January 21, 2006
Report: Public Examination of Diebold and Hart Systems
Originally posted to the Washington Citizens for Fair Elections mailing list [here http://groups.yahoo.com/group/wa-fairelections/message/565].
Howdy-
Elizabeth Walter and I carpooled down to Olympia for the Public Examination of Diebold and Hart Systems, Friday Jan 20th. Also present were John Gideon (Voters Unite) and James Zukowski (Black Box Voting). Meeting James was kind of bittersweet; having read his posts, I was eager to meet him. Alas, he's moving to Chicago next week. (We'd met John Gideon last year in Portland.)
I had no idea what to expect at this meeting. It was great to finally see the notorious DREs that we've been reading and talking about.
I most respectfully disagree with the esteemed John Gideon. I don't think his efforts were a waste of time. John does a very useful service in pointing out where the state is not complying with the laws and rules. But you can't win every fight. We're seeing the outcomes of decisions made years ago. More on that below. Further, the efforts to mitigate the problems with these systems, specifically VVPAT, were very worthwhile. That addition changed these systems from completely horrible to merely mostly horrible.
Hart/Intercivic gave a pretty slick demo. The eSlate is well-designed and well-built. It's evident that someone with an actual understanding of human factors is responsible for the whole system. After my preface that electronic voting and counting is bad, I asked a bunch of techie questions. Like the capacities for the memory cards, if the images within the eScan can be downloaded, and so forth. I was satisfied with the answers. Though I want to verify the answer on downloading images. (I didn't think to ask about the underlying operating system.)
My inner user interface designer thinks the user interface has some problems. The buttons are in the wrong places. Having multiple ballots on screen at once made for a lot of scrolling with the dial. My inner poll worker compelled me to ask about training, practice machines, if the visual ballots matched the printed ballots. Paul Miller said the state was cooking up some training materials. The smooth sales guy from Hart says they've got all that stuff. (For a price, I'm sure.)
Answering my question, Mason County Auditor Allan Brotche stated that they have an eSlate set up for their citizens to try out before the election. He left before I could ask if me and some of friends could try it out. (Who's with me?)
I also asked about the 2D barcodes. I don't recall the details of the prior controversy, but one of the vendors intended to use a proprietary barcode. I was surprised to (re)learn that in the event of a "manual" recount, where the actual VVPATs would be used, only 1/4th of the votes need be read and counted by a human. The remaining 3/4ths could be counted with a barcode scanner. Hart doesn't offer such a system. They use some standard barcode, so any scanner would work.
Which brings us to the one humorous moment during the meeting. Snohomish County Auditor Bob Terwilliger is a member of the committee. On the lack of barcode scanners, he proudly announced "Unless you have a vendor who supplies one". Paul Miller, unaware of such a system, then said "That would have to be certified by the state." Terwilliger's surprised and chagrinned "Really?" was priceless.
The Diebold demo pretty much wasn't. Whereas the Hart sales guy reveled in demonstrating his grasp of the technical terms and showing every detail, the Diebold presenter just kind of pushed some buttons, waved her arms, and said "there ya go, any questions?"
The AccuVote TSx is utter crap, inside and out. Whereas the eSlate uses a random number printed on a slip for choosing the ballot style, the AccuVote uses a "smartcard" (a credit card with a chip). Note that the smartcards will inevitably walk out with voters. Access to the card swiper is obscured by the bulk of the VVPAT printer. Obviously an afterthought, that printer just hangs there, demanding to be broken. Fortunately, access requires a key and the paper take up canister accepts a tamper seal. There's a ATM style 12-key numberic pad attachment, that just sort of rests on top. The home key, "5", has a raised dot. Which I guess makes it accessible.
John, James, Elizabeth, and I asked questions about various the Diebold and touchscreen scandals. The committee noticeably perked up. I must admit that I enjoyed that part quite a bit. Watching the Diebold people squirm was good television. And two of the committee members were obviously skeptical.
I give Diebold points, though. Initially, I was surprised they dragged one of their programmers along for the dog and pony. Watching him testify showed me why. I once worked with a guy that was amazingly elusive. You'd ask him a clear question, as in "yes or no". He'd talk on and on, not making much sense, offering all sorts of unrelated information, and kind of give both answers until he gauged which one you wanted to hear. By the time he was done, you've forgotten your question. It was amazing to watch him work over the boss. This Diebold programmer was like that. A secret weapon.
The Hursti / Leon County hack was explained away. If the proper procedures had been followed, it wouldn't be possible. James keenly pointed out that hackers don't follow the procedures. I said that even if the machines are 100% perfect, anybody with access can subvert them. That's security 101.
There was some back and forth about voting machines using "interpreted code". I wasn't aware of the controversy. And, honestly, I have no idea what that means. All code is interpreted. People talk about source code, compilers, and machine code. Fine. But all resources and data, including ballot descriptions, text strings, sound files, fonts, images, and so forth are interpreted. Sometimes we even use the term "data driven programming." I wouldn't even know how you'd legally specify what is interpreted versus not.
