Thursday, February 23, 2006
Beta Bloopers on Google Translate
Sunday, December 04, 2005
Sigh
The Bombay Ballot
-
What the U.S. can learn from India's electronic voting machines.
By Eric Weiner
Posted on Slate on Sept, 29 2004.
Some similarities with my First article on this blog, which was boingboinged and slashdoted in May 2004.
Snip from Eric Weiner's Article on Slate:
"Remember the Cold War tale of Soviet and American scientists racing to solve the problem of writing in zero gravity? NASA spent a decade and millions of dollars developing the high-tech Astronaut Pen. The Soviets solved the problem another way: They used a pencil."
Snip from My Post
"Reading this article, some of you might remember that Cold war era joke, about NASA and its multi million dollar experiment with a pen that can write in micro gravity to solve the writing problems of astronauts, and the Russian solution of using a Pencil to solve the same problem."
Eric's
"The U.S. voting machines are loaded with Windows operating systems, encryption, touch screens, backup servers, voice-guidance systems, modems, PCMCIA storage cards, etc."
Mine
"IMHO, the Diebold system is too complex for a simple and straight forward task such as voting. Windows CE, Modems, PCMCIA storage cards, Touch screen GUI, On-screen writing facility, Voice-guidance system, multiple language UI, DES Encryption, centralized voting Server, a step-by-step wizard to cast a vote, Microsoft SQL Server to store votes, Backup servers etc. are all unnecessary."
Eric's
"A voting system, whether Indian or American, is only as honest as the officials running it."
Mine
"You cannot make a system that is guaranteed as secure. A lot depends on the electoral process and the integrity of election officials."
The Slate article is also salted with racism[1], inaccuracies[2] and exaggerations[3].
[1]
"How can an impoverished nation like India, where cows roam the streets of the capital and most people's idea of high-tech is a flush toilet, succeed where we have not?"
[2]
"For decades, Indians cast their votes by marking a paper ballot with a rubber stamp."
Correction, Oct. 11, 2004:
The article originally claimed that Indians used to vote by pressing an ink-stained thumb onto a paper ballot. In fact, Indians voted by rubber-stamping a paper ballot. They pressed their ink-stained thumbs onto the ballot in order to prevent voter fraud.
[smz: this correction is also inaccurate. :-|]
[3]
"Why do the U.S. machines have so many more bells and whistles than those in India? One reason is that we can. For us, the cost of electronics is largely irrelevant (thank you, Chinese workers). This explains why your DVD player has more features than a 747."
[smz: ???]
Monday, June 07, 2004
X-Prize for innovations in vehicle technologies
In a recent Slashdot post, someone asked a question whether we will see a winner in the DARPA grand challenge in 2005? Well I want to see a winner too. Here is my winner. The best option to cross the DARPA race track is to fit a GPS navigation system, linked with some sort of mechanical device mounted on the back of a horse to pull the reins in right direction. The horse will need only directions, not precise maneuvers to clear obstacles. Yes, at the fuel stops, I will have to keep grass, and water.
Let me put my assumptions here, I think for a machine to drive a vehicle, the Urban scenario is difficult to navigate (traffic lights, moving objects) but easy to drive (tarmac), whereas rough terrain is easy to navigate (long visibility, stationary features) but difficult to drive (all wheels, traction, transmission changes, climbs).
What technologies are required to make an autonomous vehicle? after some thinking, few images flashed from the recent memory. Jeremy clarkson [on BBC] demonstrating the Mercedes' Adaptive Cruise Control on a freeway, Jeremy clarkson again, demonstrating the Terrain response of a Land Rover automatically climbing down a hill without driver's input, and a printed advertisement of a BMW travel motorcycle with cruise control. The immediate thought was, If all these technologies are available in the commercial vehicles today, then why the re-invention of wheel? But before any conclusion some research was called for. And here are the answers found.
The purpose of DARPA grand challenge is "to leverage American ingenuity to accelerate the development of autonomous vehicle technologies that can be applied to military requirements."
The last two words make all the difference. It requires the entrants to disclose their technology to the military and public, and the criteria of the race is to cross a stretch of rough terrain (very much military requirement), The output of this challenge will aid some parts of transportation industry, but not the most important, road vehicles that everyone drives.
EU is also sponsoring projects in the direction of smart vehicles and related technologies since last few years. The approach EU has taken is what impressed me to write about. There are several projects, each related to a particular function of a smart vehicle, all arranged under the 4 year EU research programs called Frameworks, Ultimately leading to (but not aimed towards) autonomous vehicles. Whereas DARPA wants to do it in the basement research mode. There is no argument about which approach is good and which is not. But it seems the EU projects, and the industry in general is producing promising results.
