How We Kept The Underdog Awards Fair
Share
I'm taking a deep dive into how the Underdog Award voting works, because I think if you're asking people to trust a result, you owe them a look at the workings.
This won't be a technical document. It's an explanation of the principles I built the thing around, and (just as importantly) what I actually checked before I announced anything.
One person, one vote, and how that's enforced
The obvious problem with any public vote is that clicking a button is easy, and clicking it two hundred times is only slightly harder.
So a vote in the Underdog Awards isn't cast when you click. It's cast when you confirm it from your email inbox. You pick your nominee, enter your email address, and you get a message with a confirmation link. Until you click that link, your vote doesn't exist as far as the count is concerned.
That single step does most of the heavy lifting. It doesn't make cheating impossible (anyone determined enough can create email addresses) but it raises the effort enormously, and effort is what deters casual ballot-stuffing.
On top of that, the system blocks repeat votes in the same category from the same address. That check happens in the database itself rather than in your browser, which matters more than it sounds: anything enforced only in the browser can be worked around by someone who knows what they're doing. Rules that live in the database can't be, because there's no path to the vote count that doesn't go through them.
The same applies to the closing deadline. When voting closed at midnight, it closed on the server. It wasn't a case of the page hiding the buttons while the back door stayed open.
Being honest about the imperfect bit
There's no vote of this kind that can't theoretically be gamed by someone with enough time and enough email addresses. Anyone who tells you otherwise is either selling something or hasn't thought about it hard enough.
What you can do is make it costly, and then, once voting has closed, go and check that nothing skewed the result.
What I checked before announcing
This is the part I think matters most. Just to be clear about what this was: I wasn't investigating anyone. I was checking that the result I was about to announce was actually the result. Votes can be distorted by all sorts of things that have nothing to do with bad faith, emails failing to arrive, a technical fault, an unnoticed quirk in how something was counted. I've also been on the receiving end of deliberate manipulation before, in other contexts, and it left me permanently unwilling to take a number at face value.
So I don't assume a clean result. I go looking for reasons it might not be, and I only announced when I couldn't find any.
Here's what I looked at.
The margin against what was still outstanding. Some people vote and never confirm - they miss the email, get distracted, or change their mind. Those votes sit in limbo, and they're the most likely thing to change an outcome.
Best Small Venue was the close one. The Feathers finished on 507 confirmed votes, with Bear Cave on 426 and 90 still unconfirmed. For Bear Cave to overturn an 81-vote gap, 82 of those 90 would have had to land - a 91% confirmation rate, far above anything realistic for votes that had already had their chance to be clicked.
Best Alternative Club Night wasn't close. Bunker 13's lead was wide enough that the runner-up couldn't have caught them even if every single outstanding vote had confirmed.
Whether any nominee looked unusual. If one nominee has a dramatically different pattern to everyone else, that's worth a hard look. I compared the proportion of unconfirmed votes across every nominee. Bear Cave came out at 17.4% and The Feathers at 18.9% - within a point and a half of each other, and in line with the field generally. If anything, the leader had slightly more outstanding than the challenger. Nothing stood out.
Whether votes arrived in suspicious bursts. Genuine voting spreads out - people see a post, tell a friend, vote over lunch. Automated voting tends to arrive in tight clusters. I checked the timing distribution and found the ordinary, scattered pattern you'd hope for.
Where the email addresses came from. A healthy vote shows the same mix of email providers you'd expect from any group of people: mostly the big consumer ones, in roughly the proportions they're used. Heavy concentration in one unusual provider would be a red flag. The spread here was entirely ordinary.
Whether the emails were actually arriving. This one has nothing to do with cheating and everything to do with fairness. If confirmation emails were bouncing, some people's votes would have failed through no fault of their own - and that would skew a result just as effectively as anything deliberate. I checked the delivery records: the only failures were obvious mistypes of email addresses, spread thinly across the whole voting period, with nothing recent. The system was reaching people.
All five checks came back clean. That's why I was comfortable announcing.
What I'm deliberately not telling you
I've explained the principles, not the exact thresholds, rules and checks. That's intentional. If I run this again (and I'd like to) a fully detailed writeup would be a useful manual for anyone wanting to beat it (I find it hard to imagine that anyone so inclined and capable couldn't find much better things to do with their time, but I've had bigger surprises in the past).
So: the mechanisms are real, they run server-side, and I audit the results afterwards. The specifics are hush hush.
Why bother with any of this?
The prize money here isn't life-changing. Nobody's retiring on it. It would have been easy to run a poll, take the top number at face value, and post a graphic.
But the whole point of the Underdog Awards is that the winners are places doing something genuinely valuable on very little money and very little recognition. A result that anyone could dismiss as rigged, or gamed, or just carelessly run would have been worth nothing to them. The award is only worth having if the process behind it holds up.
So I did my damdest to make sure it held up and now you know how.
Built in-house on Shopify, with a Supabase backend and serverless functions handling vote submission, email confirmation and the closing deadline. Nominations, voting, verification and the judges' review panel were all custom-built for these awards.
The Underdog Awards is funded by Kate's Clothing, by redirecting money that would otherwise have gone to advertising on the big platforms.