Ren Ng gave a talk on this work last April at the University of Washington, and IIRC, he argued that the resolutions of CCDs are increasing exponentially, and after a certain point, the extra resolution is pointless, so why not use that extra resolution to encode additional information not normally captured? I believe he also speculated that the rate of resolution improvements isn't nearly as high as it could be, and technology that could take advantage of the extra resolution would motivate development of extremely high-resolution CCDs.
... but the guitarist just graduated from the University of Washington's Computer Science program. Perhaps this partially explains why they tried this experiment, mentioned their encoding settings on the download page, etc.
They performed (along with The Presidents of the United States of America) at the UW this week as part of a "welcome back" concert (pictures here... ironically enough, it was partially sponsored by Dell and Napster), and it wasn't until they played "Flagpole Sitta" that many people realized who they actually were. Most people claim they don't know them or "Flagpole Sitta," but I'm sure they'd recognize it if they heard it. That's not to say that's their only good song, though -- their entire set rocked.
My understanding of the Danger hiptop is this: To put programs you write yourself on a Danger hiptop, you must become a registered developer, and even then you can not share your programs with other users unless they are also registered developers, or unless Danger gives its official stamp of approval that your application will be THE representative application for its category (calculator, etc.) in their commercial catalog.
This is pretty much the way it is, but it wasn't always that way. You used to be able to freely sign up for a developer key as long as you waived your support rights (although they would provide support in most instances anyway). I'm pretty sure two things led to the demise of that program:
1. Many high profile sites and publications (like Popular Science) gave step-by-step instructions on unlocking your hiptop/Sidekick and where to get third-party apps.
2. An application called Hiptones allowed you to add your own ringtones and circumvent T-Mobile's catalog cash cow. T-Mobile is (or was) the only provider to intentionally disable loading external ringtones via email, so the only other way to get them was to purchase them. The author of Hiptones began selling it, and this made Danger and T-Mobile very unhappy. The author and Danger quickly reached an "agreement" where Hiptones would no longer be sold or available at all, and shortly thereafter, Danger was no longer freely giving out developer keys.
I'm really tempted to blame T-Mobile and the other carriers here. From my experiences with the Danger crew, it seems like they'd really like the hiptop to be as open as possible, but the carriers are insisting that they lock it down to pad their precious pockets.
Unfortunately, many songs purchased from the iTunes Music Store will cause your iPod Shuffle to crash. Illegally stripping the DRM is the only way to get your purchases to work on it. The issue has been known to Apple for months now, and there's still no fix.
What's also particularly scary is that the iTunes Music Store Terms of Service essentially says they can close up shop or change their systems, and you'll be SOL:
d. You acknowledge that some aspects of the Service, Products, and administering of the Usage Rules entails the ongoing involvement of Apple. Accordingly, in the event that Apple changes any part of the Service or discontinues the Service, which Apple may do at its election, you acknowledge that you may no longer be able to use Products to the same extent as prior to such change or discontinuation, and that Apple shall have no liability to you in such case
From what I can tell, if you have an active DSL line provisioned by them, you're good to go. Back when I had my naked DSL installed, I plugged in a borrowed ADSL modem into the test jack in the wiring closet, plugged it into my PDA, and got online. This was a week before my line was officially activated by Covad (who did the internal wiring).
Wow. How non-intuitive is that? Thanks for the tip, though.
Strangely enough, I've never seen Run As in the context menu before today, despite trying to find it. Maybe the shift key triggered some registry change to show it on other items? Maybe I'm just blind.
When I first installed Windows on my new system, I tried creating a seperate non-admin account that I'd use for my day-to-day computing. Shortly thereafter, I added it to the Administrators group because I just couldn't take it anymore.
Installing applications was mostly a non-issue, with Windows prompting me for my Administrator password when I tried to install something that needed Administrator permissions.
