If you tell them you made a mistake they dispose of your card - ideally shredding.
They don't shred them, they are kept, the poll clerk (or the DRO can't remember which) gets a bunch of envelopes for each ballot box, one is for repudiated ballots and there are others for counted, spoiled and unused ballots. That way if there was ever a suggestion of fraud the ballots of all types can be counted and they should match the number that were printed.
You're right as long as each voter is only issued one blank ballot at a time and the machine in the booth is only marking and not counting the ballots there is no need for signalling. The voter can submit or repudiate the ballot on return to the poll clerk.
As for the "ballot box" itself you can use a system like they use for the Toronto municipal elections, the optical scanner is mounted atop a sealed box, the input of the scanner is designed so it can pull the ballot out of a privacy folder directly, the output of the scanner dumps into the box, the only way to get a ballot in there is to feed it through the scanner. The seal is only broken at the end of the day by the DRO with the scrutineers watching. Since the ballots in the box are all valid and already counted all they do at this point is place them in a sealed envelope in case of a judical recount, AFAIK since they started using this there hasn't been a recount.
If you are trying to produce a 99.9999% accurate paper trail of the electronic votes cast, allow voters to validate their entry into the paper trail while not violating voter privacy and not open the system up for more abuse or errors than a paper based ballot it isn't that easy.
The actual concern and it's a legitimate one is "the printer could jam".
It is a problem that can be dealt with but it to be analysed completely. What do you do if a printer jams or otherwise malfunctions?
This is not as simple as it looks at first blush. If you print the ballot, ask the voter to confim it and then formally record the vote how do you handle the case where a voter prints a ballot, repudiates the printout and starts again? Now they have two slips of paper. How do you make sure the correct one becomes the audit trail? Without voilating the voters privacy rights? OK so you have to make the process flow such that the voter must surrender the repudiated printout before they can start again. If you're using an ATM style printer this becomes a huge administrative nightmare for the poll workers as you have to keep track of who has voted successfully, who has repudiated a printout etc. The chances of error or abuse are simply too high.
So so instead use a printer like some ATM passbook printers where the voter must insert the blank stock ballot to be printed on. Each voter gets issued one blank and to get another they must turn over the one they have repudiated. So that takes care of preventing repudiated slips from getting into the ballot box but there is still an opportunity for someone to print a ballot, repudiate it, perhaps by mistake, leave the booth and put the ballot in the box. Now you have this person having a vote not counting in the electronic count that would count in a manual count.
And so on and so on. A printed ballot confirmation is not as simple as it sounds. The only way I can think of doing this right is as follows. You get issued a folder to contain your ballot and keep it private (just like is used with the systems we have today that use optical readers) You go into the voting booth, make your selections on the computer, it prints a ballot, you validate it or repuditate it, this lights either a green or red light on the exterior of the booth. You take your printed ballot, put it in the folder and take it back to the poll clerk, if you have a red light they feed your ballot into a shredder and give you a new blank, if you have a green light they feed your ballot into a reader and only then is the vote recorded. The actual voting machine records nothing, it is just a ballot printer but you get the benefits of computerized voting (multi language, large print, complex ballot handling etc.) without the risk of a machine failure losing legitimate votes.
Still you need to think about how do you implement appropriate controls on the paper stock to prevent its use in ballot stuffing? (there are substantial checks and controls over paper ballots where they are used to prevent ballot forgery).
The problem with any voting system, even the simplest (like we have here, mark a X in the circle next to canditates name you prefer) is that it's possible to mess up (I've scrutineered a few elections here it's amazing how people manage to spoil their ballots, many obviously intentionally as a protest but in some cases "mark an X in a circle" is too difficult to handle), the trick is to keep the error rate as low as possible and to build in opportunities for candidate scrutiny of both voting (without violating individual voter anonymity) and recounting.
Once the Brazilians figure out how lame and useless these social networking things are, their numbers will drop.
