Vote fraud and verification in New Zealand: how contests deal with it
Vote fraud in New Zealand contests is met with two documented tools: contact details collected at the ballot and a log reviewed after it. Bird of the Year asks every voter for a name and an email address, a step the competition states openly as a measure to help prevent fraudulent voting. In 2020 its organisers detected more than 1,500 fraudulent votes cast with fake email addresses from the same IP address. In a later edition, illegitimate votes briefly pushed the kiwi pukupuku to the top of the preferred bird leaderboard before Forest & Bird removed them from the competition. Removal, not a named ban on entrants, is the documented sanction.
Facts last tested on 8 October 2026.
The checks a ballot meets at entry.
The clearest documented check in a New Zealand contest is the one Bird of the Year applies at the ballot itself. Voters are asked for a name and an email address, and the competition states the reason in its own words: the details help prevent fraudulent voting. That request does two jobs at once. It ties every ballot to a contact point, and it creates a record the organiser can reread later. A vote with no attached identity is a vote nobody can check afterwards, and the organisers of a public contest know it when they design the form.
Email addresses earned their place on that form the hard way. In 2020 the organisers detected more than 1,500 fraudulent votes cast with fake email addresses from the same IP address, a batch large enough to push the little spotted kiwi to the top of the leaderboard before the result was corrected. The address field did not stop the batch on its own. What it did was leave a trail: fake addresses, one shared connection, and a pattern readable in the log once someone went looking for it after the votes had landed.
Read together, the two cases describe a two-stage structure. Stage one is collection: the form asks who is voting, at minimum a name and an email address. Stage two is review: the organiser compares the collected details against the ballot record and looks for repetition. Neither stage depends on the other catching anything in real time. A batch of fake votes can pass straight through the form and still be identified later, because the log keeps every address and every IP address the ballots arrived from, with timestamps attached.
The pattern matters for any public vote in the country, small or large. Contact details and server records are cheap to collect and expensive to argue with afterwards. A contest that skips the collection stage has nothing to compare once the count looks strange, and the 2020 case is what comparison looks like when the material exists. That is the whole documented mechanism, and it rests on ordinary form fields rather than on any specialised instrument.
- A name and email field is the front line, and Bird of the Year states that purpose openly.
- One IP address producing 1,500 ballots is exactly the pattern a log review is built to find.
- Verification does not need to block a fake vote at the door; it needs to leave a record.
- The 2020 case shows why contests ask for contact details even when nobody reviews them immediately.
Illegitimate votes get taken back out.
The second Bird of the Year case shows what removal looks like in public. Illegitimate votes briefly pushed the kiwi pukupuku to the top of the preferred bird leaderboard, and Forest & Bird then removed those votes from the competition. The word to notice is briefly. The fake lead was visible for a while, and the correction came after. A leaderboard in an ongoing contest is therefore provisional rather than final, and readers who see a sudden jump are watching a number the organiser has not yet finished checking against its own records.
Removal is the documented sanction, and it operates on votes rather than on announcements. The competition did not quietly delete the batch and say nothing; it named the illegitimate votes and stated that they had been taken out of the running. That transparency matters for anyone reading a live leaderboard in any contest in the country: a number can move down as well as up, and a downward move usually means the organiser found something during review. The final standings are the standings left after the review, never the first draft.
For an entrant, this ordering has a practical consequence. Votes that arrive in an unusual pattern carry a problem that does not expire when they are counted; they stay reviewable for as long as the organiser keeps the log. The kiwi pukupuku case involved votes already displayed on a public leaderboard, leading the whole field, and they still came out. Counting is not the last word. The audit that follows counting is, and every contest with a per-ballot record has one available to run.
- A leaderboard lead built on fake votes lasted only until Forest & Bird reviewed its records.
- Removal was announced publicly, which tells readers that a moving number can be a corrected number.
- Votes already displayed at the top of a leaderboard were still removed from the competition.
- The result that stands is the one left after removal, never the one that appeared first.
The limits of a contest's view.
An organiser running an online vote sees what its own form and server record: the ballots, the email addresses attached to them, the IP addresses they arrived through, and the timing of each submission. That is the whole toolkit, and the 2020 case shows exactly what it catches. What the organiser does not see is anything the form never asked for. It cannot see who is really behind a working email address, and it cannot see past the connection a vote came in on. Every verification rule a contest publishes lives inside that boundary.
This is why verification differs from contest to contest, and why the published rules are the real answer to the question. Bird of the Year asks for a name and an email address; another competition can ask for more or for less, under rules it writes itself. The organiser decides what to collect, what to compare, and when to review. Nothing in the documented cases points to a hidden instrument beyond those choices. The visible instruments are a form, a log, and a person willing to read both closely.
Honest limits cut in two directions here. The 2020 batch was caught, which shows the checks work on large coordinated patterns. Smaller numbers of votes are a different question, and the documented records do not say how few votes pass unnoticed or how many get caught, because organisers publish the correction and not the arithmetic behind it. Anyone evaluating a contest starts with its published rules and treats the log review as the mechanism that settles the result when the count looks wrong.
- The organiser's view is the form plus the log: ballots, addresses, IP records and timestamps.