EU projects have reached advanced level in the area of dense traffic navigation, and traffic lights recognition. Pedestrian detection, and lane departure warning system are getting built into the production vehicles all over the world. Towing of vehicles on a freeway, is already tested for trucks.
If we look at the off road scenario, manufacturers are not behind, the Land Rover's Terrain response system is mentioned below. And if we look at the brave attempt of Berkeley team to make a motorcycle run without a rider, well BMW is selling cruise control as an option in their travel motorcycles since last few years. Everything is available with the vehicle manufacturers, but it is not put together for the lack of a commercial application for autonomous vehicles. I recently read an article (can't remember where) giving a super rant (but very much understandable) about the towing concept for trucks on the freeway, the writer was frightened at the idea of driving his little car next to a 18 wheeler without a driver.
I guess, If some X-Prize kind of competition comes up for the purpose, then auto manufacturers will rush to beat each other in innovation. Today no competition exists for the innovation in vehicle technologies. Formula 1, Nascar etc. are tightly controlled with rules to make it competitive, knowingly restricting the use of new innovations. Auto shows attract people who are interested in looking at new models of the vehicles, Technological innovation can't be demonstrated on exhibition floors. Technologies like Toyota Prius' automatic parking, gets talked about for a few days and then becomes vapor. A private competition, will be good for everybody.
As long as the question of military requirements goes, the knowledge gained from the commercial technologies, can be topped with military specifications to make the ultimate rough terrain weapon. Unless there is a step by step approach to make the autonomous vehicle we are going to see little results. We may see a winner in 2005 DARPA grand challenge, but to bring the basement research to military specifications will be quite some task (not impossible!). Industry technologies can be directly introduced in military or can easily be modified to military specs.
Follow these links to read more about EU projects on Urban traffic scenario, or read the description of each link on my surfing journal.
Information
Society Technologies
European research activities are
structured around consecutive four-year programs,
or so-called Framework Programs. The Sixth
Framework Program (FP6) sets out the
priorities - including the Information Society Technologies (IST)
priority - for the period 2002-2006.
Link to Transportation
Articles and Project documents from the same site.
Computer
Vision page at a DC engineer's site
Overview about the EU projects
for Intelligent Transport.
Looking
at people
Gallery of pictures, and Links about how computer
vision looks at people in 3D, and gathers data about their motion.
DaimlerChrysler's Vision: Accident-Free Driving
If we think of it, the urban scenario requires almost real time navigation, in some cases like pedestrian detection and traffic light detection, the navigation system has to think of maneuvers a few meters ahead of the present position of the vehicle. Here are links to videos Demonstrating Traffic Sign Recognition, Pedestrian Detection, and an overview of the said technologies.
Today's top end vehicles have the Adaptive Cruise Control (ACC) system, now other research projects have improved over the first generation ACC to include the dense traffic navigation.
Second
Generation adaptive cruise control for DenseTraffic
Link to
project densetraffic.com
Off road vehicle manufacturers have put path-breaking technologies in their production vehicles. Jeremy Clarkson [on BBC] demonstrated the ability of the Range Rover [Terrain Response mode] to climb down a hill automatically at the twist of a dial. The site www.landrover.com describes it this way.
“...Just turn the Terrain Response dial to the appropriate icon to put all of our driving expertise at your fingertips. It works whether you're cruising on tarmac, driving over deep ruts, grass, snow mud or sand. When you turn the dial, Discovery 3's engine, transmission, suspension, and traction settings are configured to deliver the best driving experience...”
Discovery
3 demo [Flash]
Click "Driver", then Navigate the
"Four Hotspots" to find "Dial an offroad expert"
I couldn't find Information on BMW's cruise control for motorcycles.
Thursday, May 27, 2004
Many ways to Internet
This post is done with a GPRS modem of my Nokia 6510. This completes my usage of all the modes of Internet access available in India. I used the Plain Old Dial-up access, Leased Circuits of various speeds (at work ofcourse!), Cable (Crap), CDMA modems at 115kbps, and now GPRS at slow but cheapest rate. I describe my experience here.
Dial-up
Government Telecom companies are the most honest
in this category, they haven't reached the bandwidth oversell
threshold, since they have big-fat gateways to the hub ISPs
of the Internet. They don't use the Innovative billing rules that are
often used (mostly without informing the users) by Private ISPs. The
general quality of service and reliability is good. Support is not
very good, but for dial-up services the support is not much required,
and good reliability of the service makes it not an important factor.