You think so? Have a look at fotolog.net, for some reason Brazilians really took to it and in so doing almost destroyed it. Nothing to do with language, a large percentage of them (and there were some very notable exceptions) were teenagers using it for "social networking" and killing it with volume, very few were using it for the site's original intended purpose, to make photo blogs, they were just using the photo comment fields to message each other and a lot of them were cheating to get around the limits on the free accounts.
I don't half wonder if the issue at orkut is similar, it's not the language so much as the sheer volume of no content teenage blather....
I didn't measure the speeds and they did vary wildly, so all I have to go on is a gut feel that it was usably fast.
Keep in mind that it is still a trial, almost ended once but now extended "until further notice" and it is only available on the VIA express trains between Toronto and Montreal (first train in the AM and 5PM) the milk run trains do not have this.
Well I'm another Wi-Fi on the train user, somewhat irregularly but I can say at least on the PointShot trial that runs on VIA rail between Toronto and Montreal there is no need to limit oneself to "small text files and e-mail" the performance varies as you roll but it seemed to be better than dialup at all times.
$10 sounds lower that what they are actually achieving but at $35/bbl market price they are making money. Keep in mind also that the current extraction technology used for turning tar sand into what they call synthetic crude, uses a lot of natural gas, if methane prices keep going up too so will this.
Well actually before it was an Adabas product (Called Adabas-D, it was NOT the flagship Adabas database) it was a Nixdorf product.
Doesn't change the fact that until recently SAP owned the commercial rights and SAP execs in Europe were using SAPDB to openly attack Oracle and IBM and Microsoft saying in effect that databases were a commodity so use our product which is "good enough". A message that disappeared suspisciously about the time these talks were going on.
Oh yea, that will work. Take the two companies in the industry most infected with 'not invented here' disease and try to put them together.
Does explain the SAPDB sale to MySQL a little more rationally though. That was one piece of baggage MS would not have tolerated.
I also suspect that WGIII and Uncle Fester took a hard look at the install base, evaluated their chances of actually converting some of the largest customers, overestimated it by at least double and still realized they'd be buying into supporting a product on competitive operating system platforms and databases for a basically a decade at least. Further noticed that many of these customers have ahem connections that they'd rather not mess with (it's rumoured that Haliburton is or was the largest single instance SAP system in the world, this appeared on a chart at one SAP conference and then disappeared for future appearances of the same presentation).
Alpha would blow the doors off PPC64 if had been developed.
Probably a true statment for PPC chips (G5/970), more debatable for the IBM POWER chips.
It would easily be past 2 GHz today and would also have SMT and/or multiple cores on a single die.
Well DEC certainly wasn't going to get there. Due to KO's shortsightedness in regards to partners, licensing and UNIX, it was already limping out of the gate in terms of volume and market acceptance, to cite only the best known example of this can you imagine where it would be today if KO hadn't told Apple to bugger off and sent them into the arms of IBM and Motorola?
By the time he was turfed they were already running out of money to finish the next generation Alpha FAB (can't recall now if this was the EV5 or EV6 fab) cash hunger led them to the legal settlement with Intel, not that they had much choice at that point, but it was a disaster for Alpha going from being the shining star in DEC's crown to being another one of Intel's ugly stepchildren. By the time Compaq got hold of DEC they were already too late.
SuSe SLES8 has been able to run on those machines for a while now (well actually what SuSe says is p660 but the p660 is just the H80 with faster clock speed chips) See SuSe Linux Enterprise Server 8 for IBM iSeries/pSeries
If you have a web-based client, why are you needing Citrix at all?
Network resources, I've seen apps where the bandwidth to carry the rendered browser screen with Citrix is multiple times less than would be required to carry the http traffic.
OS/400 (or whatever they call it these days)
Well for a few more weeks it's till called OS/400, when the next release (V5R3) ships they will change the name to i5OS.
I wouldn't assume these questions are necessarily answered in the full report.
