released
·
· Score: 2, Informative
I've been using Bibble for my photography needs. It handles raw files quite nicely and includes Noise Ninja. I know it's not open source, but it's an excellent package with good workflow support. It has all the tweaks for changing white balance, fixing lens distortion, exposure compensation, curves, and features for removing blemishes.
It's also available for the Mac and Windoze. The professional version, which I use, includes all three. Additionally, since I've bought it I can't count how many free upgrades I've downloaded with useful new features. They also seem fairly responsive to their users needs.
I haven't spent much time with Krita, but it also handles raw files and has 16-bit per color support (48 bit RGB) and multiple color spaces.
As for printing, I've heard there's new printer support coming out (if not already out) that looks better on Linux than Windows, at least with Epson printers (I don't know about other brands).
My logs quickly fill up too. A lot of it comes out of asia, China in particular. There's one IP address that is especially bad. Doing a google search had that subnet turn up in a several year old Department of Homeland Security document. I think a lot of countries either don't care or actively encourage it.
Having dealt with both VxWorks and a commercial embedded Linux I would recommend against VxWorks. My experience with their support is it's almost non-existant and it's missing a ton of functionality and has had a lot of bugs.
For our new project we are using buildroot, which is free. It will automatically download all the various tools and libraries, build the cross compiler and everything else.
If you need help setting this up, I suggest contacting one of the many consultants available to get you up and running. Once you're up and running, just go with a consultant when you're stuck. Our experience with a commercial embedded Linux vendor has been pretty bad with respect to support and I've heard similar complaints about other vendors as well.
The problem has to do with ground. With -48 volts, ground is about the same as chassis ground. If you change it such that -48 is now zero and ground is 48 volts then there could be potential problems (no pun intended) since now your ground and chassis ground differ by 48 volts.
A lot of telco equipment is designed to run on -48 volts DC and PC and server power supplies are readily available at this voltage.
The advantage of -48 over 12 volts is that there will be less loss through resistance and smaller conductors can be used. Of course, there is a greater risk of electric shock, but I would think -48 would be pretty safe.
48 volts is also the standard for Power over Ethernet (IEEE 802.3af). This may not be compatible, though, since telcos run -48, not +48, though some equipment can operate with either (though some cannot).
I'm surprised it doesn't offer component output. Even my old ReplayTV offers progressive component out. Component in would be awesome, as would HDMI. The thing I'm interested in is something that can record HDTV streams as well as the Open Cable (QAM) and have support for a smart card to handle the encryption.
Most cameras can take decent pictures. More megapixels doesn't mean better photographs either since there are so many factors involved. You can take great photos with almost any camera, assuming it has basic focusing control and a usable lens, which most do.
Where it becomes a challenge is when you're in situations with low light, where only the better cameras have low enough noise, or when you need a particular lens.
I have seen many excellent pictures taken by cheap consumer point and shoot cameras. I have also seen crappy photos taken by expensive SLRs. A good camera can't fix taking pictures of bad subject matter or someone who doesn't know how to properly use a camera for the situations they're shooting in.
A lot of people are buying high-MP cameras without fully understanding what they're buying. A 8MP camera won't make a huge difference over a 6MP camera, and a 6MP camera isn't twice as good as a 3MP camera. The differences are usually in things like how sensitive the sensor is and how low noise it is, and how good the lens is.
I see ads in the back of some of my magazines for a Bell and Howell 10MP* camera, but one will find very few reviews. The only thing I could find was that it was only a 3MP camera (* the small text says interpolated) and the resulting images, if it would take them at all, looked like someone was standing in front of a funhouse mirror.
I started off with a 3MP Canon Elph and took a lot of good pictures with it. After a couple years I finally broke down and bought a Nikon D70s, but by then I had totally outgrown the Canon, finding myself limited by the lack of control, the high amount of noise, limited flash capability and the shutter lag, among other issues. I also have spent more on lenses than the camera back. The main reason I went with Nikon over Canon was I already had Nikon film gear and was familiar with it and had some lenses for it. If I didn't already have Nikon I could have just as easily gone with Canon. Both take great pictures.
Some people argue between Canon and Nikon. It doesn't really make a whole lot of difference. Nice specs are one thing, but rather than arguing specs or shooting test charts go out and shoot some real pictures and have fun, whatever camera you use, even that little point-and-shoot.