Private ISPs on the other hand, are mostly into Billing fraud, you
buy a 100 Hours package, and they charge you in innovative ways, -15
minutes for every connection, -2 minutes for every MB of download, -n
minutes per hour if you are in a B-Class town (!!!), -n minutes if
you dare use it during the peak hours, etc.. Speed is okay,
most of the ISPs entered the business when Indian government started
distributing licenses at the fees of 1 Rupee
(2 cents US). These ISPs use bandwidth from Government Gateways, some
do have their own gateways, but they use
the submarine cables of government telecom companies. Those who use
satellite links are the lousiest, you know the delay in satellite
communication. Support is excellent, you get to a phone monkey very
quickly, but they read a scripted dialog. No problem gets solved
unless a number of people report the same problem, a lone caller with
a problem is always shooed away. Classic BOFH material. Among the
Dial-up ISPs, Government Owned VSNL which is now privatized
and called Tata Indicom is the best and most professional (for ISDN
too). In Private ISPs, Satyam is worst.
Leased circuits
Mostly reliable but very expensive.
Equipment and installation costs are high, but it works well.
Normally Government telecom companies budget the bandwidth separately
for the leased circuits based Internet subscribers, and overselling
is very less. At work, I have found a 2Mbps (advertised) leased line
to be good enough for an office of 100 people. ofcourse, it depends
on the usage, employee culture, and filters at firewall ;-).
Cable
In the cities, Cable TV providers and some ISPs
have introduced the Cable Internet. Most of the ISPs use the local
cable providers to run CAT5 (cheapest) cables between buildings, and
put hubs and switches wherever necessary. In most cities the cabling
is overhead, using the street light poles, so it is very much prone
to weather and vandalism. On top of that, in case of a blackout (very
often) the hubs and switches go down, so even if you have a
uninterrupted power supply at your home you
can't use the Cable Internet. In more annoying cases, switches and
hubs are often installed at somebody's home in your neighborhood,
and if that person switches off power to the equipment and goes on a
vacation, you are disconnected from the world.
CDMA
There is only one provider which gives Internet
Access in most parts of the country using
3G CDMA wireless technology. The service is good enough, but not very
attractively priced. There were reports on Indian rant sites, that
access for only well-known ports is
available, but I think they have rectified
it. Support is not worth mentioning, I
mean, it is a giant petrochemicals company
that diversified into Tech business, what do they know about Customer
Care?. This however is the best, and
cheapest option for road warriors.
GSM/GPRS
Its offered by almost all GSM providers, and it
comes at a cost of nearly 500 rupees (12$
US) per month. Speed is less than attractive, but it works. I am
using it from home, with no availability of a fixed Internet
connection, GPRS at this cost is a good compromise. My phone is
always connected to the Power outlet, and always pointing its IRDA to
my Laptop computer. It keeps on checking my mail, and RSS feeds. At
12$ per month.
Update: A cell phone with active GPRS connection causes noise in your computer speakers, I am forced to mute the volumes on my laptop.
Thursday, May 13, 2004
Indian EVM compared with Diebold
I am writing this while the results are coming out for the Indian Assembly Election of 2004. 8 news channels are showing results updating like stock prices on the screen. Yes a 3 second update. The counting is although very simple and fast because Electronic voting machines were used. But counting from 1.5 Million voting machines is expected to take almost 3 to 4 hours, since, the Electoral Process needs to be followed in all its bureaucratic steps. The numbers tell that the ruling party (BJP) is trailing against its main opposition (INC established by Gandhi). Nobody has doubts about the fairness of the elections. Re-voting is announced in few stations, due to various reasons. But over all the biggest democracy in the world has lived up to its expectations. The ruling party, over-confident of its good economic track record, declared the elections 8 months ahead of schedule, riding on the general “feel-good” factor going on in the country these days. But now it seems, the voters are not that gullible, the opposition INC has got 216 seats out of 539 and will be invited to prove the majority and form a government. Indian democracy is of the Parliamentary type, not the Jeffersonian Model (Presidential type) practiced in America. In the Indian Equation, if a party gets 272 seats out of 539, they can form a government. and guess what, we may see an Italian born woman as Prime minister of India! If this is not democratic then what is?
Last few months have brought very serious discussions on the Net regarding the use of Electronic Voting, and the security of it. In the USA, the saga related to Diebold and its opposition is well known. I do not know the electoral process in the United States, but I attempt here to compare the Technology used by the Indian Election commission and the Diebold AccuVote system. I present here the Information I have about the Indian system, and the information about Diebold got from the web.