The Diebold programmer tried to explain away the Hursti hack by saying it was part of the report generator, which while "interpreted code", only has read-only access to the data. I'm pretty sure that's complete crap. I loved Microsoft Access when it came out, as only a geek could. And I know that there's no security "sandbox" which would prevent code from getting full access to the host system. Maybe there's some other safeguards, like file permissions or something, but that's kind of moot. Buttoning down Windows boxes is a notoriously hard thing for system administrators to do. I can't imagine elections officials having the time or knowledge to do so.
I haven't looked into specifics of that hack. I'm really not that interested. I once spent some time looking through source code I've downloaded from blackboxvoting.org. Yup, it just screams "hack me". In fact, the whole system says that. If not the Hursti hack, there's a zillion other potential exploits. For instance, the AccuVote TSx uses Microsoft Windows and can display images. There's many examples of exploits where a bad (malicious) image installs a worm, trojan, horse, or virus. That could happen accidently. Imagine the potential for someone with trusted (physical) access, script kiddie level knowledge of hacking of Windows and Microsoft Access, and malicious intent. You could change the results. Screw up the ballot definitions. Crash machines. Or just fail silently. Pretty much anything.
I thought the Diebold programmer's retort to John about "computer scientists making law" was impolite. Otherwise, I thought he was a pretty nice guy. He answered all my questions one on one. I believe that one of John Gideon's strongest arguments is that these voting machines don't meet the requirement of an error rate of less than one 1 in 500,000. I asked if Diebold gathers that kind of data. He said he didn't know. Probably not. Because when you do that kind of total quality management (TQM) and zero defect kind of engineering, it's an all or nothing effort. In other words, everyone involved would know the answer.
I loved John Gideon's comment afterwards to the effect "You've just spent 5 minutes minimizng a very serious problem while ignoring everything else." Unfortunately, because this information is asymmetrical, unless you know what the Diebold guy knows, you really can't call bullshit, regardless what your gut tells you.
John Gideon pointed out the source code isn't in escrow, as required. I didn't really follow the explanations for why that hasn't happened. I'd really like for that to happen.
Paul Miller pointed out the technical difficulties of verifying the code being used is the same as that certified (held in escrow?). Miller says they've gotten the same hashkey utility the authorities (ITA? NASED? too many acronyms for me) use. That's how online software distribution is usually done. Unfortunately, hashkeys are easy to "forge". And they can't help detect tampering. While I'm uncomfortable with this solution, it's probably sufficient for now. (I've got a different idea, based on using CD-ROMs, that I'll work through and write up later. Though it's moot without new gear. Which I oppose on principle...)
There was also some thrashing on HAVA, ADA, waiting for rules and clarification from the feds, getting changes from the vendors in a timely fashion, and so forth. The committee member from Disabled Accessible Technology (?) was more or less making the "good enough" argument. Having not read all the statutes and so forth, I accept John Gideon's statement that the AccuVote TSx isn't eligible for certification based on the letter of the law. (I'd love to see actual usability tests of these systems by the disabled. It wouldn't be pretty.)
Having seen the AccuVote TSx and the eSlate side by side, I can't imagine why anyone would choose the AccuVote TSx. I intend to ask Dean Logan next time I see him. It probably has something to do with "legacy systems". Because King Co uses Diebold software for ballots and counting and whatever.
As I said to Paul Miller, talking about the need to upgrade the Diebold backend software (GEMS?), it's completely implausible that these systems aren't data compatible. This is exactly the kind of tight spot one finds themselves using proprietary gear from sole source system integrators. Every single industry segment adopting computer IT goes through this same crisis. Eventually, the customers revolt and demand open standards.
I really enjoyed chatting with Paul Miller and Pam Floyd (Paul's boss). Paul Miller appreciated my use of the cliche "ambiguous rules, ruthlessly enforced."
I've met Floyd before and we talked about voter registration databases and such. She's been very open and helpful. I like her. I promised her I'd have a bunch of questions for her after we get past the VBM brouhaha. I hoping Elizabeth, who knows a lot more about the felons and challenges and whatnot, can relate her conversation.
There are some things that I didn't understand about this whole electronic voting machine charade. Floyd and Miller point out that the systems certified comply with the law. That's a pretty good point. (And that she relies on her lawyers to interpret the law. Also a good point.) My inner QA Manager would call the activity I witnessed part of "customer acceptance testing". Having not seen the state's test plan, I imagine the logic and accuracy test only verifies that the parts are working. Nothing more. And the state isn't equipped to do more than that.
Pam Floyd said that legislators have approached her with various questions. Such as the costs of setting up a testing lab. Wow. I told her that there's no equation where you feed in the parameters, turn the crank, and get a QA/test budget. Every product I've worked on had different testing goals, requirements, and burdens. Further, unless the state has access to the source code and development systems, I can't imagine the state being able to independently actually test any product. In fact, I can't imagine the federal agencies (whoever they are) doing it either. (I've seen plenty of fake testing. What my buddy Troy called "monkey motion".)
The answers we get is dependent on the questions we ask. We're just seeing the consequences of questions asked and assumptions made years ago. If we want different answers, we need to ask different answers.
Said another way, we need different laws. The effort to add the VVPAT feature is a great example of that.