However, almost everything else was a giant pain in the ass. If I wanted to use any of the control panels, I either had to log out/log back in as Administrator, use Terminal Services to connect to localhost and log in as Administrator, create yet another shortcut to run it as Administrator, or use the runas command. None of those options are nearly as slick as Windows Installer asking me for my Administrator password. Why they couldn't use the same model is beyond me.
It's not only the control panels that I had problems with. If I wanted to use Windows Update, I had to be Administrator, and it gave me no easy way to become Administrator. If I wanted to develop and debug something in Visual Studio, I either had to be Administrator or be in the debuggers group, which essentially gives you free access to poke at the system any way you like. And of course, numerous applications and games have copy protection systems that require system drivers and services to work.
Of course, LUA doesn't do a damn thing against network-based attacks.
In the end, it's much easier to run as Administrator and drop priviledges when running certain applications.
My former high school had a guideline that you should spend twice as long on homework as you do in the classroom, and virtually every teacher followed that guideline.
Let's do the math.
Every day, you had 348 minutes (5 hours, 48 minutes) of instruction. Lunch, passing time, announcements, etc. push the total time up to 7.5 hours. Using the 2:1 homework:class time ratio, you were expected to spend 11.6 hours a DAY on homework. We're now up to over 19 hours.
Now, factor in transportation time, eating, getting ready in the morning, and of course, the extracurricular activites that they said were needed for colleges to seriously consider you.
You are quite correct. I observed the same thing with the Futuremark forums.
A lot of people on LiveJournal where whining about it, but LiveJournal correctly includes a Cache-Control header, and after several extensive tests, I've found Google's Web Accelerator to not cache anything it shouldn't.
When you first request a page, it sends the request to Google along with some of the request headers (which may contain cookies). Google then sends back a response with a special X-Google-Cache-Control header that instructs the client what to do next. In LiveJournal's case, it sends back X-Google-Cache-Control: remote-fetch, which causes the client to directly fetch the page from LiveJournal. The page contents are not transferred back to Google. Subsequent loads of the page cause only a few bytes to be exchanged with the Web Accelerator server.
Interestingly enough, with a packet sniffer, you can see what it prefetches. When you go to Google.com, it begins fetching hotmail.com, ebay.com, and cnn.com. That says a lot about the typical user.
If you want to link to somewhere without affecting its PageRank, you can use the ref="nofollow" attribute in your link tag. In fact, it looks like slashcode adds it automagically if you don't use your karma bonus. However, since it's automatically added, I'm not sure if it'll pass it through when you post with a karma bonus.
Of course, whether people should deliberately use this is up for debate.
Last week, Dr. Drew Endy from MIT gave a talk to the University of Washington's CSE department on Building Biological Systems (PowerPoint slides are here).
At first glance, building biological systems seems like a pretty daunting task. You have all of these As, Ts, Gs, and Cs, and your task is to figure out how to order them to make your system work as specified. And unlike computers that were engineered by humans, the biological mechanisms that work on DNA aren't completely understood.
However, a promising method of engineering biological systems is to abstract them into systems, devices, and parts. One of the interesting things they're doing is building a repository of biological parts, available at http://parts.mit.edu/. These parts use a standardize way of communicating with each other, allowing you to combine them easily.
Using these parts, college students are able to engineer biological systems in a single quarter. In fact, there's been a few intercollegiate biological engineering competitions, linked to from the MIT Parts site.
Many years ago, my parents took me on a tour of the US Bureau of Engraving and Printing. In the building, they had a gift shop where you could buy all sorts of things. One of the things they sold was a sheet of uncut $2 bills. Apparently, you can now buy them directly from the government over the Internet.
In case you're wondering, they are indeed legal currency. I'd personally like to see someone try to pay with an entire sheet.
From what I understand, in the future, the EPCglobal system (which is being positioned as the replacment for UPC/EAN barcodes) will use DNS as the core technology behind their ONS (Object Name Service) system. So, if a store sees a new EPCglobal RFID tag and wants to know what it is, it can simply execute a DNS query to find out. I'll admit that I haven't looked very closely at the proposed architecture, but this seems like a recipe for disaster when combined with DNS poisoning, especially if automated checkout systems are used that read the EPCglobal tag on every item. I could imagine someone poisoning a store's DNS server, changing product information, such as the price, and causing all sorts of havok.