Reading this article, some of you might remember that Cold war era joke, about NASA and its multi million dollar experiment with a pen that can write in micro gravity to solve the writing problems of astronauts, and the Russian solution of using a Pencil to solve the same problem. IMHO, the Diebold system is too complex for a simple and straight forward task such as voting. Windows CE, Modems, PCMCIA storage cards, Touch screen GUI, On-screen writing facility, Voice-guidance system, multiple language UI, DES Encryption, centralized voting Server, a step-by-step wizard to cast a vote, Microsoft SQL Server to store votes, Backup servers etc. are all unnecessary. All geeks know that a smaller and simple system is more secure, more code means more cost, more chances for bugs, more threats to security. You cannot make a system that is “guaranteed” as secure. A lot depends on the electoral process and the integrity of election officials.
The Indian Electronic Voting Machines (EVM) are designed and developed by two Government Owned Defense Equipment Manufacturing Units, Bharat Electronics Limited (BEL) and Electronics Corporation of India Limited (ECIL). Both systems are identical, and are developed to the specifications of Election Commission of India.
The System is a set of two devices running on 6V batteries. One device, the Voting Unit is used by the Voter, and another device called the Control Unit is operated by the Electoral Officer. Both units are connected by a 5 meter cable. The Voting unit has a Blue Button for every candidate, the unit can hold 16 candidates, but up to 4 units can be chained, to accommodate 64 candidates. The Control Units has Three buttons on the surface, namely, one button to release a single vote, one button to see the total umber of vote casted till now, and one button to close the election process. The result button is hidden and sealed, It cannot be pressed unless the Close button is already pressed.
The voting unit has a list of candidate's names and their Party Symbols pasted on the surface, and a Blue button to cast a vote faces ever candidate's name. The Party Symbols (like a Lotus, an elephant, a horse etc.) are approved by the election commission to be unique, All political parties use these symbols while campaigning, and illiterate people can identify their candidates by looking at his symbol, and pressing the blue button in front of his symbol.
Here is how the voting process goes,
-
The Voter is Identified, by his Government Issued Voter Identity Card, or his Public Distribution System's Ration Card, when he enters the polling station.
-
Voter's finger is marked with a special ink, in such a way that the ink spreads from finger skin to nail in a small dot. One cannot remove this Ink without hurting himself. The Ink washes away in two week's time.
-
The Electoral Officer then Presses a button on his Control Unit, that releases a single ballot, for the voter to use, this of course is electronic so it just enables the Voting unit to register one Vote.
-
Now Voter enters the voting Booth, and preses a Button in front of name and Election Symbol of the Candidate. This action blinks an LED in front of the candidate's name and sounds a loud and long Beep, that declares that the vote is casted.
Notes: The System accepts only 5 votes in a minute. The Indian Election process is distributed in such a way that there are never more than 1500 voters for a single polling booth. So, even if armed men capture the polling station, they cannot cast 1500 bogus votes in less than 5 hours, and Indian police is not as slow as the bollywood movies project them to be. No voter has to travel more than 2 Kilometers to cast his vote. Its fairly easy for an election officer or opposition political agents to identify people who attempt to appear twice with different identity. (The Ink on the finger is the main reason).
And here is how the results are obtained from the machines.
-
After the voting is over, electoral officer presses the Close switch on the Control Unit, after which no votes are registered by the unit. The total number of the Votes registered are noted by all stake holders (political party agents) and then the control units are put into its own special carrying case, and sealed for transport.
-
Control Units from all Polling stations are transported to the nearest District headquarters.
-
On the day of counting the seals of the Control Units are opened. The control unit has a Results Button which is physically secured by a protective seal, this button is pressed to obtain the results. The Machine gives the Serial number of the Candidate, and the votes that he has won.
-
The Election commission takes a decision to ask for a re-election if the machines are found to be tempered with. Or if the count of signatures or thumb impressions (yes, India's illiterate also take part in the democracy) on the voter register do not tally with the number of votes registered by the Voting Machine. In this election, about a 100 polling booths, (I think) were asked to conduct the election again. This number is small, for the size of Indian elections.
-
In case of disputes, the machines are preserved for the courts to decide upon, other machines are used for next election after reseting the memory.
Diebold system works on Microsoft software, it has no seals on locks and panels to detect a tempering. It has a keyboard interface (!!!) and the server was tested to have “Blaster” virus. One report on Wired says a lady stumbled upon some files from Diebold, and found that the votes were stored in MS Access files. It also has a PCMCIA SanDisk card for local storage. A touchscreen GUI and a network connection to send the results to a server after encrypting it with DES.