A couple things are now obvious to me. The people in charge of our elections are wedged in. There's top down requirements, unfunded mandates, and deadlines. There's upcoming elections, with long lead times. There's shifting public interest and opinion. Sprinkle controversy, high tech paranoia, and allegations of conflict of interest on top. None of the available choices are particularly appealing.
The result is what we see today.
I'm against 100% vote by mail. But given the circumstances, I think we've dodged a bullet with a the widespread deployment of DREs. The immediate task is to work towards ensuring the risks of VBM are mitigated.
Longer-term, we need to devise and offer alternatives.
I've asked a few people about using open source software for elections. I think the reason alternatives like VotePad and AutoMark haven't been considered is because of the proprietary lock-in. Replacing the voting machines isn't enough. An open source alternative would have to be an end-to-end functional replacement of existing systems.
When I asked Paul about using open source system in our state, he explained that we can only use stuff approved by the feds, but that other states permit it. Interesting. Another law we'll have to change.
Another avenue would be to get fed certification. That would be fun in and of itself. Especially if the effort had a loudmouth standing on tables demanding to know why industry funded testing groups are dragging their feet (or whatever). That would make good television.
Cheers, Jason Aaron Osgood / Seattle WA
PS- I'll link to the minutes when I find them.
Howdy-
Elizabeth Walter and I carpooled down to Olympia for the Public Examination of Diebold and Hart Systems, Friday Jan 20th. Also present were John Gideon (Voters Unite) and James Zukowski (Black Box Voting). Meeting James was kind of bittersweet; having read his posts, I was eager to meet him. Alas, he's moving to Chicago next week. (We'd met John Gideon last year in Portland.)
I had no idea what to expect at this meeting. It was great to finally see the notorious DREs that we've been reading and talking about.
I most respectfully disagree with the esteemed John Gideon. I don't think his efforts were a waste of time. John does a very useful service in pointing out where the state is not complying with the laws and rules. But you can't win every fight. We're seeing the outcomes of decisions made years ago. More on that below. Further, the efforts to mitigate the problems with these systems, specifically VVPAT, were very worthwhile. That addition changed these systems from completely horrible to merely mostly horrible.
Hart/Intercivic gave a pretty slick demo. The eSlate is well-designed and well-built. It's evident that someone with an actual understanding of human factors is responsible for the whole system. After my preface that electronic voting and counting is bad, I asked a bunch of techie questions. Like the capacities for the memory cards, if the images within the eScan can be downloaded, and so forth. I was satisfied with the answers. Though I want to verify the answer on downloading images. (I didn't think to ask about the underlying operating system.)
My inner user interface designer thinks the user interface has some problems. The buttons are in the wrong places. Having multiple ballots on screen at once made for a lot of scrolling with the dial. My inner poll worker compelled me to ask about training, practice machines, if the visual ballots matched the printed ballots. Paul Miller said the state was cooking up some training materials. The smooth sales guy from Hart says they've got all that stuff. (For a price, I'm sure.)
Answering my question, Mason County Auditor Allan Brotche stated that they have an eSlate set up for their citizens to try out before the election. He left before I could ask if me and some of friends could try it out. (Who's with me?)
I also asked about the 2D barcodes. I don't recall the details of the prior controversy, but one of the vendors intended to use a proprietary barcode. I was surprised to (re)learn that in the event of a "manual" recount, where the actual VVPATs would be used, only 1/4th of the votes need be read and counted by a human. The remaining 3/4ths could be counted with a barcode scanner. Hart doesn't offer such a system. They use some standard barcode, so any scanner would work.
Which brings us to the one humorous moment during the meeting. Snohomish County Auditor Bob Terwilliger is a member of the committee. On the lack of barcode scanners, he proudly announced "Unless you have a vendor who supplies one". Paul Miller, unaware of such a system, then said "That would have to be certified by the state." Terwilliger's surprised and chagrinned "Really?" was priceless.
The Diebold demo pretty much wasn't. Whereas the Hart sales guy reveled in demonstrating his grasp of the technical terms and showing every detail, the Diebold presenter just kind of pushed some buttons, waved her arms, and said "there ya go, any questions?"
The AccuVote TSx is utter crap, inside and out. Whereas the eSlate uses a random number printed on a slip for choosing the ballot style, the AccuVote uses a "smartcard" (a credit card with a chip). Note that the smartcards will inevitably walk out with voters. Access to the card swiper is obscured by the bulk of the VVPAT printer. Obviously an afterthought, that printer just hangs there, demanding to be broken. Fortunately, access requires a key and the paper take up canister accepts a tamper seal. There's a ATM style 12-key numberic pad attachment, that just sort of rests on top. The home key, "5", has a raised dot. Which I guess makes it accessible.
John, James, Elizabeth, and I asked questions about various the Diebold and touchscreen scandals. The committee noticeably perked up. I must admit that I enjoyed that part quite a bit. Watching the Diebold people squirm was good television. And two of the committee members were obviously skeptical.