Here's my output from Top, sorted by virtual memory usage. This system has been up for weeks. The java task is the Eclipse IDE. X has been up for a long time. Look at Thunderbird and Firefox. Also look at the amount of runtime they have received compared to the two copies of Konqueror that are eating up 228 and 209MB. With well over 10 times the usage, they're not much bigger than Firefox. Thunderbird is just as bad. I have to restart it frequently as well.
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 25617 root 16 0 478m 112m 5500 R 6.6 8.9 2736:49 X 25883 aaronw 15 0 227m 120m 21m S 0.3 9.6 271:54.25 konqueror 25871 aaronw 15 0 209m 123m 19m S 0.3 9.8 327:27.31 konqueror 3156 aaronw 15 0 202m 89m 19m S 0.0 7.0 13:45.00 thunderbird-bin 26844 aaronw 15 0 173m 75m 21m S 0.0 5.9 18:43.62 firefox-bin
You must be a Firefox developer, because the developers keep denying that there is a problem, yet many users complain about the memory leak. Just do a Google search. The developers keep coming back that it's a feature. No, a feature is when you limit your caches to what the user specifies, not arbitrarily gobble up all of their available memory. I had my cache set to 32MB, since I don't often go back to previous pages and when I do I have a fast connection. Some people don't have problems, but many others, like myself, do. I guess it depends partially on the browsing habits and the sites you go to. I tend to open and close a lot of tabs, which seems to exasperate the problem in Firefox, and also hit some sites that are heavy with Javascript. Note that this was with Firefox 1.5.0.6. With only some light browsing it's currently hovering at 161MB, and I haven't visited many sites or used it much.
I have 1.25GB of memory in my system. According to top, Firefox was consuming over 800MB and Thunderbird will often chew up over 200MB.
What I meant to say about Konqueror not being bloated then listing all these things is that for web browsing it only loads the KHTML component, whereas for a text file it loads the editor component. Each component itself is generally not that big. I think the memory footprint of KHTML is a lot smaller than Gecko, hence it being popular for use in embedded devices.
Konqueror looks like a swiss army knife, but it's really just a container for various parts, and new parts can be added to it without having to modify any code in Konqueror due to the nature of KDE.
I have been using Konqueror a lot. I got fed up with Firefox when after 24 hours of browsing it was consuming 800MB of RAM, even though the cache size was set at 32MB. I have found that Konqueror is often faster, uses a lot less memory, and is generally more stable. There are a few sites where I have had trouble, but I've also had problems with some sites with Firefox (and a few of those worked with Konqueror). Konqueror has gotten a lot better, especially 3.5. A number of additional fixes went into 3.5.4. My only real complaint is that the adblock feature needs a lot of work to catch up with the Firefox extension.
The file dialog for Konqueror, when I download and save binaries, is infinitely better than the one in Firefox. The UI on Konqueror is also much easier to customize, adding or removing buttons at will. Some of the buttons I find quite useful, like scaling the web page larger or smaller. I also like the fact that plug-ins run as a separate process than the browser and I can run them niced. It also means I can run a 64-bit browser and integrate 32-bit plugins.
I also like the bookmark toolbar better in Konqueror. I can easily add folders or book marks to any folder I want with only a couple clicks.
As a file browser, Konqueror is actually quite nice. It's not the big bloated mess people make it out to be. In fact, if anything is a big bloated mess, it's Firefox. Konqueror uses kparts, so that if, for example, I open a.c file in it, it loads the shared libraries for the editor, or if I click on a multimedia file, it loads kaffeine. Just about everything in KDE is a part, so they can be reused. PDFs are also great in Konqueror when it uses kpdf instead of that bloated Acrobat mess.
Hell, I can't even open more than one instance of Firefox, even on different machines if my home directory is shared over a network. Konqueror has no such problems.
The Konqueror browser I'm typing this from has 18 open tabs and has been open for probably about a week or two. It's consuming 475MB of virtual memory and 116MB of resident memory, but I have had a *lot* more tabs open in the past. I can rarely keep Firefox going for more than 24 hours or so, and it gobbles up memory at an astronomical rate (even 1.5.0.6).
As far as rendering web sites goes, I believe Firefox had problems with Slashdot for the longest time, while Konqueror did not.
One of my father's hard drives could not quite spin up. It was an old 76MB Micropolis MFM drive (110MB with a RLL controller). Anyway, we decided to try putting it in the freezer and cooling it down, since it would spin up if it was cool. This worked and we extracted all of the data off of the drive.
My Micropolis drive ran at least 7 years before I finally retired it, it had two bad sectors new and two bad sectors when I retired it.