In any case the math does work, if we make the not unreasonable assumption that for both Windows and Linux IA64 adoption is similar and that it is low enough to be in the noise we get the following.
((4.13+1.02)*.325)+1.17 = 2.84 leaving HP with.23B of the unaccounted for stuff which would be the remaining dribs of VMS and Tandem stuff. I think it's safe to assume that a large chunk of the remaining.71B is IBM iSeries nee AS/400 and that there are some other minor issues like a few remaining dribs of proprietary file and print NOS and so on.
A much more interesting question is does Gartner double count zLinux and Linux and Mainframe and how are they going to account going forward for Linux vs. proprietary UNIX when customers can buy a box that will a) run both and b) decide themselves how to slice up the box's capacity and can change it at any time...
I don't think it's so much Oracle vs. SQL Server as it is a multiplatform DBMS vs. a uniplatform one.
With Oracle and DB2 and several open source options lined up on one side and SQL Server on the other.
Still today most applications are limited by the scaling of a single node database and every time a company with MSSQL realizes they are running out of gas and and facing a choice between an expensive and risky conversion to another platform or an expensive and risky rehost onto Windows Datacenter they don't make the "one platform" mistake again (or one hopes, in reality of course some PHB heavy shops do exactly this over and over again).
I also think that the day most businesses actually really care about which DBMS they use is somewhat past. Databases are becoming more commoditized and more and more people are buying applications (or application suites, or application server platforms) first and then asking what's the "normal" database to run this with.
As long as MS and ORCL continue to make products that are directly competetive with the likes of SAP and Peoplesoft the answer these companies are increasingly going to make is "DB2 or Open Source".
Can you please clarify, "On an IBM eServer in VM" could be any of half a dozen diffent things. Do you actually mean Oracle on Linux on eServer zSeries running as a guest OS under VM? Which Linux? SuSe?
definitely ran on Ultrix later (or maybe by then it was OSF/1)
Keep in mind that Ultrix and OSF/1 (later Digital UNIX) were completely separate UNIX implementations.
OSF/1 was a new UNIX implementation that was released ONLY on Alpha machines (or rather, the Digital version of OSF/1 was only released on Alpha, the IBM and HP versions of OSF/1 were respecively called AIX and HP/UX but I digress).
Ultrix ran on VAXen and on the MIPS chip DECstation/DECserver machines (it also was beta tested but never shipped on PDP-11, I know I was a beta tester).
I still recall that RDB for UNIX was not publically released on Ultrix but I may be wrong, by the time it was starting to hit the streets I wasn't paying attention to databases that didn't run under SAP, including Ingres, Sybase and RDB.
RDB was SQL but it ran only of VMS for a very long time, I don't recall that it ever ran on Ultrix. DEC bundled a runtime license for Ingres (called Ultrix-SQL) with some versions of Ultrix and it was bundled into some of the Ultrix versions of some of the Polycenter products.
RTFA, it says Ingres was the first [non-SQL] relational database, and that SQL was added later. Ingres used a Query language called QUEL.
Now it says that CA added SQL which if I'm remembering isn't true, SQL was in the product well before CA bought it.
Ingres was made by Relational Technology Inc. (at one point in the early eighties there were three database companies that had names containing "relational" and they all eventually changed their names to that of their product (Ingres, Oracle and Informix).
I wrote an application in PC-Ingres in 1986 that used QUEL, I stopped paying attention shortly after that as I went to work for Oracle. Then in 1991 when I left Oracle to go to DEC Ingres was on my radar again as we resold it as "ULTRIX-SQL" and obviously by that point it had gained SQL capabilities. Sometime after that Ingres was in financial trouble and got bought by ASK because they had an application that was based on Ingres and felt they couldn't afford to have them go out of business. Later CA bought ASK.