I give Diebold points, though. Initially, I was surprised they dragged one of their programmers along for the dog and pony. Watching him testify showed me why. I once worked with a guy that was amazingly elusive. You'd ask him a clear question, as in "yes or no". He'd talk on and on, not making much sense, offering all sorts of unrelated information, and kind of give both answers until he gauged which one you wanted to hear. By the time he was done, you've forgotten your question. It was amazing to watch him work over the boss. This Diebold programmer was like that. A secret weapon.
The Hursti / Leon County hack was explained away. If the proper procedures had been followed, it wouldn't be possible. James keenly pointed out that hackers don't follow the procedures. I said that even if the machines are 100% perfect, anybody with access can subvert them. That's security 101.
There was some back and forth about voting machines using "interpreted code". I wasn't aware of the controversy. And, honestly, I have no idea what that means. All code is interpreted. People talk about source code, compilers, and machine code. Fine. But all resources and data, including ballot descriptions, text strings, sound files, fonts, images, and so forth are interpreted. Sometimes we even use the term "data driven programming." I wouldn't even know how you'd legally specify what is interpreted versus not.
The Diebold programmer tried to explain away the Hursti hack by saying it was part of the report generator, which while "interpreted code", only has read-only access to the data. I'm pretty sure that's complete crap. I loved Microsoft Access when it came out, as only a geek could. And I know that there's no security "sandbox" which would prevent code from getting full access to the host system. Maybe there's some other safeguards, like file permissions or something, but that's kind of moot. Buttoning down Windows boxes is a notoriously hard thing for system administrators to do. I can't imagine elections officials having the time or knowledge to do so.
I haven't looked into specifics of that hack. I'm really not that interested. I once spent some time looking through source code I've downloaded from blackboxvoting.org. Yup, it just screams "hack me". In fact, the whole system says that. If not the Hursti hack, there's a zillion other potential exploits. For instance, the AccuVote TSx uses Microsoft Windows and can display images. There's many examples of exploits where a bad (malicious) image installs a worm, trojan, horse, or virus. That could happen accidently. Imagine the potential for someone with trusted (physical) access, script kiddie level knowledge of hacking of Windows and Microsoft Access, and malicious intent. You could change the results. Screw up the ballot definitions. Crash machines. Or just fail silently. Pretty much anything.
I thought the Diebold programmer's retort to John about "computer scientists making law" was impolite. Otherwise, I thought he was a pretty nice guy. He answered all my questions one on one. I believe that one of John Gideon's strongest arguments is that these voting machines don't meet the requirement of an error rate of less than one 1 in 500,000. I asked if Diebold gathers that kind of data. He said he didn't know. Probably not. Because when you do that kind of total quality management (TQM) and zero defect kind of engineering, it's an all or nothing effort. In other words, everyone involved would know the answer.
I loved John Gideon's comment afterwards to the effect "You've just spent 5 minutes minimizng a very serious problem while ignoring everything else." Unfortunately, because this information is asymmetrical, unless you know what the Diebold guy knows, you really can't call bullshit, regardless what your gut tells you.
John Gideon pointed out the source code isn't in escrow, as required. I didn't really follow the explanations for why that hasn't happened. I'd really like for that to happen.
Paul Miller pointed out the technical difficulties of verifying the code being used is the same as that certified (held in escrow?). Miller says they've gotten the same hashkey utility the authorities (ITA? NASED? too many acronyms for me) use. That's how online software distribution is usually done. Unfortunately, hashkeys are easy to "forge". And they can't help detect tampering. While I'm uncomfortable with this solution, it's probably sufficient for now. (I've got a different idea, based on using CD-ROMs, that I'll work through and write up later. Though it's moot without new gear. Which I oppose on principle...)
There was also some thrashing on HAVA, ADA, waiting for rules and clarification from the feds, getting changes from the vendors in a timely fashion, and so forth. The committee member from Disabled Accessible Technology (?) was more or less making the "good enough" argument. Having not read all the statutes and so forth, I accept John Gideon's statement that the AccuVote TSx isn't eligible for certification based on the letter of the law. (I'd love to see actual usability tests of these systems by the disabled. It wouldn't be pretty.)
Having seen the AccuVote TSx and the eSlate side by side, I can't imagine why anyone would choose the AccuVote TSx. I intend to ask Dean Logan next time I see him. It probably has something to do with "legacy systems". Because King Co uses Diebold software for ballots and counting and whatever.
As I said to Paul Miller, talking about the need to upgrade the Diebold backend software (GEMS?), it's completely implausible that these systems aren't data compatible. This is exactly the kind of tight spot one finds themselves using proprietary gear from sole source system integrators. Every single industry segment adopting computer IT goes through this same crisis. Eventually, the customers revolt and demand open standards.
I really enjoyed chatting with Paul Miller and Pam Floyd (Paul's boss). Paul Miller appreciated my use of the cliche "ambiguous rules, ruthlessly enforced."
I've met Floyd before and we talked about voter registration databases and such. She's been very open and helpful. I like her. I promised her I'd have a bunch of questions for her after we get past the VBM brouhaha. I hoping Elizabeth, who knows a lot more about the felons and challenges and whatnot, can relate her conversation.
