You'll need at least two boxes running in different locations. But you can have multiple dns records pointing at the same boxes. So a surfer looks up your site, tries record 1 on box 1, then record 2 on box 2, then record 3 perhaps on box 1 again. In other words, even if box 1 and 2 are offline temporarilly perhaps the 3rd - 6th attempts will succeed. Now, say you wanted to do some network re-configuration. Having the DNS under your control allows you to set the TTL or time-to-live on the records to very small values. Try a different configuration without worries that some bozo's computer somewhere will have the wrong ip address for your new box for 24 hours till the default record expires. Say you have a redundant off-site backup of all your sites - a short TTL plus DNS under your control will allow you to almost instantly redirect all users to your backup system if needed. Many other benefits as well, plus it is better to keep these things under your control assuming you can handle the burden of setting it up.
They're not random clicks. Presumably you click on icons that you like or that catch your eye. A nice icon will likely lead you to a nice site too. So the votes are for the icons themselves, not the sites that host the icons per se. Thanks for the bookmark, see you again soon hopefully.
Depends upon your browser. In IE 6 most load. Mozilla, Firebird and Opera virtually all load. The icons are actual icons, so some browsers do not support them. You're probably only seeing the icons that happen to be gifs or jpegs.
If you really want to browse the web, try this: The world's best Surf Engine
IconSurf.com,
where you surf the web using website icons.
When you want to explore instead of doing targeted searches
or visiting the same old sites, this is the way to do it.
Shameless plug of my new toy, but it's pretty revolutionary as well.
I'd appreciate any feedback, as the site is just a little over 1 week old.
Thanks for the advice. I started working on the submission form yesterday - it will be available in a day or two as I need to work on some details. I think I will keep the icons at 32 pixels, since 16 is a bit too small for the eyes (but as a visitor you can specify 16 already as an option). Also, many icons do come in 32 bit size and benefit from being larger. As far as serving icons from my site, I believe that is impossible legally. All of these icons are presumably copyrighted, so the most I can do is link to them. Any legal advice about that would be appreciated from other slashdotters.
Tired of searching and only finding what you were looking for? Try my new creation - the world's best Surf Engine, where you surf the web using website icons. When you just want to explore, I believe this is the best and most fun way to do it. I'd appreciate any feedback, as the site is only 1 week old.
I know no-one will see this response because the parent poster will not be modded up, but
your second point deserves some explanation. A Bayesian spam filter would never work if it were not able to overcome the trainer's frequent and unavoidable mistakes. Take a simple and impossible situation as an example: My friend sends me 20 emails I deem to be important, consisting solely of the word "BayesianRules". As an infallible human, I save 19 in the good folder and one in the spam folder. Now I train my filter. Since no-one else used that made up word, it is assigned the probability 19/20=95% as being good. Since my filter is conservative, it only calls an email bad if its' probability of being good is less than 1%. Lo and behold, my misclassified email has a 95% chance of being classified as good by the filter if it only looked at the message content. The filter can easily identify its training flaws. Now consider all of the information in the email header, my friend's signature, the actual content, etc. With one piece of information the filter is great, with many it is nearly infallible. In practice, my filter reverts to a flip-flop -- the content of real and spam mail is so different it is childs play to tell them apart.
When I realized over a year ago that spam was starting to be a huge headache for me, I started saving all my spam and good mail to separate directories, in preparation for using a Bayesian filter. At that time I was getting 20 per day, now I get 350, of which a few make it to my inbox. Anyway, I read Paul Graham's plan for spam and decided to write my own filter, and built in a feature where it would check my classification. Lo and behold, about 5% of the mail I classified was identified by my filter as being incorrectly classified. The filter was correct in almost all cases - I was either misinterpreting the emails or ending up saving them to the wrong directories after correctly categorizing them. Now, whenever someone wants to use my filter, I first require them to classify by hand all their mail for a few weeks. Once they run my program they are amazed - they can't believe they made so many mistakes, and they are instant believers in the power of Bayesian filters. My point is that in implementing these spam reports, the ISPs MUST take human error into account, and only penalize mass-senders if over (roughly) 5% of a given sender's recipients complains.