This language we speak has a couple of interesting attributes, one of them being that the same word used in different contexts can have diffrerent meanings. The other is that usage changes over time and we're expected to adapt. One hears this comment about the "incorrect" use of organic in this context occasionally, and I've yet to hear it from someone who didn't turn out, from further contact to be an utter boob...
Slashdot Mirror
They don't shred them, they are kept, the poll clerk (or the DRO can't remember which) gets a bunch of envelopes for each ballot box, one is for repudiated ballots and there are others for counted, spoiled and unused ballots. That way if there was ever a suggestion of fraud the ballots of all types can be counted and they should match the number that were printed.
You're right as long as each voter is only issued one blank ballot at a time and the machine in the booth is only marking and not counting the ballots there is no need for signalling. The voter can submit or repudiate the ballot on return to the poll clerk.
As for the "ballot box" itself you can use a system like they use for the Toronto municipal elections, the optical scanner is mounted atop a sealed box, the input of the scanner is designed so it can pull the ballot out of a privacy folder directly, the output of the scanner dumps into the box, the only way to get a ballot in there is to feed it through the scanner. The seal is only broken at the end of the day by the DRO with the scrutineers watching. Since the ballots in the box are all valid and already counted all they do at this point is place them in a sealed envelope in case of a judical recount, AFAIK since they started using this there hasn't been a recount.
If you are trying to produce a 99.9999% accurate paper trail of the electronic votes cast, allow voters to validate their entry into the paper trail while not violating voter privacy and not open the system up for more abuse or errors than a paper based ballot it isn't that easy.
The actual concern and it's a legitimate one is "the printer could jam".
It is a problem that can be dealt with but it to be analysed completely. What do you do if a printer jams or otherwise malfunctions?
This is not as simple as it looks at first blush. If you print the ballot, ask the voter to confim it and then formally record the vote how do you handle the case where a voter prints a ballot, repudiates the printout and starts again? Now they have two slips of paper. How do you make sure the correct one becomes the audit trail? Without voilating the voters privacy rights? OK so you have to make the process flow such that the voter must surrender the repudiated printout before they can start again. If you're using an ATM style printer this becomes a huge administrative nightmare for the poll workers as you have to keep track of who has voted successfully, who has repudiated a printout etc. The chances of error or abuse are simply too high.
So so instead use a printer like some ATM passbook printers where the voter must insert the blank stock ballot to be printed on. Each voter gets issued one blank and to get another they must turn over the one they have repudiated. So that takes care of preventing repudiated slips from getting into the ballot box but there is still an opportunity for someone to print a ballot, repudiate it, perhaps by mistake, leave the booth and put the ballot in the box. Now you have this person having a vote not counting in the electronic count that would count in a manual count.
And so on and so on. A printed ballot confirmation is not as simple as it sounds. The only way I can think of doing this right is as follows. You get issued a folder to contain your ballot and keep it private (just like is used with the systems we have today that use optical readers) You go into the voting booth, make your selections on the computer, it prints a ballot, you validate it or repuditate it, this lights either a green or red light on the exterior of the booth. You take your printed ballot, put it in the folder and take it back to the poll clerk, if you have a red light they feed your ballot into a shredder and give you a new blank, if you have a green light they feed your ballot into a reader and only then is the vote recorded. The actual voting machine records nothing, it is just a ballot printer but you get the benefits of computerized voting (multi language, large print, complex ballot handling etc.) without the risk of a machine failure losing legitimate votes.
Still you need to think about how do you implement appropriate controls on the paper stock to prevent its use in ballot stuffing? (there are substantial checks and controls over paper ballots where they are used to prevent ballot forgery).
The problem with any voting system, even the simplest (like we have here, mark a X in the circle next to canditates name you prefer) is that it's possible to mess up (I've scrutineered a few elections here it's amazing how people manage to spoil their ballots, many obviously intentionally as a protest but in some cases "mark an X in a circle" is too difficult to handle), the trick is to keep the error rate as low as possible and to build in opportunities for candidate scrutiny of both voting (without violating individual voter anonymity) and recounting.