Slashdot Mirror
The problem with the machines is you need to look at a lot of the people running the polling places. Machines should not be used unless they're so simple to set up that your grandmother could set it up.
I am also hoping that the democratic candidate wins as secretary of state for California. She has proven to be highly receptive to the Open Voting Consortium, which advocates a fully open-source voting system with a good paper trail. Remember, machines are also used to scan the paper ballots as well. The OVC solution looks pretty good as far as machines go. It prints a human and machine readable ballot and is designed to handle many different voting methods like instant runoff. Not only that, it can run on commodity hardware and so far one voting machine manufacturer has signed on.
By being open and available for experts to study, it will result in a much more transparency.
Paper ballots are not perfect either. Though they are easily humanly readable, they are also easily humanly fallable. I.e. if grandma doesn't fill in the oval completely or too lightly it might not be read, or if they accidently mark where it shouldn't be marked. The only solution for this is to have the ballots scanned at each precinct so the voter can verify their vote.
-Aaron
While I haven't played around much with Krita, I did see that it loads Nikon RAW NEF files correctly. I don't know about other cameras, but I would guess it uses the code from dcraw which handles many different raw formats.
You probably don't even need OCR. I read up on what Ironport does and if you can just detect things like the random dots and the line-type art (i.e. text) in the image one could probably catch a lot of it. It should be a lot less processor intensive as well.
I continue to have very good luck using Baysian filtering. I run DSpam, and even though it's an old version, it is highly effective. It seems to be resistant to poisoning that spammers do using random text since it also looks at how long it's been since a word has been used. If a word hasn't been used in several weeks then it forgets it. This allows it to adapt much more quickly.
I also found that blocking China, Russia, Nigeria and a few other countries also helps, as do RBLs.
Maybe on a bad day 1-2 spams get through, and in all the years I've run it I have only had 3 false positives, far less than I get with Thunderbird.
Part of the problem is a lot of spam filters need to catch up. I have been running DSpam which has been very effective, far more effective than Thunderbird. Unlike most baysian filters I've looked at, it forgets words if they are not used in a spam for a certain amount of time. This helps prevent poisoning. What seems to be more effective for the spammers is their use of image spam, with random dots and whatnot in it, though some filters now look for that (i.e. Ironport).
The problem I have with Thunderbird is I see a high percentage of false positives, even with a lot of training.
-Aaron
Some are less than that. The infra-red CCD is 64kilopixels, only 256x256 but has taken some gorgeous pictures. The number of pixels isn't everything, but more what you can do with those pixels and how sensitive they are. A coworker has an awesome picture on the wall that he took with the Hubble as a grad student that used about a week of time on the Hubble using the infrared camera to photograph the Orion nebula to see star formation.
Also, more pixels isn't always better. It's the quality of those pixels and the optics. Larger pixels tend to have less noise.
-Aaron
Another benefit of this is that it is possible to run plug-ins at nice CPU level and limit their resources. I.e. one can use ulimit to limit the amount of memory a plug-in can allocate.
The problem I think is that a lot of browser plug-ins won't work with 64-bit support, i.e. Flash. Konqueror solved this problem by making plugins run in a separate process context than the browser, so while the browser is 64-bit, it handles 32-bit binary plugins just fine. It has an added benefit that if a plugin goes berserk it doesn't take out the browser and I can kill the plugin task without affecting the browser.
This is another complaint I have about Firefox. They try and hide a lot of the configuration in about:config, rather than have the configuration options part of the GUI configuration dialog. This includes other things like cache settings and whatnot as well. Now granted, to keep from cluttering up the main configuration they might want to be put into an advanced configuration section.
For those who are interested in seeing a proper voting system put together, check out the Open Voting Consortium. They have a free, open-source voting platform that addresses all of the concerns. It has a verifiable paper trail as well as support for blind users and multiple languages.
I personally have donated money to this organization and believe they are doing the right thing in addressing the current mess we have now.
Their paper trail has a really nice feature in that it also prints a bar code for a quick machine recount of the ballots as well as a human readable output.
-Aaron
I've been using Bibble for my photography needs. It handles raw files quite nicely and includes Noise Ninja. I know it's not open source, but it's an excellent package with good workflow support. It has all the tweaks for changing white balance, fixing lens distortion, exposure compensation, curves, and features for removing blemishes.