The Indian EVM is just plain circuit, with some assembly code. A few LEDs, and two Seven Segment LED displays. One EVM can list 16 candidates, but up to 4 EVMs can be Linked to accommodate 64 candidates. (In a country of a billion people its possible to have 64 candidates for one single constituency.)
Diebold has received its share of criticism from Techies and Paranoids. Techies are Concerned about the vulnerability of the system. Some concerns are right, like having a network to communicate votes to a central server, exposes the system to unimaginable risk. I mean, we all know how safe is a windows box on a network ;-). But, some criticism is just not right, like the keyboard interface, and card reader jamming etc. No, electoral officer in his sane mind would allow a voter to walk into a booth with a keyboard in hand, and would not let him be inside the booth long enough to duplicate a smart card or to open the voting machine to do some EPROM Programming, or to run a forceful algorithm to break DES.
India's leading daily newspaper, carried an article on the eve of the elections, saying that the microchip containing the code of the EVM machines can be copied in minutes. The article was titled, “Winning Elections Made Easy”. The article was written by an Indian Professor living and teaching in America. But I don't think it is possible, as long as the Electoral system and Election officials function to their expectations. Usually Indian elections take place during school Vacations, and Teachers are recruited by the Indian Election Commission to perform duty as election officers.
I guess, The differences in both technologies are as follows.
Device type
EVM: Embedded with Assembly code
Diebold: Embedded with Windows CE, and C++ code
Visual Output
EVM: Single LED against each candidate's name
Diebold: Color Touchscreen, with GUI software
Operating System/ Software
EVM: None, the Assembly code to register number of votes is all it has.
Diebold: Windows CE, and C++ code stored on the Internal Memory and
PCMCIA cards.
Records/ Audits
EVM: The Voting unit doesn't store anything, the control unit records
the number of votes casted for each candidate against his serial number. No
record to link person-to-vote.
Diebold: Internal ribbon printer. And PCMCIA storage for records and
audit trials. Additionally the GEMS server also stores the votes and audits.
Accessibility
EVM: Blind people are allowed to bring an escort into the polling booth
to help them vote.
Diebold: Optional Audio component to assist the visually impaired.
“Magnify” feature to enlarge the text. I guess all Windows CE Accessibility
features are available on these systems.
Control
EVM: Control Unit accumulates the votes, it is a device with flash
storage and seven segment LED displays. They are connected to voting units with
a 5 meter cable, the Unit has a switch to issue a ballot for a voter.
Diebold: Two GEMS servers one primary and a backup, for every polling
station, that connects to the voting units to “load the ballots” (!!) and then
voting units work independently. They are again connected at the time of
results.
Security of Access
EVM: Physical security is ensured by the electoral officers. Unit is
sealed during transport.
Diebold: GEMS servers have access through Supervisory Smart cards, and
PINs, some users have login and password access.
Ballot Issue
EVM: Ballot is issued by Electoral officer by pressing a button on the
Control Unit. It allows the voter to press one button on the voting unit.
Diebold: Voter access smart card is issued in an envelope for a
terminal. Voter can put it in the assigned terminal and cast his/her vote.
Storage of Votes
EVM: In Internal Non removable memory of the Control Units. All control
units are transported physically to the counting center.
Diebold: In a PCMCIA card hidden in the Voting Unit. Results are
“transmitted” using modems to the counting center.
Cost of the System
EVM: About 10500 Rs. (230$) for 1 control Unit + one voting unit.
Diebold: About 3300$.
Power Supply
EVM: 6V alkaline batteries
Diebold: electricity
Capacity
EVM: 3840 Votes (the electoral process distributes one polling station
for not more than 1500 voters) so its large enough.
Diebold: Over 35000 votes.
Localization/Languages
EVM: The Voting unit has a Non tear-able printed sticker, which is
printed in any of the 18 (yes 18) constitutional languages, spoken in the
region. The Election Symbol of the candidate allows people who cannot read that
language or cannot read any language at all to vote by pressing the button
against the symbol.
Diebold: Supports more than 8 different languages using GEMS software.
Developed by
EVM: State owned Defense Equipment manufacturing units.
Diebold: Private company.
Buyer/ Owner
EVM: Election commission of India buys it and ownes it to conduct
elections everywhere in India.
Diebold: Individual states / counties buy the systems, and use it to
take part in elections, each state can decide based on its law, which system to
adopt.
Links
How E-Voting threatens democracy [Wired]
Indian Electronic Voting Machine
Presentation on EVM [Election Commission of India]