I'm really not surprised this wasn't a success. A lot of companies blindly go after "emerging markets" without really understanding them. In particular, price isn't as big of a deal as some people think it is. For example, people vastly underestimate the buying power of people in India. Even if everyone was able to afford a computer, what would they do with them? They have no training, no experience, and no support infrastructure.
Interestingly enough, there are some business models that work well. Take the "village PC" model. One person in the village buys a computer (possibly with village assets), supports it, rents out time on it, etc. Everyone in the village, regardless of their technical expertise, benefits from the technology. This model has also worked well for mobile phones.
Last quarter, there were two good talks on technology for emerging and "invisible" markets here at the University of Washington. The first is a talk by Eric Brewer (UC Berkeley) entitled The Case for Technology for Developing Regions. An abstract, video, and MP3 of the talk are available from that site. The other talk was given by John Sherry of Intel's People and Practices Research Group. PowerPoint slides, an abstract, a suggested reading list, a discussion wiki, and more can be found here. I highly encourage you to check these talks out.
I recall reading a post in Jeff Pulver's blog that was a letter from the WUTC on E911/VoIP testing:
Jeff,
I'm sure you are aware that a E911 VoIP trial is ongoing in King County
(greater Seattle area). The King County 911 office along with an ILEC,
Intrado, Vonage, and others have already completed 911 calls that route to
the correct Primary PSAP, carried the correct call back number, and the
correct address information. Dynamic routing (within one hour) of "changed"
address information when a user moves locations, is the last test and is
scheduled for next week (today the information takes a week to be updated
by industry). The method is unusual and still needs to be worked through
the standards organizations but proves that 911 issues for VoIP can be
resolved by cooperation between government and interested companies. If
approved by NENA and ATIS, this method of routing 911 calls will bring
better 911 service to VoIP users in all states.
The initial coordination of this continuing successful project was started
at the VON conference in Boston more than a year ago. Inviting staff from
the Washington State Utility Commission, as well as other state regulatory
commissions, allowed for the type of communication that will build
networks and resolve just these kinds of issues in the future between
industry and the regulatory world.
Hopefully the difficult issues like E-911 cost recovery will be resolved in
the same cooperative manner. Thanks again for making it possible for state
regulators to attend both the Boston and Santa Clara VONs. I will miss the
communication between attendees. I will also miss the excellent parties.
Cheers,
Bob Williamson
Senior Member Technical Staff
Washington Utilities and Transportation Commission
Olympia, WA.
So, it seems like they're working on it. Apparently, Speakeasy's VoIP service already works with 911/E911:
Does Speakeasy Voice Over IP/Home support 911 emergency calling?
Yes, Speakeasy's VOIP offering provides for standard 911 and E-911 emergency dialing.
Our customers have told us that a reliable 911 solution was key to switching from traditional phone lines to VOIP. This mission-critical requirement drove us to design a service that could deliver on this promise.
With Speakeasy Voice Over IP/Home, unlike with most VOIP providers, your 911 calls are routed to the PSAP (Public Safety Answering Point) for your area using standard 911 trunks (not the General Access Line). Your address and telephone number are provided to the 911 operator by way of E-911. In addition, you do not need to register for 911 dialing, as your phone number and address are automatically registered when we complete your order. We have made the design decision to offer voice service only in areas where this level of integration was possible. These features are equivalent to what the local telephone company would provide with the following exceptions: traditional 911 dialing is not possible in the event of a power loss or loss of connectivity. We suggest that you keep a UPS (Uninterruptible Power Supply) connected to your telephone adapter and DSL modem to minimize the effect of power outages on your phone service, and we recommend that you always keep a cell phone as backup for emergencies. Please review our Terms of Service carefully for details about 911 and E-911 dialing.
T-Mobile offers unlimited data and SMS on their Sidekick plan. I'm pretty sure they offer unlimited SMS to encourage people to use it instead of email/IM, which take up more air time/bandwidth.