It's also available for the Mac and Windoze. The professional version, which I use, includes all three. Additionally, since I've bought it I can't count how many free upgrades I've downloaded with useful new features. They also seem fairly responsive to their users needs.
I haven't spent much time with Krita, but it also handles raw files and has 16-bit per color support (48 bit RGB) and multiple color spaces.
As for printing, I've heard there's new printer support coming out (if not already out) that looks better on Linux than Windows, at least with Epson printers (I don't know about other brands).
-Aaron
After reading this article, it sounds like the circumstantial evidence may be fairly strong.
Actually, in this area I don't know that 100K is all that bad, especially considering that houses usually go for 500K+, though a lot of areas of Oakland are considerably cheaper. Oakland is a sore spot in the Bay Area with a high crime rate compared to most other areas. There are many areas of Oakland where I and most people I know will avoid because it just isn't safe. Oakland also has a fairly high homicide rate. See c rime.asp?city=Oakland&state=CA</a> for some statistics. The murder rate is 3.5 times the national average. The city I live in, about 25 miles south is 0.13 times the national average for murder. Berkeley, which is right next store to Oakland is 0.75 times the national average, and Alameda who also borders Oakland has had zero murders, with a population of 74K people. San Jose, a city with about twice the population of Oakland is 0.42 times the national average. Now there are a number of really nice and expensive parts of Oakland, where the murder rate is a lot lower, such as in the Oakland hills, which just indicates how bad that 3.5 times average is. There are areas of Oakland where you just don't want to go.
I hope that she is found alive and that if not that Hans Reiser had nothing to do with it. I also don't have a lot of confidence in the Oakland police department.
My logs quickly fill up too. A lot of it comes out of asia, China in particular. There's one IP address that is especially bad. Doing a google search had that subnet turn up in a several year old Department of Homeland Security document. I think a lot of countries either don't care or actively encourage it.
Having dealt with both VxWorks and a commercial embedded Linux I would recommend against VxWorks. My experience with their support is it's almost non-existant and it's missing a ton of functionality and has had a lot of bugs.
For our new project we are using buildroot, which is free. It will automatically download all the various tools and libraries, build the cross compiler and everything else.
If you need help setting this up, I suggest contacting one of the many consultants available to get you up and running. Once you're up and running, just go with a consultant when you're stuck. Our experience with a commercial embedded Linux vendor has been pretty bad with respect to support and I've heard similar complaints about other vendors as well.
Why is it that nobody seems to benchmark these chips in 64-bit mode? All the benchmarks seem to be 32-bit only.
The problem has to do with ground. With -48 volts, ground is about the same as chassis ground. If you change it such that -48 is now zero and ground is 48 volts then there could be potential problems (no pun intended) since now your ground and chassis ground differ by 48 volts.
A lot of telco equipment is designed to run on -48 volts DC and PC and server power supplies are readily available at this voltage.
The advantage of -48 over 12 volts is that there will be less loss through resistance and smaller conductors can be used. Of course, there is a greater risk of electric shock, but I would think -48 would be pretty safe.
48 volts is also the standard for Power over Ethernet (IEEE 802.3af). This may not be compatible, though, since telcos run -48, not +48, though some equipment can operate with either (though some cannot).
I'm surprised it doesn't offer component output. Even my old ReplayTV offers progressive component out. Component in would be awesome, as would HDMI. The thing I'm interested in is something that can record HDTV streams as well as the Open Cable (QAM) and have support for a smart card to handle the encryption.
Most cameras can take decent pictures. More megapixels doesn't mean better photographs either since there are so many factors involved. You can take great photos with almost any camera, assuming it has basic focusing control and a usable lens, which most do.
Where it becomes a challenge is when you're in situations with low light, where only the better cameras have low enough noise, or when you need a particular lens.
I have seen many excellent pictures taken by cheap consumer point and shoot cameras. I have also seen crappy photos taken by expensive SLRs. A good camera can't fix taking pictures of bad subject matter or someone who doesn't know how to properly use a camera for the situations they're shooting in.
A lot of people are buying high-MP cameras without fully understanding what they're buying. A 8MP camera won't make a huge difference over a 6MP camera, and a 6MP camera isn't twice as good as a 3MP camera. The differences are usually in things like how sensitive the sensor is and how low noise it is, and how good the lens is.
I see ads in the back of some of my magazines for a Bell and Howell 10MP* camera, but one will find very few reviews. The only thing I could find was that it was only a 3MP camera (* the small text says interpolated) and the resulting images, if it would take them at all, looked like someone was standing in front of a funhouse mirror.
