Robots 'Stuffed' the Hydrogen Ballot Box
Young man runs from fighting near Donetsk airport in troubled Ukraine. The IP addresses of some of the 'Yes' votes came from an ISP in this embattled city. Others came from Belarus, across the border in Russia.
I should have looked into this months ago. When, starting last November, our monthly reader poll responses jumped from a few hundred to more than a few thousand, I should have guessed that something was amiss. It took a reader expressing his own suspicions that 'pro-hydrogen trolls' were electronically skewing the vote on our most recent question about the future of hydrogen fuel cells to spur me to action. The result was my blog 'Stuffing the Hydrogen Fuel Cell Ballot Box.' I explained that while there were a few instances of people voting more than once, the evidence suggested that it wasn't enough to sway the votes one way or the other.
However, a much deeper dive into the question the last two days has revealed that the votes are, in fact, being electronically manipulated from afar, but not by imagined 'pro-hydrogen trolls.'
Let me take you back to Fall of last year. In October, 2013, we asked readers, "Should electric car maker Tesla be worried now that General Motors says it aims to launch competing productions?" 630 people responded with 61.43% answering yes, 27.62% answering 'no', and 10.95% checking 'Uncertain.'
That was a pretty typical level of response we'd seen over the years. In November, we then asked, "With the number of electric vehicles on the road in America now under 200,000, was President Obama wise in setting a goal of 1 million EVs by 2015?"
Suddenly responses jumped to 10,895 with 98.55% answering 'yes'. Puzzled by this, I started to wonder what was going on. At the time, however, I had no way to track where each vote was coming from. When the December poll was similarly skewed, I decided it was time to act. I wrote code to capture the IP address of each vote.
January 2014 showed something a little less lopsided, as the total number of votes dropped to 8,302 and the yes vote to 90.79%. Part of me wanted to believe that the new format, which made the poll more visible on the web site that accounted for the jump in total votes, but the disproportionality of the votes troubled me. Each month, the skew was the same, overwhelmingly 'yes' to every poll question. Example:
"Is the addition of a fuel cell or gasoline range extender to an electric car worth the added expense to you?" -- 93.34% 'Yes'.
"Would you buy a Tesla electric car if it is priced less than $35,000US?" -- 96.93% 'Yes'.
"As an American citizen, if federal tax credits went away, would you be less inclined to buy an electric car?" -- 97.46% 'Yes'.
"Should taxpayers, via government grants, continue to fund the build-out of public charging infrastructure?" -- 99.16% 'Yes'.
"New EPA CO2 rules will require dramatic shifts away from coal by US electric utilities by 2030. Do you believe this is an appropriate move by the federal government?" -- 99.22% ' Yes'.
It really didn't matter what the question, the result was always the same: 'Yes'. Do you support President Obama's policy initiatives? Yes. Will hydrogen fuel cells win out over battery electric cars? Yes.
As I explained in the original 'Stuffing the Ballot Box' piece, while I did find some multiple voting going on, it didn't appear to be enough to influence the general outcome of the vote. A cursory look through the database of IP addresses didn't reveal anything unusual. All the addresses seemed pretty random, which you'd expect, so I reported my initial findings in the piece.
Still, those huge numbers and skewing troubled me. There was something fishy going on here, my gut told me. For several years now, we've experienced a variety of attacks from abroad. Why? I haven't a clue. I decided to dig deeper. I began with the first moment the hydrogen poll went live and began looking up the origin of the IP address linked to each vote, 99% of them 'Yes.'
The poll went live at 1:12 PM local time with my own test 'Uncertain' vote. The very next vote was a 'No' vote that came at 1:14:58 PM local time. The first 'Yes' vote came at 1:47 PM, followed at 1:56 PM by the second 'No'. The next two 'Yes' votes came two minutes later, followed 5 minutes later by the third and fourth 'No' votes. For the first hour, the results were what you'd expect: 3 'Yes', 4 'No', and my one 'Uncertain'.
Then at 2:18 PM, the 'Yes' votes began rolling in. The next 'No' vote wouldn't happen for another three hours at 5:14 PM local time.
So, where were these votes coming from? Using Network Solution's Whois/IP lookup revealed a distinct trend: virtually all of them were coming from an organization based in Amsterdam, Netherlands called Réseaux IP Européens Network Coordination Centre, otherwise know as RIPE.net. Other votes were coming from Brisbane, Australia. That organization is APNIC or Asia Pacific Network Information Centre. A third group also kept cropping up: LACNIC, short of Latin America and Caribbean Network Information Center. These are loosely based bodies who oversee regional network resource allocation. In effect, they are information clearing houses for their regional Internets. They aren't actually responsible for traffic origination. That comes from local ISPs within their geographic regions, so to find out where those votes were coming from, beyond the fact they were originating somewhere in either Europe, Asia or Latin America, I had to dig deeper still.
It would be tempting to look for a larger conspiracy. Amsterdam? Isn't that where Royal Dutch Shell is based? Maybe they're behind this, just as the reader suspecting 'pro-hydrogen trolls' wondered. But if that's the case, then what's the connection in Brisbane and Montevideo, Mexico, here LACNIC is based?
It would appear, at least from where I am at just this moment after having randomly looked up dozens of IP addresses using RIPE, APNIC and LACNIC's IP lookup databases, that none of this has anything to do with any particular anti-EV political or economic agenda. Not when the sources of these votes come, quite literally from all over the globe, from troubled cities in eastern Ukraine to Belarus to Turkey. The same with the APNIC originating addresses: China, Taiwan, Korea.
The very fact that 97-99% of the votes are always 'Yes', irrespective of the question being asked, clearly points to mindless computer robots whose only mission is to cast a vote on EV World, likely just to mess with us. Again, why? I haven't a clue. We just aren't that important to the scheme of things.
What continues to trouble me though is the concerted planning and effort that's gone into this totally worthless attack. I can't believe that there are thousands of individuals from Donetsk to Taipei, who may not even speak English, who are deliberately expending energy on this. My personal suspicion is there are one of two possibilities: their computers have been either willingly or unwittingly hijacked to randomly cast a 'Yes' vote on EV World polls, or there's a program running out there that 'spoofs' all these addresses, again randomly injecting a 'Yes' vote just for the sheer spite of it.
Come on, guys, do you really think EV World is that important? I don't. It also get's me to wondering, if this is happening to such an inconsequential web site, who else are they targeting? Or maybe we're just an 'easy mark' for hacking.
I'll probably never know who or why they've decided to pick on EV World - and we've been the target for other attacks in the past - but it sure would be nice if they'd find something more productive to do with their time and obvious talent. How about you comrades help me make the world a better place? I really could use the help.
Anyway, I thought you'd all appreciate an explanation of what's been going on. Hopefully, we can resume asking your views on important questions to the EV world, eliciting real human responses instead of from mindless 'robots.'
UPDATE: The introduction of a Captcha procedure when voting now appears to have rectified the problem. I deleted all the votes subsequent to that 2:18 surge of 'Yes' votes, and as of Sunday morning, July 13th, the tally seems much more representative of reader views. Of the 50 votes so far, 10 have voted 'Yes', 34 'No', and 6 'Uncertain.' If you had votes previously, chances are your votes, along with all the 'bot votes was deleted. Please consider voting again.
blog comments powered by Disqus