Once the Brazilians figure out how lame and useless these social networking things are, their numbers will drop. You think so? Have a look at fotolog.net, for some reason Brazilians really took to it and in so doing almost destroyed it. Nothing to do with language, a large percentage of them (and there were some very notable exceptions) were teenagers using it for "social networking" and killing it with volume, very few were using it for the site's original intended purpose, to make photo blogs, they were just using the photo comment fields to message each other and a lot of them were cheating to get around the limits on the free accounts. I don't half wonder if the issue at orkut is similar, it's not the language so much as the sheer volume of no content teenage blather....
I didn't measure the speeds and they did vary wildly, so all I have to go on is a gut feel that it was usably fast. Keep in mind that it is still a trial, almost ended once but now extended "until further notice" and it is only available on the VIA express trains between Toronto and Montreal (first train in the AM and 5PM) the milk run trains do not have this.
Well I'm another Wi-Fi on the train user, somewhat irregularly but I can say at least on the PointShot trial that runs on VIA rail between Toronto and Montreal there is no need to limit oneself to "small text files and e-mail" the performance varies as you roll but it seemed to be better than dialup at all times.
$10 sounds lower that what they are actually achieving but at $35/bbl market price they are making money. Keep in mind also that the current extraction technology used for turning tar sand into what they call synthetic crude, uses a lot of natural gas, if methane prices keep going up too so will this.
RTFA, this is about selling NEW books at discunt prices. Nothing about used books.
RTFA this decision is about selling NEW books. It does not conflict with Right of First Sale.
It's more science than tecnology perse but a great show and they even have stuff in Ogg.
Doesn't change the fact that until recently SAP owned the commercial rights and SAP execs in Europe were using SAPDB to openly attack Oracle and IBM and Microsoft saying in effect that databases were a commodity so use our product which is "good enough". A message that disappeared suspisciously about the time these talks were going on.
Does explain the SAPDB sale to MySQL a little more rationally though. That was one piece of baggage MS would not have tolerated.
I also suspect that WGIII and Uncle Fester took a hard look at the install base, evaluated their chances of actually converting some of the largest customers, overestimated it by at least double and still realized they'd be buying into supporting a product on competitive operating system platforms and databases for a basically a decade at least. Further noticed that many of these customers have ahem connections that they'd rather not mess with (it's rumoured that Haliburton is or was the largest single instance SAP system in the world, this appeared on a chart at one SAP conference and then disappeared for future appearances of the same presentation).
Probably a true statment for PPC chips (G5/970), more debatable for the IBM POWER chips.
It would easily be past 2 GHz today and would also have SMT and/or multiple cores on a single die.
Well DEC certainly wasn't going to get there. Due to KO's shortsightedness in regards to partners, licensing and UNIX, it was already limping out of the gate in terms of volume and market acceptance, to cite only the best known example of this can you imagine where it would be today if KO hadn't told Apple to bugger off and sent them into the arms of IBM and Motorola?
By the time he was turfed they were already running out of money to finish the next generation Alpha FAB (can't recall now if this was the EV5 or EV6 fab) cash hunger led them to the legal settlement with Intel, not that they had much choice at that point, but it was a disaster for Alpha going from being the shining star in DEC's crown to being another one of Intel's ugly stepchildren. By the time Compaq got hold of DEC they were already too late.
SuSe SLES8 has been able to run on those machines for a while now (well actually what SuSe says is p660 but the p660 is just the H80 with faster clock speed chips) See SuSe Linux Enterprise Server 8 for IBM iSeries/pSeries
Network resources, I've seen apps where the bandwidth to carry the rendered browser screen with Citrix is multiple times less than would be required to carry the http traffic.
OS/400 (or whatever they call it these days) Well for a few more weeks it's till called OS/400, when the next release (V5R3) ships they will change the name to i5OS.