There are some things that I didn't understand about this whole electronic voting machine charade. Floyd and Miller point out that the systems certified comply with the law. That's a pretty good point. (And that she relies on her lawyers to interpret the law. Also a good point.) My inner QA Manager would call the activity I witnessed part of "customer acceptance testing". Having not seen the state's test plan, I imagine the logic and accuracy test only verifies that the parts are working. Nothing more. And the state isn't equipped to do more than that.
Pam Floyd said that legislators have approached her with various questions. Such as the costs of setting up a testing lab. Wow. I told her that there's no equation where you feed in the parameters, turn the crank, and get a QA/test budget. Every product I've worked on had different testing goals, requirements, and burdens. Further, unless the state has access to the source code and development systems, I can't imagine the state being able to independently actually test any product. In fact, I can't imagine the federal agencies (whoever they are) doing it either. (I've seen plenty of fake testing. What my buddy Troy called "monkey motion".)
The answers we get is dependent on the questions we ask. We're just seeing the consequences of questions asked and assumptions made years ago. If we want different answers, we need to ask different answers.
Said another way, we need different laws. The effort to add the VVPAT feature is a great example of that.
A couple things are now obvious to me. The people in charge of our elections are wedged in. There's top down requirements, unfunded mandates, and deadlines. There's upcoming elections, with long lead times. There's shifting public interest and opinion. Sprinkle controversy, high tech paranoia, and allegations of conflict of interest on top. None of the available choices are particularly appealing.
The result is what we see today.
I'm against 100% vote by mail. But given the circumstances, I think we've dodged a bullet with a the widespread deployment of DREs. The immediate task is to work towards ensuring the risks of VBM are mitigated.
Longer-term, we need to devise and offer alternatives.
I've asked a few people about using open source software for elections. I think the reason alternatives like VotePad and AutoMark haven't been considered is because of the proprietary lock-in. Replacing the voting machines isn't enough. An open source alternative would have to be an end-to-end functional replacement of existing systems.
When I asked Paul about using open source system in our state, he explained that we can only use stuff approved by the feds, but that other states permit it. Interesting. Another law we'll have to change.
Another avenue would be to get fed certification. That would be fun in and of itself. Especially if the effort had a loudmouth standing on tables demanding to know why industry funded testing groups are dragging their feet (or whatever). That would make good television.
Cheers, Jason Aaron Osgood / Seattle WA
PS- I'll link to the minutes when I find them.
January 19, 2006
Milkrun
I had another pleasant converation with another person who opposed the monorail. She was relating how a new property near her residence didn't have enough parking. You know, because to live in Seattle, couples pretty much need two cars.
I just couldn't help myself. "Yea, it's too bad we don't have mass transit," I quipped.
This person claimed the bus works fine. If people believed that, then why does a couple in Seattle need two cars?
She also said the bus from downtown to Ballard is great, it's never full, and they could just add another bus. I've heard similar stuff from other anti-monorail types. Like Elliot Ave is wide open and the trip (presumably by car) only takes 10 minutes.
Well, yea. Ideally, the Seattle-Ballard route is speedy and convenient. But that's not been my experience. I work in Ballard and used to volunteer downtown about once a week. Despite what the schedules for the 17 and 18 say, that's easily a 30 minute trip. The buses come in 20 or 30 minute intervals. My success rate for catching any particular bus run never measureably improved. So I learned to allow myself at least 45 minutes to get to work. I think the monorail would have reduced that 15 minutes.
Also... The buses were always super full. I have thing about stinky people. And some claustrophobia. Both worsen during the summer. So I'd occasionally hop off and wait for the next bus.
I've never been able to understand why the Metro routes are all milkruns. Each and every stop. Each and every light. No offense, but when there's one or more disabled riders, it's lights out. Just getting people onto the bus, with all the fuss with the fares, is a huge waste of time.
Express routes and more buses aren't the fix. Transportation systems should be set up as hubs and spokes. The busiest routes between any two hubs gets replaced with something faster. Like a monorail. (Hey! What a great idea!)
I just couldn't help myself. "Yea, it's too bad we don't have mass transit," I quipped.
This person claimed the bus works fine. If people believed that, then why does a couple in Seattle need two cars?
She also said the bus from downtown to Ballard is great, it's never full, and they could just add another bus. I've heard similar stuff from other anti-monorail types. Like Elliot Ave is wide open and the trip (presumably by car) only takes 10 minutes.
Well, yea. Ideally, the Seattle-Ballard route is speedy and convenient. But that's not been my experience. I work in Ballard and used to volunteer downtown about once a week. Despite what the schedules for the 17 and 18 say, that's easily a 30 minute trip. The buses come in 20 or 30 minute intervals. My success rate for catching any particular bus run never measureably improved. So I learned to allow myself at least 45 minutes to get to work. I think the monorail would have reduced that 15 minutes.
Also... The buses were always super full. I have thing about stinky people. And some claustrophobia. Both worsen during the summer. So I'd occasionally hop off and wait for the next bus.
I've never been able to understand why the Metro routes are all milkruns. Each and every stop. Each and every light. No offense, but when there's one or more disabled riders, it's lights out. Just getting people onto the bus, with all the fuss with the fares, is a huge waste of time.