I use firefox, you insensitive clod. I have as many start pages as I like!
I actually use only two: drpa.us and peter-a-andrews.com/links.html, both of which are my personal pages.
She mentions that the radiation exposure in Kiev during the first few days was equivalent to about a year's worth of radiation at Chernobyl now. The bastards did not inform the populace until the wind blew into Europe and radiation alarms started going off, igniting international alarm. My wife, a child at the time, was belatedly rushed out of town along with all the children in Kiev a week later. I can't prove a link, but the fact is my wife had cancer surgery just last week. I'm sure that coal and gas are worse for the environment, and I support nuclear energy as a cleaner alternative, but a freak accident combined with a stupid reaction of a government made matters much worse than they should have been. People will be suffering due to Chernobyl for decades and centuries to come.
I have been keeping detailed records of all email received for the past 450 days. For the first 350 days, the volume of spam I have received has increased exponentially. For the past 100 days, my spam rate has levelled off at approximately 275 messages per day. I have not cancelled any domain names or changed my internet habits in any way, but my spam quantities are no longer increasing. This is great news, since I will now not need to improve my Bayesian filter as early as I had expected. Currently, about 3-5 spams reach my inbox weekly, with no false positives in the past 6 months.
Angel will be sorely missed. Hopefully another network can pick it up, for at least a few years. Beyond that, Angel will age too much for his immortality to be believable when comparing the first Buffy episodes with the final Angel season.
Or, can anyone say spinoff? The Buffy universe is way too special to end.
In the grid computing analogy, 100 cheap computers working together can produce the equivalent output of one computer 100 times as powerful. The same cannot be said of amateur telescopes working together. For a simplistic example, adding two images both of signal-to-noise 10:1 will result in an image with best signal-to-noise of 20:1.4, which is not twice as good as two single images. Modern large-scale telescopes are so great that all the amateurs in the world would have trouble stitching together a comparable image. Best for amateurs to concentrate in areas the big boys can't match, and not try to win the signal-to-noise (or deep image) game.
My father (and posts above) mentioned a drastic recent improvement in the spam situation as an MSN user, but I have seen no decrease in my considerable (300/day) spam numbers as a multi-domain owner. He claims he specifically requested no filtering to be done on his emails. I therefore conclude that either major spammers are avoiding Microsoft accounts, or that Microsoft is filtering user's emails without permission.
Hey. I get a fair amount of spam, but I am not afraid. It is all filtered. You can see some recent ones at drpa.us/spam.html. Try to send me an email, and check if it gets through! You can also see a plot of my daily spam frequency for the last 400 days or so at drpa.us/spam0.jpg. Advice to all: start saving all your spam and good mail in separate folders. The more you save, the easier it is for a smart filter to automatically identify them. And many thanks to Paul Graham for teaching us all the Bayesian solution (we just need to listen).
Now we need someone equally giving and selfeless as Linus to put GNU-AI into Linux before it's "invented" elsewhere and patented. To the victor goes the spoils. I hope we all win. Just think of the legal battles against AI-backed efforts to keep AI in the hands of the few once its existence is known of.
I've heard it said that the Linux development team is the largest collaborative project in the world or something. Many of these experts should make an open list of algorithms which exhibit behavior likely to be useful for AI, and the open development corps should be invited to build it. Think prior art. We need to ensure we'll have it if someone else discovers it first. But as long as we realize there's a race, with N times the developers and the bottom-up nature of y(our) organizational model (and the mryiad of possibilities this allows), we can beat them to this prize and the ones to follow.
Sorry for the off-topic rant, but I couldn't find the paranoid AI / anti-commercial-exploitation category. Birth of the computer and Linux-open source connection is the best I could do.
I just bought a camera yesterday that outputs quicktime movies but I don't want to download that real player or quicktime player crap. Question: What is the most portable video format and how can I convert my movies to that format? I am looking for the movie equivalents of the.jpg format and linux convert program. This is for the convenience of all the people who will eventually download my movies - I don't want to force them to use a proprietary product. Thanks a bunch. Either answer me here or email me at pa@drpa.us.