As an added plus, the Sidekick stores SMS messages on your SIM, so they can't be retrieved should someone discover your password.;)
I'm pretty sure that's a standard warning that was adopted back when everyone connected to mainframes. Still, it might be time to break out the tin-foil hats.
According to some LiveJournal employees, a massive UPS exploded. From IRC:
<rahaeli> As far as we can tell, a UPS exploded.
Their site now says that they're buying their own UPSes, because this is the second time that the entire data center has lost power. Details on the first outage can be found here (a Google cache since LJ is down).
For the paranoid: This has nothing to do Six Apart buying LJ. They're still in the same "world-class" data center they've been in for years.
A Technical Look At Google
on
Defining Google
·
· Score: 5, Insightful
If you're looking for a technical look at Google's inner workings, I highly suggest you view this talk given by one of Google's Distinguished Engineers at the University of Washington. He talks about how Google stores all of its data (the Google File System), and how massive amounts of data are processed (MapReduce), among other things.
1. Some RFID chips (such as EPCglobal Gen2 tags) have a kill password that can be issued to command erasure of the entire PROM, or otherwise totally disable the tag. Gen2 tags have a 32-bit kill password.
2. Virtually all RFID chips can be killed by subjecting them to strong RF radiation. This will induce a lethal voltage and blow out the internal circuitry. Some existing EAS systems use this technique to disable their tags.
About two months ago, Jeff Dean (an employee of Google) gave a talk at the University of Washington about the inner workings of Google. One thing he mentioned was Google Print and how they scan books: they slice 'em up into individual pages, and then feed them through a scanner. This doesn't seem like an acceptable way to archive a library's collection. So, how are they scanning them in? Why not use this method for Google Print?
Slashdot Mirror
Ren Ng gave a talk on this work last April at the University of Washington, and IIRC, he argued that the resolutions of CCDs are increasing exponentially, and after a certain point, the extra resolution is pointless, so why not use that extra resolution to encode additional information not normally captured? I believe he also speculated that the rate of resolution improvements isn't nearly as high as it could be, and technology that could take advantage of the extra resolution would motivate development of extremely high-resolution CCDs.
... but the guitarist just graduated from the University of Washington's Computer Science program. Perhaps this partially explains why they tried this experiment, mentioned their encoding settings on the download page, etc.
They performed (along with The Presidents of the United States of America) at the UW this week as part of a "welcome back" concert (pictures here... ironically enough, it was partially sponsored by Dell and Napster), and it wasn't until they played "Flagpole Sitta" that many people realized who they actually were. Most people claim they don't know them or "Flagpole Sitta," but I'm sure they'd recognize it if they heard it. That's not to say that's their only good song, though -- their entire set rocked.
1. Many high profile sites and publications (like Popular Science) gave step-by-step instructions on unlocking your hiptop/Sidekick and where to get third-party apps.
2. An application called Hiptones allowed you to add your own ringtones and circumvent T-Mobile's catalog cash cow. T-Mobile is (or was) the only provider to intentionally disable loading external ringtones via email, so the only other way to get them was to purchase them. The author of Hiptones began selling it, and this made Danger and T-Mobile very unhappy. The author and Danger quickly reached an "agreement" where Hiptones would no longer be sold or available at all, and shortly thereafter, Danger was no longer freely giving out developer keys.
I'm really tempted to blame T-Mobile and the other carriers here. From my experiences with the Danger crew, it seems like they'd really like the hiptop to be as open as possible, but the carriers are insisting that they lock it down to pad their precious pockets.
What's also particularly scary is that the iTunes Music Store Terms of Service essentially says they can close up shop or change their systems, and you'll be SOL:
From what I can tell, if you have an active DSL line provisioned by them, you're good to go. Back when I had my naked DSL installed, I plugged in a borrowed ADSL modem into the test jack in the wiring closet, plugged it into my PDA, and got online. This was a week before my line was officially activated by Covad (who did the internal wiring).