In any case the math does work, if we make the not unreasonable assumption that for both Windows and Linux IA64 adoption is similar and that it is low enough to be in the noise we get the following.
((4.13+1.02)*.325)+1.17 = 2.84 leaving HP with .23B of the unaccounted for stuff which would be the remaining dribs of VMS and Tandem stuff. I think it's safe to assume that a large chunk of the remaining .71B is IBM iSeries nee AS/400 and that there are some other minor issues like a few remaining dribs of proprietary file and print NOS and so on.
A much more interesting question is does Gartner double count zLinux and Linux and Mainframe and how are they going to account going forward for Linux vs. proprietary UNIX when customers can buy a box that will a) run both and b) decide themselves how to slice up the box's capacity and can change it at any time...
With Oracle and DB2 and several open source options lined up on one side and SQL Server on the other.
Still today most applications are limited by the scaling of a single node database and every time a company with MSSQL realizes they are running out of gas and and facing a choice between an expensive and risky conversion to another platform or an expensive and risky rehost onto Windows Datacenter they don't make the "one platform" mistake again (or one hopes, in reality of course some PHB heavy shops do exactly this over and over again).
I also think that the day most businesses actually really care about which DBMS they use is somewhat past. Databases are becoming more commoditized and more and more people are buying applications (or application suites, or application server platforms) first and then asking what's the "normal" database to run this with.
As long as MS and ORCL continue to make products that are directly competetive with the likes of SAP and Peoplesoft the answer these companies are increasingly going to make is "DB2 or Open Source".
Can you please clarify, "On an IBM eServer in VM" could be any of half a dozen diffent things. Do you actually mean Oracle on Linux on eServer zSeries running as a guest OS under VM? Which Linux? SuSe?
Keep in mind that Ultrix and OSF/1 (later Digital UNIX) were completely separate UNIX implementations.
OSF/1 was a new UNIX implementation that was released ONLY on Alpha machines (or rather, the Digital version of OSF/1 was only released on Alpha, the IBM and HP versions of OSF/1 were respecively called AIX and HP/UX but I digress).
Ultrix ran on VAXen and on the MIPS chip DECstation/DECserver machines (it also was beta tested but never shipped on PDP-11, I know I was a beta tester).
I still recall that RDB for UNIX was not publically released on Ultrix but I may be wrong, by the time it was starting to hit the streets I wasn't paying attention to databases that didn't run under SAP, including Ingres, Sybase and RDB.
Relational Technologies Inc. became RTI became Ingres. Later ASK bought Ingres, even later CA bought ASK.
RDB was SQL but it ran only of VMS for a very long time, I don't recall that it ever ran on Ultrix. DEC bundled a runtime license for Ingres (called Ultrix-SQL) with some versions of Ultrix and it was bundled into some of the Ultrix versions of some of the Polycenter products.
Which may be a big part of why MySQL bought SAPDB now MAXDB from SAP.
Now it says that CA added SQL which if I'm remembering isn't true, SQL was in the product well before CA bought it.
Ingres was made by Relational Technology Inc. (at one point in the early eighties there were three database companies that had names containing "relational" and they all eventually changed their names to that of their product (Ingres, Oracle and Informix).
I wrote an application in PC-Ingres in 1986 that used QUEL, I stopped paying attention shortly after that as I went to work for Oracle. Then in 1991 when I left Oracle to go to DEC Ingres was on my radar again as we resold it as "ULTRIX-SQL" and obviously by that point it had gained SQL capabilities. Sometime after that Ingres was in financial trouble and got bought by ASK because they had an application that was based on Ingres and felt they couldn't afford to have them go out of business. Later CA bought ASK.
This language we speak has a couple of interesting attributes, one of them being that the same word used in different contexts can have diffrerent meanings. The other is that usage changes over time and we're expected to adapt. One hears this comment about the "incorrect" use of organic in this context occasionally, and I've yet to hear it from someone who didn't turn out, from further contact to be an utter boob...