When I needed a chat program, I just wrote one that uses a continuous html output. When you view the page the program gives you the latest comments, then hangs until more comments are added by someone else. Then it writes the output over the still open connection and hangs again... The program dies upon prolonged inactivity or if the same ip address requests the page again. When you submit a comment, a new page view gives you what you just entered and anything else typed in by others. I really fear for my webserver by posting this link (due to the way I implemented the chat, it is not good for large numbers of users at once), so it will "disappear" temporarily if trouble ensues:
My chat program.
Let me know if you like it, and I'll pass you the code, written in Perl. It is very minimalistic, but intended to be at least viewable on any browser that can show a page before the whole document is loaded. No download of software, such as the latest java version (and short simple code) is a big plus.
The killer DRM application will be when I can create my resume using professional quality open source tools (like this) and make it impossible for the bastards to convert it to Word.
Re:Tried that one once
on
Superball!
·
· Score: 2, Interesting
I tried it too, except mine was a 5cm diameter steel ball wrapped tightly in 3cm more of rubber bands. Utter demolition of anything and everything in the way of that ball after dropping several stories and ricocheting in a stairwell. If you rolled the ball down a long hallway (quickly) it would (appear to) return to you (after large bang) twice as quickly, bouncing madly.
A nice heavy steel ball also does wonders when rolled slowly on a wavy floor. It appears to have a mind of its own, and drives the dog nuts when it rolls back and forth and changes direction on an apparently level floor.
Here's a real desk worthy of slashdot. I designed it myself and am using it right now.
Go see my computer table for some pictures. Why this strange contraption? It allows me to lie flat on my back with my desk suspended above me. I attach my computer to the rotatable desktop for perfect keyboard and display positioning. It sure helps me keep working while I recover from back problems. Not cool, but somewhat practical.
Slashdot Mirror
See this or this. RMS and many others are all over it.
Did anyone notice that the stock was tanking BEFORE the press release? I guess they put out a negative release to justify the insider-trading fall.
You'll need at least two boxes running in different locations. But you can have multiple dns records pointing at the same boxes. So a surfer looks up your site, tries record 1 on box 1, then record 2 on box 2, then record 3 perhaps on box 1 again. In other words, even if box 1 and 2 are offline temporarilly perhaps the 3rd - 6th attempts will succeed. Now, say you wanted to do some network re-configuration. Having the DNS under your control allows you to set the TTL or time-to-live on the records to very small values. Try a different configuration without worries that some bozo's computer somewhere will have the wrong ip address for your new box for 24 hours till the default record expires. Say you have a redundant off-site backup of all your sites - a short TTL plus DNS under your control will allow you to almost instantly redirect all users to your backup system if needed. Many other benefits as well, plus it is better to keep these things under your control assuming you can handle the burden of setting it up.
They're not random clicks. Presumably you click on icons that you like or that catch your eye. A nice icon will likely lead you to a nice site too. So the votes are for the icons themselves, not the sites that host the icons per se. Thanks for the bookmark, see you again soon hopefully.
Depends upon your browser. In IE 6 most load. Mozilla, Firebird and Opera virtually all load. The icons are actual icons, so some browsers do not support them. You're probably only seeing the icons that happen to be gifs or jpegs.
If you really want to browse the web, try this: The world's best Surf Engine IconSurf.com, where you surf the web using website icons. When you want to explore instead of doing targeted searches or visiting the same old sites, this is the way to do it. Shameless plug of my new toy, but it's pretty revolutionary as well. I'd appreciate any feedback, as the site is just a little over 1 week old.
Thanks for the advice. I started working on the submission form yesterday - it will be available in a day or two as I need to work on some details. I think I will keep the icons at 32 pixels, since 16 is a bit too small for the eyes (but as a visitor you can specify 16 already as an option). Also, many icons do come in 32 bit size and benefit from being larger. As far as serving icons from my site, I believe that is impossible legally. All of these icons are presumably copyrighted, so the most I can do is link to them. Any legal advice about that would be appreciated from other slashdotters.
