Friends: I need to offer a written explanation of the reason this "soft-boot" occurred (see client comments after Ken's). Since this resulted in a voter's vote not being cast and we have no feasible way of knowing which voter this was, my client has requested a letter of explanation. Can I have a "technical layman"'s explanation of this event? Thanks. Les

Additional Comments From Ken Clark 2002-10-28 16:29

RCR#2196 Improve critical error recovery from soft reboot

[There is really no hope of reproducing this. Instead of listing as a bug, I am making a venue change to RCR/enhancement with a different summary.]

The worst part about this is that the voter walked away without having their vote counted. We have to prevent this at all costs.

Lets assume the analysis below is accurate (it sounds pretty plausible). Here are a few concrete improvements to the software that could help:

We should write a ‘token’ to nonvolatile storage (flash) when we accept a voter card. When the voter casts their ballot, we delete the token. At power up, the software checks for the token and knows we crashed (for whatever reason) during voting if one is there. Then we could log the event more accurately. We could also force poll worker intervention when the unit comes back up, so humans are aware of the problem at the point that it happens.

At power up, right now we just unilaterally eject an inserted card without fanfare and without even reading it. This isn’t good enough. We should read the card, log that there was a card in the machine at boot, and log the details of the card. Then we should pop up an error message and require poll worker intervention.

Ideally we would include some audio clues that a poll worker could hear as well, but alas the current internal speaker isn’t up to the task.

If we had something like the above, then at least the poll workers would have known about the problem and could have issued the voter a new card and asked him/her to vote again.

November 5, 2002 Advance Voting-Johnson County, Kansas

Here is the scenario:

A voter inserted their voter card in a touchscreen. Somehow the voter card

was ejected without the cast ballot being pushed...immediately after the

machine audit tape indicates that the machine was rebooted - however, the

machine was never turned off and on. The Supervising Judge (SJ) has the only key

and has instructions to not turn the machines off if there is a reported

problem. They know to just close the lid and call our office. That

location has had several of the problem machines this time and the SJ has

been following those instructions.

Our internal brainstorming leads us to this conclusion. Because of the

voter receipt and the date and time stamp in ESM (voter registration system)

for the computerized voter checking we are able to determine which voter experienced this problem.

It is an elderly voter - we telephoned the voter to do a "voter satisfaction

survey" - they loved the system and placed their voter card in the plastic

container and received their "I Voted" sticker on the way out. This leads

us to believe that the voter's ballot appeared on the screen - the precinct

was validated to the voter receipt by the machine judge and the voter

receipt was placed in the brown envelope for that voting machine. Sometime

during voting, the machine did an internal reboot and ejected the voter card

to the voter. The voter is elderly and simply took the card and placed it

in the container and left. The machine log indicates that the vote was not

cast and that the machine was turned back on. The SJ did not open the

swinging door and turn the machine off and on. I am concerned about this -

can you check with your software folks and get an answer for us. This

appears to be something like a soft reboot happening within the machine

which triggers the voter card to eject just like physically turning the

machine off and on or pushing the Cast Ballot button. As you will recall,

we had several instances in August where the machine ejected the voter card

after the voter voted for just one candidate. We understand that this

problem was also reported by other jurisdictions in their September

primaries.

We need your assistance to problem solve this one.

Thanks..Connie

Need AVTS Soft Boot Explanation

Additional Comments From Ken Clark 2002-10-28 16:29

RCR#2196 Improve critical error recovery from soft reboot