- No documented case in the country needed anything beyond contact details and server records to find a batch.
- What a form never collects, an organiser never sees, and that boundary shapes every verification rule.
- Organisers publish corrections rather than detection rates, so the record shows what was caught and nothing else.
What the rules do with breaches.
The documented sanction is removal of votes. In the kiwi pukupuku case, Forest & Bird took the illegitimate votes out of the competition and let the corrected standings stand in their place. Whether the people behind those votes faced a further ban is not stated in the record, and this page does not guess. What the record does establish is that a contest will reverse a result its own review has discredited, mid-competition, at the very top of the board, in front of the same audience that watched the fake lead appear.
Disqualification sits inside the organiser's own rules rather than inside any external watchdog. A contest publishes how votes are counted, and the same document governs what happens when votes breach the count: they are removed, and an entrant whose votes were removed loses the standing those votes had created. That is the chain, and every link of it belongs to the organiser running the vote. No appeal body appears in the documented cases, and no external review of a contest leaderboard is described anywhere in them.
One boundary closes the subject. Cheating in the electoral sense, the kind attached to national elections, is a separate matter with its own law and its own institutions, and none of it is covered on this page. Inside contests and public polls, the mechanics are plainer: a published rule, a collected detail, a log that keeps everything, and an organiser willing to delete votes its review has judged illegitimate. The 2020 batch and the kiwi pukupuku batch both ended the same way, struck off the board and replaced by corrected numbers.
- Removal of votes is the documented penalty; further bans are not described in the published record.
- A contest's own published rules govern disqualification, and the organiser enforces them without an outside body.
- Electoral cheating of the national kind is a separate subject under separate law; this page covers contests and public polls only.
- Both documented cases ended with illegitimate votes deleted and corrected standings published in their place.
Where the rules stand
The page is built as a risk-check. Zone one covers verification at the ballot: the contact details a form collects and the IP records a server keeps. Zone two follows the votes out the other side, where illegitimate ballots are removed and a public leaderboard is corrected. Zone three sets out what an organiser can and cannot see with these tools, and where the published rules draw the line. Zone four closes with disqualification and the boundaries of the record. The six concerns below answer the questions readers bring about fake votes, IP checks, audits and removal, each answered in its first sentence from the documented cases.
What we refuse
This site does not sell votes, does not arrange votes for any contest, and never explains how to evade the checks described here: no bots, no proxy switching, no duplicate accounts, no captcha workarounds. The page makes no claim that any voting method guarantees a result, and it states no detection rate, because organisers publish corrections rather than statistics. Cheating in the electoral sense belongs to other bodies of law and is out of scope here, as are paid surveys and television talent shows. Nothing on this page is advice on obtaining votes; it is a record of what organisers catch and remove.
Typical questions
The documented answer combines a form and a log. Bird of the Year collects a name and an email address with every ballot, stating that the details help prevent fraudulent voting. In 2020 its organisers detected more than 1,500 fraudulent votes cast with fake email addresses from the same IP address. Detection came from the record the form created: repeated addresses and one shared connection, readable in the log. A contest that collects nothing has nothing to compare, which is why contact fields are the standard opening move in a public vote.
IP checking is documented. The 1,500-vote batch of 2020 was identified partly through votes arriving from a single IP address, which means the organisers read IP data when reviewing ballots. Captchas are not named in the documented records, and this page does not claim them for any specific competition. What the record shows instead is email verification at entry and address comparison afterwards. An organiser wanting a stronger gate adds further steps under its own published rules, but the cases on record rest on the two documented ones.
The kiwi pukupuku case is the documented example. Illegitimate votes briefly placed the bird at the top of the preferred bird leaderboard, and Forest & Bird then removed the votes from the competition. Removal happens to the votes, and the standings are corrected accordingly; a lead built on removed votes disappears with them. The public record names the removal and the reason, so readers see a corrected board rather than a quiet edit. Suspicious votes stay reviewable for as long as the organiser keeps the log open.
Yes, and the two Bird of the Year cases are audits in everything but name. In each one, organisers went back over ballots already cast and separated illegitimate votes from legitimate ones. An audit of this kind needs three things: a per-ballot record, a detail that identifies each ballot, and a reviewer with the authority to correct the result. The name and email fields supply the second, the log supplies the first, and Forest & Bird supplied the third by removing the offending votes from the standings.
The verified steps are two. First, collection: the ballot form asks for a name and an email address, with the stated purpose of helping to prevent fraudulent voting. Second, review: the organiser compares the collected details and the IP records after votes arrive, which is how the 2020 batch was found. Neither step belongs to a secret toolkit; both are visible in the competition's own statements. Steps beyond these two exist in online contests generally, but the cases on record rest on collection and review.
The documented cases show votes disqualified rather than entrants named. Forest & Bird removed the illegitimate votes from the competition and published the correction; the record does not state that any person was banned from future editions. What it does state is that a top-of-the-leaderboard lead was reversed once the review finished. For an entrant the lesson is direct: votes that fail the organiser's review are struck, the standings move, and the organiser's published rules govern whatever follows for the people responsible.