I started off with a 3MP Canon Elph and took a lot of good pictures with it. After a couple years I finally broke down and bought a Nikon D70s, but by then I had totally outgrown the Canon, finding myself limited by the lack of control, the high amount of noise, limited flash capability and the shutter lag, among other issues. I also have spent more on lenses than the camera back. The main reason I went with Nikon over Canon was I already had Nikon film gear and was familiar with it and had some lenses for it. If I didn't already have Nikon I could have just as easily gone with Canon. Both take great pictures.
Some people argue between Canon and Nikon. It doesn't really make a whole lot of difference. Nice specs are one thing, but rather than arguing specs or shooting test charts go out and shoot some real pictures and have fun, whatever camera you use, even that little point-and-shoot.
You must be a Firefox developer, because the developers keep denying that there is a problem, yet many users complain about the memory leak. Just do a Google search. The developers keep coming back that it's a feature. No, a feature is when you limit your caches to what the user specifies, not arbitrarily gobble up all of their available memory. I had my cache set to 32MB, since I don't often go back to previous pages and when I do I have a fast connection. Some people don't have problems, but many others, like myself, do. I guess it depends partially on the browsing habits and the sites you go to. I tend to open and close a lot of tabs, which seems to exasperate the problem in Firefox, and also hit some sites that are heavy with Javascript. Note that this was with Firefox 1.5.0.6. With only some light browsing it's currently hovering at 161MB, and I haven't visited many sites or used it much.
I have 1.25GB of memory in my system. According to top, Firefox was consuming over 800MB and Thunderbird will often chew up over 200MB.
What I meant to say about Konqueror not being bloated then listing all these things is that for web browsing it only loads the KHTML component, whereas for a text file it loads the editor component. Each component itself is generally not that big. I think the memory footprint of KHTML is a lot smaller than Gecko, hence it being popular for use in embedded devices.
Konqueror looks like a swiss army knife, but it's really just a container for various parts, and new parts can be added to it without having to modify any code in Konqueror due to the nature of KDE.
I have been using Konqueror a lot. I got fed up with Firefox when after 24 hours of browsing it was consuming 800MB of RAM, even though the cache size was set at 32MB. I have found that Konqueror is often faster, uses a lot less memory, and is generally more stable. There are a few sites where I have had trouble, but I've also had problems with some sites with Firefox (and a few of those worked with Konqueror). Konqueror has gotten a lot better, especially 3.5. A number of additional fixes went into 3.5.4. My only real complaint is that the adblock feature needs a lot of work to catch up with the Firefox extension.
.c file in it, it loads the shared libraries for the editor, or if I click on a multimedia file, it loads kaffeine. Just about everything in KDE is a part, so they can be reused. PDFs are also great in Konqueror when it uses kpdf instead of that bloated Acrobat mess.
The file dialog for Konqueror, when I download and save binaries, is infinitely better than the one in Firefox. The UI on Konqueror is also much easier to customize, adding or removing buttons at will. Some of the buttons I find quite useful, like scaling the web page larger or smaller. I also like the fact that plug-ins run as a separate process than the browser and I can run them niced. It also means I can run a 64-bit browser and integrate 32-bit plugins.
I also like the bookmark toolbar better in Konqueror. I can easily add folders or book marks to any folder I want with only a couple clicks.
As a file browser, Konqueror is actually quite nice. It's not the big bloated mess people make it out to be. In fact, if anything is a big bloated mess, it's Firefox. Konqueror uses kparts, so that if, for example, I open a
Hell, I can't even open more than one instance of Firefox, even on different machines if my home directory is shared over a network. Konqueror has no such problems.
The Konqueror browser I'm typing this from has 18 open tabs and has been open for probably about a week or two. It's consuming 475MB of virtual memory and 116MB of resident memory, but I have had a *lot* more tabs open in the past. I can rarely keep Firefox going for more than 24 hours or so, and it gobbles up memory at an astronomical rate (even 1.5.0.6).
As far as rendering web sites goes, I believe Firefox had problems with Slashdot for the longest time, while Konqueror did not.
One of my father's hard drives could not quite spin up. It was an old 76MB Micropolis MFM drive (110MB with a RLL controller). Anyway, we decided to try putting it in the freezer and cooling it down, since it would spin up if it was cool. This worked and we extracted all of the data off of the drive.
My Micropolis drive ran at least 7 years before I finally retired it, it had two bad sectors new and two bad sectors when I retired it.