Wow. How non-intuitive is that? Thanks for the tip, though.
Strangely enough, I've never seen Run As in the context menu before today, despite trying to find it. Maybe the shift key triggered some registry change to show it on other items? Maybe I'm just blind.
When I first installed Windows on my new system, I tried creating a seperate non-admin account that I'd use for my day-to-day computing. Shortly thereafter, I added it to the Administrators group because I just couldn't take it anymore.
Installing applications was mostly a non-issue, with Windows prompting me for my Administrator password when I tried to install something that needed Administrator permissions.
However, almost everything else was a giant pain in the ass. If I wanted to use any of the control panels, I either had to log out/log back in as Administrator, use Terminal Services to connect to localhost and log in as Administrator, create yet another shortcut to run it as Administrator, or use the runas command. None of those options are nearly as slick as Windows Installer asking me for my Administrator password. Why they couldn't use the same model is beyond me.
It's not only the control panels that I had problems with. If I wanted to use Windows Update, I had to be Administrator, and it gave me no easy way to become Administrator. If I wanted to develop and debug something in Visual Studio, I either had to be Administrator or be in the debuggers group, which essentially gives you free access to poke at the system any way you like. And of course, numerous applications and games have copy protection systems that require system drivers and services to work.
Of course, LUA doesn't do a damn thing against network-based attacks.
In the end, it's much easier to run as Administrator and drop priviledges when running certain applications.
My former high school had a guideline that you should spend twice as long on homework as you do in the classroom, and virtually every teacher followed that guideline.
Let's do the math.
Every day, you had 348 minutes (5 hours, 48 minutes) of instruction. Lunch, passing time, announcements, etc. push the total time up to 7.5 hours. Using the 2:1 homework:class time ratio, you were expected to spend 11.6 hours a DAY on homework. We're now up to over 19 hours.
Now, factor in transportation time, eating, getting ready in the morning, and of course, the extracurricular activites that they said were needed for colleges to seriously consider you.
Now, when do you sleep?
You are quite correct. I observed the same thing with the Futuremark forums.
A lot of people on LiveJournal where whining about it, but LiveJournal correctly includes a Cache-Control header, and after several extensive tests, I've found Google's Web Accelerator to not cache anything it shouldn't.
When you first request a page, it sends the request to Google along with some of the request headers (which may contain cookies). Google then sends back a response with a special X-Google-Cache-Control header that instructs the client what to do next. In LiveJournal's case, it sends back X-Google-Cache-Control: remote-fetch, which causes the client to directly fetch the page from LiveJournal. The page contents are not transferred back to Google. Subsequent loads of the page cause only a few bytes to be exchanged with the Web Accelerator server.
Interestingly enough, with a packet sniffer, you can see what it prefetches. When you go to Google.com, it begins fetching hotmail.com, ebay.com, and cnn.com. That says a lot about the typical user.
If you want to link to somewhere without affecting its PageRank, you can use the ref="nofollow" attribute in your link tag. In fact, it looks like slashcode adds it automagically if you don't use your karma bonus. However, since it's automatically added, I'm not sure if it'll pass it through when you post with a karma bonus. Of course, whether people should deliberately use this is up for debate.
Last week, Dr. Drew Endy from MIT gave a talk to the University of Washington's CSE department on Building Biological Systems (PowerPoint slides are here).
At first glance, building biological systems seems like a pretty daunting task. You have all of these As, Ts, Gs, and Cs, and your task is to figure out how to order them to make your system work as specified. And unlike computers that were engineered by humans, the biological mechanisms that work on DNA aren't completely understood.
However, a promising method of engineering biological systems is to abstract them into systems, devices, and parts. One of the interesting things they're doing is building a repository of biological parts, available at http://parts.mit.edu/. These parts use a standardize way of communicating with each other, allowing you to combine them easily.
Using these parts, college students are able to engineer biological systems in a single quarter. In fact, there's been a few intercollegiate biological engineering competitions, linked to from the MIT Parts site.