Tired of searching and only finding what you were looking for? Try my new creation - the world's best Surf Engine, where you surf the web using website icons. When you just want to explore, I believe this is the best and most fun way to do it. I'd appreciate any feedback, as the site is only 1 week old.
I'm waiting for DRM so I can prevent companies from converting my beautiful latex-generated pdf-format resume to a Microsucks word nightmare.
I know no-one will see this response because the parent poster will not be modded up, but your second point deserves some explanation. A Bayesian spam filter would never work if it were not able to overcome the trainer's frequent and unavoidable mistakes. Take a simple and impossible situation as an example: My friend sends me 20 emails I deem to be important, consisting solely of the word "BayesianRules". As an infallible human, I save 19 in the good folder and one in the spam folder. Now I train my filter. Since no-one else used that made up word, it is assigned the probability 19/20=95% as being good. Since my filter is conservative, it only calls an email bad if its' probability of being good is less than 1%. Lo and behold, my misclassified email has a 95% chance of being classified as good by the filter if it only looked at the message content. The filter can easily identify its training flaws. Now consider all of the information in the email header, my friend's signature, the actual content, etc. With one piece of information the filter is great, with many it is nearly infallible. In practice, my filter reverts to a flip-flop -- the content of real and spam mail is so different it is childs play to tell them apart.
When I realized over a year ago that spam was starting to be a huge headache for me, I started saving all my spam and good mail to separate directories, in preparation for using a Bayesian filter. At that time I was getting 20 per day, now I get 350, of which a few make it to my inbox. Anyway, I read Paul Graham's plan for spam and decided to write my own filter, and built in a feature where it would check my classification. Lo and behold, about 5% of the mail I classified was identified by my filter as being incorrectly classified. The filter was correct in almost all cases - I was either misinterpreting the emails or ending up saving them to the wrong directories after correctly categorizing them. Now, whenever someone wants to use my filter, I first require them to classify by hand all their mail for a few weeks. Once they run my program they are amazed - they can't believe they made so many mistakes, and they are instant believers in the power of Bayesian filters. My point is that in implementing these spam reports, the ISPs MUST take human error into account, and only penalize mass-senders if over (roughly) 5% of a given sender's recipients complains.
I use firefox, you insensitive clod. I have as many start pages as I like! I actually use only two: drpa.us and peter-a-andrews.com/links.html, both of which are my personal pages.
She mentions that the radiation exposure in Kiev during the first few days was equivalent to about a year's worth of radiation at Chernobyl now. The bastards did not inform the populace until the wind blew into Europe and radiation alarms started going off, igniting international alarm. My wife, a child at the time, was belatedly rushed out of town along with all the children in Kiev a week later. I can't prove a link, but the fact is my wife had cancer surgery just last week. I'm sure that coal and gas are worse for the environment, and I support nuclear energy as a cleaner alternative, but a freak accident combined with a stupid reaction of a government made matters much worse than they should have been. People will be suffering due to Chernobyl for decades and centuries to come.
I have been keeping detailed records of all email received for the past 450 days. For the first 350 days, the volume of spam I have received has increased exponentially. For the past 100 days, my spam rate has levelled off at approximately 275 messages per day. I have not cancelled any domain names or changed my internet habits in any way, but my spam quantities are no longer increasing. This is great news, since I will now not need to improve my Bayesian filter as early as I had expected. Currently, about 3-5 spams reach my inbox weekly, with no false positives in the past 6 months.
Angel will be sorely missed. Hopefully another network can pick it up, for at least a few years. Beyond that, Angel will age too much for his immortality to be believable when comparing the first Buffy episodes with the final Angel season. Or, can anyone say spinoff? The Buffy universe is way too special to end.
In the grid computing analogy, 100 cheap computers working together can produce the equivalent output of one computer 100 times as powerful. The same cannot be said of amateur telescopes working together. For a simplistic example, adding two images both of signal-to-noise 10:1 will result in an image with best signal-to-noise of 20:1.4, which is not twice as good as two single images. Modern large-scale telescopes are so great that all the amateurs in the world would have trouble stitching together a comparable image. Best for amateurs to concentrate in areas the big boys can't match, and not try to win the signal-to-noise (or deep image) game.