Express routes and more buses aren't the fix. Transportation systems should be set up as hubs and spokes. The busiest routes between any two hubs gets replaced with something faster. Like a monorail. (Hey! What a great idea!)
January 18, 2006
Washington Public Campaigns
We saw Wyatt (last name?) from Washington Public Campaigns speak last night at the 42nd Legislative District Dems meeting. He showed the Bill Moyers video The Road To Clean Elections, talked a bit about WPC's efforts, and answered some questions.
I'm totally in favor of public financing of campaigns. I'll get the details wrong, but here goes: You qualify for a race by gathering 1,000 $5 donations, you swear off other fund raising or using your own money, and then the public pays for your campaign.
The results have been very impressive. Public financing opens up the playing field. The results have been much greater female and minority representation, a rollback of a lot of legislation favoring special interests and the wealthy, and increased civic participation.
We definitely need CLEAN Elections for Washington State, at every level.
Find out more:
Washington Public Campaigns
Fair Elections
Fixing Democracy: The Clean Election Movement
I'm totally in favor of public financing of campaigns. I'll get the details wrong, but here goes: You qualify for a race by gathering 1,000 $5 donations, you swear off other fund raising or using your own money, and then the public pays for your campaign.
The results have been very impressive. Public financing opens up the playing field. The results have been much greater female and minority representation, a rollback of a lot of legislation favoring special interests and the wealthy, and increased civic participation.
We definitely need CLEAN Elections for Washington State, at every level.
Find out more:
Washington Public Campaigns
Fair Elections
Fixing Democracy: The Clean Election Movement
Vote by Mail Presentation Tues 1/24 @ 6-9pm
Washington Citizens for Fair Elections invites you to our upcoming
meeting. The topic will be Vote By Mail: How it works, the
mechanics, the benefits, and the risks. It will be very relevant
given the proposal to shift King County and the entire state to vote-
by-mail. There will be a presentation followed by question and
answer discussion.
If you have questions or issues you'd like addressed in the
presentation, please submit them to fairelections@julieg.net.
Washington Citizens for Fair Elections is a new, multi-partisan group
forming to educate people on issues around elections, and to work for
free, fair, transparent, verifiable elections.
Time & Place
Tuesday Jan 24th, 2006 6:00pm - 9:00pm in Room 209
University Heights Center for the Community Center
5031 University Way NE
Seattle, Washington 98105
http://uhcca.org/
Agenda
6:00 Doors Open (Food and drink is potlatch, if you wish to join in.)
6:30 Meet and Greet (Introductions)
7:00 Group Business
7:15 Vote By Mail Presentation
8:00 Q & A
8:30 Wrap up
PS- Our wa-fairelections mailing list is currently hosted on Yahoo Groups here. Everyone is welcome.
meeting. The topic will be Vote By Mail: How it works, the
mechanics, the benefits, and the risks. It will be very relevant
given the proposal to shift King County and the entire state to vote-
by-mail. There will be a presentation followed by question and
answer discussion.
If you have questions or issues you'd like addressed in the
presentation, please submit them to fairelections@julieg.net.
Washington Citizens for Fair Elections is a new, multi-partisan group
forming to educate people on issues around elections, and to work for
free, fair, transparent, verifiable elections.
Time & Place
Tuesday Jan 24th, 2006 6:00pm - 9:00pm in Room 209
University Heights Center for the Community Center
5031 University Way NE
Seattle, Washington 98105
http://uhcca.org/
Agenda
6:00 Doors Open (Food and drink is potlatch, if you wish to join in.)
6:30 Meet and Greet (Introductions)
7:00 Group Business
7:15 Vote By Mail Presentation
8:00 Q & A
8:30 Wrap up
PS- Our wa-fairelections mailing list is currently hosted on Yahoo Groups here. Everyone is welcome.
January 09, 2006
DesignGridLayout 0.1
I released DesignGridLayout version 0.1 today. The announcement on JavaLobby generated a fair amount of traffic. I've even gotten some pretty good feedback. Very pleasing.
Something interesting. At JavaOne 2005 there was a session called Layout Roundup and Futures. It compares the effort (lines of code) required to reproduce JGoodies Skeleton using the various grid-based layout managers. So I did a quick mockup using DesignGridLayout. (I'll add it to the Examples in CVS.) Here's the numbers:
Something interesting. At JavaOne 2005 there was a session called Layout Roundup and Futures. It compares the effort (lines of code) required to reproduce JGoodies Skeleton using the various grid-based layout managers. So I did a quick mockup using DesignGridLayout. (I'll add it to the Examples in CVS.) Here's the numbers:
| Layout Manager | LOC |
| DesignGridLayout | 20 |
| FormLayout | 30 |
| TableLayout | 39 |
| Packer | 51 |
| GridBagLayout | 66 |
| GroupLayout | 78 |
| SpringLayout | 93 |
January 08, 2006
Judging Alito
Date: Sun, Jan 8, 2006
From: Jason Aaron Osgood
To: Senator Patty Murray
Subj: Alito
Senator Murray-
Fresh on the heels of you waving Roberts through the confirmation process, I'm eager to hear your rationalizations for giving Alito a pass, allowing yet another explicitly anti-Constitution radical judge to determine Constitutional issues. As I've written to you before, and I'm sure you'd agree, democracy is overrated, the rule of law is a nuisance, women should stay at home, reproductive rights are a sin against God, and the President is supreme ruler, and so forth.
