In mid-April the Palm Coast City Council agreed to survey residents about an issue that’s divided the population over the years: should the city relax its ban on commercial vehicles parked in residential driveways?
The city conducted its survey in May. It posed just the one question: “Would you be in favor of amending the Code of Ordinance Sec. 44-34 (C) to allow ONE commercial vehicle (passenger car, panel van, pickup truck or similar) with advertising markings to park in a residential driveway?”
The results are in: there were 16,491 responses. How they split is a mosaic of caveats, because the survey was flawed.
Taking the raw numbers alone, it’s unequivocal: 70 percent of the responses were opposed to relaxing the rule. But when the survey is weighed for manipulation by those who took the survey, the results are almost dead even: 49.32 percent for relaxing the rule, 49.29 percent for not relaxing it, and 1.38 percent undecided.
What does that mean, weighed for manipulation, and why should there be two sets of results?
The answer is in the survey’s design. It was web-based. Many such surveys lock down a response from a particular browser after one participation. Palm Coast’s did not. It allowed the same respondents to take part as many times as they wished. A respondent could sit at a computer all day and click “No” again and again–as many respondents did. Some did the reverse, clicking yes, but not nearly as many.
The city’s IT staff could see all this by analyzing responses from their IP addresses. Every vote was tied to an IP address. It was obvious when the vote was jiggered.
For example, one Palm Coast-based IP address returned 2,062 No votes, eight undecided, and one Yes vote. That IP address alone accounted for 12.5 percent of the raw votes. It was obviously a manipulated vote. For starters, no single IP address in the city can account for over 2,000 people. Even taking account of large organizations, such as the school board or City Hall, where an organization has one IP address, no organization (or business) has anywhere near that many people in one office complex. Someone with too much time on their hands, or a group of people, systematically clicked No over 2,000 times.
The same pattern is repeated with almost 150 IP addresses accounting for thousands more votes, each address returning from tens to hundreds of lopsided votes one way or the other, with zero vote in the opposing column–a mathematical impossibility in a normal vote, especially about an issue that has divided the city more or less down the line for years.
The top 20 busiest IP addresses, which accounted for almost 6,000 votes between them, or 35 percent of the vote, broke down this way: 5,593, or 95.7 percent, opposed, 21 undecided, 228 (or 3.9 percent) for. Just two IP addresses accounted for all but three of the votes in favor, just as all the votes opposed had almost no “yes” votes to go along with them: again, indications of clear and unquestionable manipulation.
In all, votes were traced back to 5,000 IP addresses. Those 20 IP addresses accounted for 35 percent of the vote, but represented only 0.4 percent of all IP addresses participating: more indications of willful manipulation.
The survey could have been determined to be useless at that point. But the city’s IT department then applied its weighing method to essentially neutralize the manipulators. Every single one of the votes from the 5,000 IP addresses was weighed to an average.
For example, where an IP address produced 473 votes against and zero for, the weighted result was 1 for, 0 against.
Where an IP address produced 631 against, 8 undecided and 0 for, the weighted average was 0.9958 against, 0.0042 undecided, and 0 for. And so on.
Similarly, where an IP address returned a single vote, for or against, that vote counted as one vote, for or against. In other words, every IP address was reduced to its weighted result, and those results were then added up for the weighted total.
That total showed the 49-49 split, but for its decimal difference. Those final percentages “represent a total average of the averaged responses for all IP addresses,” the city’s analysis stated.
Weighting that way is not foolproof: for example, many a household would have had two or three people taking the survey. If they split their votes, that would have been reflected in the weighed results. But if they did not–if three actual people in the same house voted No, the total would still read as one No vote, not three. But in the same way, when manipulators attempted to run up the score in their favor, their votes, too, were averaged out.
The 49-49 split reflects the current split on the council: two council members want the rule relaxed (Ed Danko and Victor Barbosa), and two don’t (Nick Klufas and Eddie Branquinho), though they all agreed to the survey. When they agreed to it, they were not aware of the survey’s flaws.
City staff appears aware enough of the survey’s problems. The administration will present the numbers to the council on July 13. The matter may all be moot after the special July 27 election. Whoever is elected will provide a swing vote regarding numerous issues at the city, the commercial vehicle rule being one of them.
The current rule forbids commercial trucks with commercial lettering of a certain size from parking in residential driveways when they are not on a job assignment. If such commercial trucks do park for lunch or overnight, they are required to be covered with a tarp.
The proposed relaxation of the rule that council members have discussed would not be a free-for-all: larger trucks would still be banned regardless. But smaller work trucks such as work vans would be permitted, as long as the number of such vehicles is limited to one in a driveway.