My father (and posts above) mentioned a drastic recent improvement in the spam situation as an MSN user, but I have seen no decrease in my considerable (300/day) spam numbers as a multi-domain owner. He claims he specifically requested no filtering to be done on his emails. I therefore conclude that either major spammers are avoiding Microsoft accounts, or that Microsoft is filtering user's emails without permission.
Hey. I get a fair amount of spam, but I am not afraid. It is all filtered. You can see some recent ones at drpa.us/spam.html. Try to send me an email, and check if it gets through! You can also see a plot of my daily spam frequency for the last 400 days or so at drpa.us/spam0.jpg. Advice to all: start saving all your spam and good mail in separate folders. The more you save, the easier it is for a smart filter to automatically identify them. And many thanks to Paul Graham for teaching us all the Bayesian solution (we just need to listen).
Now we need someone equally giving and selfeless as Linus to put GNU-AI into Linux before it's "invented" elsewhere and patented. To the victor goes the spoils. I hope we all win. Just think of the legal battles against AI-backed efforts to keep AI in the hands of the few once its existence is known of.
I've heard it said that the Linux development team is the largest collaborative project in the world or something. Many of these experts should make an open list of algorithms which exhibit behavior likely to be useful for AI, and the open development corps should be invited to build it. Think prior art. We need to ensure we'll have it if someone else discovers it first. But as long as we realize there's a race, with N times the developers and the bottom-up nature of y(our) organizational model (and the mryiad of possibilities this allows), we can beat them to this prize and the ones to follow.
Sorry for the off-topic rant, but I couldn't find the paranoid AI / anti-commercial-exploitation category. Birth of the computer and Linux-open source connection is the best I could do.
Get a bag from Tumi - they make some awesome, high quality products. Expensive, though. See Tumi website for details.
I just bought a camera yesterday that outputs quicktime movies but I don't want to download that real player or quicktime player crap. Question: What is the most portable video format and how can I convert my movies to that format? I am looking for the movie equivalents of the .jpg format and linux convert program. This is for the convenience of all the people who will eventually download my movies - I don't want to force them to use a proprietary product. Thanks a bunch. Either answer me here or email me at pa@drpa.us.
When I needed a chat program, I just wrote one that uses a continuous html output. When you view the page the program gives you the latest comments, then hangs until more comments are added by someone else. Then it writes the output over the still open connection and hangs again... The program dies upon prolonged inactivity or if the same ip address requests the page again. When you submit a comment, a new page view gives you what you just entered and anything else typed in by others. I really fear for my webserver by posting this link (due to the way I implemented the chat, it is not good for large numbers of users at once), so it will "disappear" temporarily if trouble ensues: My chat program. Let me know if you like it, and I'll pass you the code, written in Perl. It is very minimalistic, but intended to be at least viewable on any browser that can show a page before the whole document is loaded. No download of software, such as the latest java version (and short simple code) is a big plus.
The killer DRM application will be when I can create my resume using professional quality open source tools (like this) and make it impossible for the bastards to convert it to Word.
I tried it too, except mine was a 5cm diameter steel ball wrapped tightly in 3cm more of rubber bands. Utter demolition of anything and everything in the way of that ball after dropping several stories and ricocheting in a stairwell. If you rolled the ball down a long hallway (quickly) it would (appear to) return to you (after large bang) twice as quickly, bouncing madly.
A nice heavy steel ball also does wonders when rolled slowly on a wavy floor. It appears to have a mind of its own, and drives the dog nuts when it rolls back and forth and changes direction on an apparently level floor.
Here's a real desk worthy of slashdot. I designed it myself and am using it right now. Go see my computer table for some pictures. Why this strange contraption? It allows me to lie flat on my back with my desk suspended above me. I attach my computer to the rotatable desktop for perfect keyboard and display positioning. It sure helps me keep working while I recover from back problems. Not cool, but somewhat practical.