With the confirmation of Alito, we'll be just that much closer to tryanny and theocratic rule. I can't wait. Thank you for doing your part.
Cheers, Jason Aaron Osgood / Seattle WA
PS- You're not all bad. For instance, thanks for voting against Bush's Folly in Iraq, and yet still be the most effective supporter of our troops.
From: Jason Aaron Osgood
To: Senator Patty Murray
Subj: Alito
Senator Murray-
Fresh on the heels of you waving Roberts through the confirmation process, I'm eager to hear your rationalizations for giving Alito a pass, allowing yet another explicitly anti-Constitution radical judge to determine Constitutional issues. As I've written to you before, and I'm sure you'd agree, democracy is overrated, the rule of law is a nuisance, women should stay at home, reproductive rights are a sin against God, and the President is supreme ruler, and so forth.
With the confirmation of Alito, we'll be just that much closer to tryanny and theocratic rule. I can't wait. Thank you for doing your part.
Cheers, Jason Aaron Osgood / Seattle WA
PS- You're not all bad. For instance, thanks for voting against Bush's Folly in Iraq, and yet still be the most effective supporter of our troops.
January 03, 2006
Strategies for Automated Screenshots
Years ago, seems like another lifetime, I approached automated build nirvana. Twice. Not to be confused with version control nirvana, mind you. The only piece missing was automated, as in scripted, screenshots.
On one project (code named Cherry, appropriately named Pandora), I was told to do something about localization. The solution we came up with (Troy Langley, Lauren Bish did all the work) was pretty cool. I have yet to see its equal. The nightly build would put together the localized resouce bundles. Extract the IDs from the source tree then query the translation database. (There was some special sauce to support the translator's use cases.) This allowed us to separate the engineering and translation work flows. Previously, there'd be a "UI freeze" months before release. With this new system, translators could update their database whenever and the changes would be reflected in the next build. I figure it saved a couple hundred hours.
On my next projects (Color Central, TrapWise), after some layoffs, I had the immense pleasure of playing technical writer. Apart from being too busy, it was a lot of fun. FrameMaker is the total boss DTP solution, even though it hadn't been significantly updated in a decade. I haven't kept up, but I don't think there's another DTP product that can replace FrameMaker's stylesheets and links. (The last version of InDesign I used was 1.5, so things may have changed.) It's pretty much a programming language onto itself. Very cool. Knowing how the serious doc people do this kind of thing, seeing stuff like XMLFO is a huge disappointment.
Inheriting that work, I just treated the docs like any other deliverable. You've got source code, fonts, some resources, etc. that get stuffed into source control. Suck down the "source tree", "compile" the PDFs, move everything to the CD image, etc. Just like normal. Our tech doc stuff got very efficient.
The only piece missing from both of these uber builds was automated screenshots. I wanted to completely eliminate the "UI freeze". Find a doc, UI design, or localization bug? Then fix it now. All the docs, promo materials, and so forth get updated in the next build.
So now I'm playing with automated screenshots for DesignGridLayout. I haven't settled on a strategy for determining gutters and margins (pulled from the LAF); I anticipate changes when (if) people start to weigh in. And creating those screenshots are a major hassle.
There's two methods: use java.awt.Robot.createScreenCapture(...) and overriding JFrame's paint. Neither are completely satisfying. The Robot can't handle non-rectangular areas, so the rounded corners on Mac OS X are a problem. The issue I have with overriding the paint(...) method is I don't want the build script stuff to be dependent on app code, or vice versa.
Does anybody else have automated screenshots? Yea? How'd you handle it?
On one project (code named Cherry, appropriately named Pandora), I was told to do something about localization. The solution we came up with (Troy Langley, Lauren Bish did all the work) was pretty cool. I have yet to see its equal. The nightly build would put together the localized resouce bundles. Extract the IDs from the source tree then query the translation database. (There was some special sauce to support the translator's use cases.) This allowed us to separate the engineering and translation work flows. Previously, there'd be a "UI freeze" months before release. With this new system, translators could update their database whenever and the changes would be reflected in the next build. I figure it saved a couple hundred hours.
On my next projects (Color Central, TrapWise), after some layoffs, I had the immense pleasure of playing technical writer. Apart from being too busy, it was a lot of fun. FrameMaker is the total boss DTP solution, even though it hadn't been significantly updated in a decade. I haven't kept up, but I don't think there's another DTP product that can replace FrameMaker's stylesheets and links. (The last version of InDesign I used was 1.5, so things may have changed.) It's pretty much a programming language onto itself. Very cool. Knowing how the serious doc people do this kind of thing, seeing stuff like XMLFO is a huge disappointment.
Inheriting that work, I just treated the docs like any other deliverable. You've got source code, fonts, some resources, etc. that get stuffed into source control. Suck down the "source tree", "compile" the PDFs, move everything to the CD image, etc. Just like normal. Our tech doc stuff got very efficient.