Many years ago, my parents took me on a tour of the US Bureau of Engraving and Printing. In the building, they had a gift shop where you could buy all sorts of things. One of the things they sold was a sheet of uncut $2 bills. Apparently, you can now buy them directly from the government over the Internet.
In case you're wondering, they are indeed legal currency. I'd personally like to see someone try to pay with an entire sheet.
From what I understand, in the future, the EPCglobal system (which is being positioned as the replacment for UPC/EAN barcodes) will use DNS as the core technology behind their ONS (Object Name Service) system. So, if a store sees a new EPCglobal RFID tag and wants to know what it is, it can simply execute a DNS query to find out. I'll admit that I haven't looked very closely at the proposed architecture, but this seems like a recipe for disaster when combined with DNS poisoning, especially if automated checkout systems are used that read the EPCglobal tag on every item. I could imagine someone poisoning a store's DNS server, changing product information, such as the price, and causing all sorts of havok.
I'm really not surprised this wasn't a success. A lot of companies blindly go after "emerging markets" without really understanding them. In particular, price isn't as big of a deal as some people think it is. For example, people vastly underestimate the buying power of people in India. Even if everyone was able to afford a computer, what would they do with them? They have no training, no experience, and no support infrastructure.
Interestingly enough, there are some business models that work well. Take the "village PC" model. One person in the village buys a computer (possibly with village assets), supports it, rents out time on it, etc. Everyone in the village, regardless of their technical expertise, benefits from the technology. This model has also worked well for mobile phones.
Last quarter, there were two good talks on technology for emerging and "invisible" markets here at the University of Washington. The first is a talk by Eric Brewer (UC Berkeley) entitled The Case for Technology for Developing Regions. An abstract, video, and MP3 of the talk are available from that site. The other talk was given by John Sherry of Intel's People and Practices Research Group. PowerPoint slides, an abstract, a suggested reading list, a discussion wiki, and more can be found here. I highly encourage you to check these talks out.
T-Mobile offers unlimited data and SMS on their Sidekick plan. I'm pretty sure they offer unlimited SMS to encourage people to use it instead of email/IM, which take up more air time/bandwidth. As an added plus, the Sidekick stores SMS messages on your SIM, so they can't be retrieved should someone discover your password. ;)
I'm pretty sure that's a standard warning that was adopted back when everyone connected to mainframes. Still, it might be time to break out the tin-foil hats.
According to some LiveJournal employees, a massive UPS exploded. From IRC:
<rahaeli> As far as we can tell, a UPS exploded.
Their site now says that they're buying their own UPSes, because this is the second time that the entire data center has lost power. Details on the first outage can be found here (a Google cache since LJ is down).
For the paranoid: This has nothing to do Six Apart buying LJ. They're still in the same "world-class" data center they've been in for years.
If you're looking for a technical look at Google's inner workings, I highly suggest you view this talk given by one of Google's Distinguished Engineers at the University of Washington. He talks about how Google stores all of its data (the Google File System), and how massive amounts of data are processed (MapReduce), among other things.
There are two routes to killing an RFID chip:
1. Some RFID chips (such as EPCglobal Gen2 tags) have a kill password that can be issued to command erasure of the entire PROM, or otherwise totally disable the tag. Gen2 tags have a 32-bit kill password.
2. Virtually all RFID chips can be killed by subjecting them to strong RF radiation. This will induce a lethal voltage and blow out the internal circuitry. Some existing EAS systems use this technique to disable their tags.
About two months ago, Jeff Dean (an employee of Google) gave a talk at the University of Washington about the inner workings of Google. One thing he mentioned was Google Print and how they scan books: they slice 'em up into individual pages, and then feed them through a scanner. This doesn't seem like an acceptable way to archive a library's collection. So, how are they scanning them in? Why not use this method for Google Print?
Anything that intentionally degrades the quality of something (be it audio or printed matter) is bad.
This is actually part of a research project going on at the University of Washington Computer Science and Engineering department.
That sure looks like a tax on access to me.