The only piece missing from both of these uber builds was automated screenshots. I wanted to completely eliminate the "UI freeze". Find a doc, UI design, or localization bug? Then fix it now. All the docs, promo materials, and so forth get updated in the next build.
So now I'm playing with automated screenshots for DesignGridLayout. I haven't settled on a strategy for determining gutters and margins (pulled from the LAF); I anticipate changes when (if) people start to weigh in. And creating those screenshots are a major hassle.
There's two methods: use java.awt.Robot.createScreenCapture(...) and overriding JFrame's paint. Neither are completely satisfying. The Robot can't handle non-rectangular areas, so the rounded corners on Mac OS X are a problem. The issue I have with overriding the paint(...) method is I don't want the build script stuff to be dependent on app code, or vice versa.
Does anybody else have automated screenshots? Yea? How'd you handle it?
January 02, 2006
500 Greatest Albums
We visited with my Uncle Randy over the weekend. Nice visit. Driving around Bellingham and Lynden is always bittersweet for me. Seeing grandpa Osgood's farm subdivided for tasteless California style mansions always galls me.
Randy received "Rolling Stone's 500 Greatest Albums of All Time". I couldn't resist leafing through and kibitzing. Having reviewed this collection, I can briefly summarize its contents thusly:
What a crock of shit.
From memory... Prince's "Purple Rain" ranks higher than "1999". U2's "All That You Can't Leave Behind" is in the top 100, whereas "War" (one of their two worthwhile albums) is something like 273. "Thriller" bests "Off the Wall". Nearly every Beattle's album is higher than any Elvis album. Led Zepplin's albums are buried deep in the list. There's a bunch of demographic expanding crap, to placate the numbed esthetics of the masses. Like Radiohead, Green Day, Weezer, and Lauryn Hill. Please. And we'll just have to agree to disagree about Bob Dylan's contributions to the arts.
I do appreciate that lots of seminal bands and musicians get a nod. Muddy Waters, Chuck Berry, Talking Heads, Bob Marley, Kraftwerk, etc. That's appropriate and necessary.
My brother Greg has always been a fan of Rolling Stone magazine. Ditto highschool classmate Dan G, whose senior year ambition was to be a contributing writer. Me, I never really got it. Best I can tell, the "culturally relevant" parts of the magazine are written and edited by self-important scenesters, poseurs, wannabes, and fanboys. I dunno. Like a 60's / 70's soft-intellect wank circle archetype used by Mondo 2000, Wired, and GQ.
Some of the political and current events articles (over the years) weren't so bad.
Were I going to do the book, it'd be much different. And better, of course. An actual readership vote would have been cool. Something thematic would have been cool. Like "500 albums you should take with you to Mars". Or "100 albums representing early rock through heavy metal to punk". Or "50 rap albums you can't live without". Compiling a list of favorite albums from musicians would be great. Say top 10 from 50 artists. Or ranking albums and songs by popularity (sales, airplay, times covered, etc).
The virtue of my ideas is they're both informative and interesting. Not some banal self-indulgent wankfest.
Why do I have to come up with all the good ideas? I should be getting paid for my efforts. Geesh.
Randy received "Rolling Stone's 500 Greatest Albums of All Time". I couldn't resist leafing through and kibitzing. Having reviewed this collection, I can briefly summarize its contents thusly:
What a crock of shit.
From memory... Prince's "Purple Rain" ranks higher than "1999". U2's "All That You Can't Leave Behind" is in the top 100, whereas "War" (one of their two worthwhile albums) is something like 273. "Thriller" bests "Off the Wall". Nearly every Beattle's album is higher than any Elvis album. Led Zepplin's albums are buried deep in the list. There's a bunch of demographic expanding crap, to placate the numbed esthetics of the masses. Like Radiohead, Green Day, Weezer, and Lauryn Hill. Please. And we'll just have to agree to disagree about Bob Dylan's contributions to the arts.
I do appreciate that lots of seminal bands and musicians get a nod. Muddy Waters, Chuck Berry, Talking Heads, Bob Marley, Kraftwerk, etc. That's appropriate and necessary.
My brother Greg has always been a fan of Rolling Stone magazine. Ditto highschool classmate Dan G, whose senior year ambition was to be a contributing writer. Me, I never really got it. Best I can tell, the "culturally relevant" parts of the magazine are written and edited by self-important scenesters, poseurs, wannabes, and fanboys. I dunno. Like a 60's / 70's soft-intellect wank circle archetype used by Mondo 2000, Wired, and GQ.
Some of the political and current events articles (over the years) weren't so bad.
Were I going to do the book, it'd be much different. And better, of course. An actual readership vote would have been cool. Something thematic would have been cool. Like "500 albums you should take with you to Mars". Or "100 albums representing early rock through heavy metal to punk". Or "50 rap albums you can't live without". Compiling a list of favorite albums from musicians would be great. Say top 10 from 50 artists. Or ranking albums and songs by popularity (sales, airplay, times covered, etc).
The virtue of my ideas is they're both informative and interesting. Not some banal self-indulgent wankfest.
Why do I have to come up with all the good ideas? I should be getting paid for my efforts. Geesh.
