The green from Emisar lights are what I would have picked for green. The “green” here in the pictures look more like the sand color from Convoy.
Purple and white for me.
I wonder what these would look like after being baked.
Today’s totals with 175 votes:
- 105 purple
- 105 blue
- 72 green
- 68 white
+1
Wish they made one in red…
Purple or blue with TK logo. Clear anno should be a choice.
Purple and Green
Purple and Green
Yes, but this is misleading TK. With these numbers, it is entirely possible that no voters actually prefer purple over any other color (although I doubt that is the case). If everyone who voted chose purple as the second choice in a pair, it would still garner 60 points, per your analysis, which would have made it at least tied for highest preference, even though it was no one’s first choice. This is the problem with combinations versus ranked choice, and why I don’t think this poll is technically predictive, unless you assume that all voters intend to buy two flashlights.
Original option seems missing?
I don’t think it is intended : Which two colors of FW3A you prefer?
I responded to the question but don’t plan to buy any. 
Voting system mechanics are something I’m really interested in, but I’m not sure this is a great place to get into detail. Long story short though… the desired style of vote was a “vote for two” system. But BLF doesn’t have that, so Neal used a workaround. Every possible combination of two was listed, so people can use one vote to select two choices.
The intended method of counting is to sum up the total votes for each of the four colors, where each person has two votes. So that’s what I did.
More generally, the “vote for one” / plurality / first-past-the-post system has some major problems when there are more than two options in the poll. It is basically a very buggy system which has a tendency to select winners which don’t accurately represent the preferences of the people.
Instead of using such a flawed system, I generally prefer to use a ranked choice voting system, such as Schulze Condorcet or Ranked Pairs Condorcet. I’ve used those a few times in the past, like for choosing the FW3A color temperatures and emitter types to recommend to Lumintop.
However, I don’t normally use the phrase “ranked choice voting” because most of the time when people say that they’re talking about Instant Runoff Voting… which has some major problems. In practice, IRV tends to be almost as bad as a simple vote-for-one style of ballot. It gives the appearance of being better, since it has a nice ranked ballot which allows people to express detailed preferences… but it then throws out most of that data. It’s not even used. The entire algorithm is optimized to throw out as much ballot data as possible with even taking it into consideration. So the results are typically … not good.
There is another good option though, and it’s much simpler than a ranked choice system. Approval Voting is as simple as “vote for one or more”, which makes it quite easy to implement and understand. And in practice, it almost always selects the same winner as a Condorcet-based system, especially when there are a lot of votes. This is quite different than the current “vote for one” system, which almost never selects the Condorcet winner.
(For reference, Condorcet may sound fancy, but it’s a very simple idea: If a candidate would beat every other candidate in a one-on-one vote, they should win the election.)
An example might help, if the previous post didn’t make much sense.
Imagine Starfleet is electing a new admiral. Five people are running: Kirk, Picard, Sisko, Janeway… and Riker. In this scenario, there’s a lot of disagreement about which captain is the best, but everybody thought Riker was pretty decent.
So, we have about 25% of the population ranking each of the four captains as their first choice, while nearly everyone ranked Riker as their second choice, and the remaining three preferences are pretty random. Let’s say the first choice slot was 27% Picard, 26% Kirk, 24% Sisko, 23% Janeway, 0% Riker. Second choice was 100% Riker. Third to fifth place preferences were otherwise random.
Makes sense, I hope? That’s simply describing how people would fill out a ranked ballot.
The interesting part is … How does each method count this?
In plurality/first-past-the-post/vote-for-one voting, people would argue (rationally, because game theory and Duverger’s Law) that Sisko and Janeway are wasted votes, then fight aggressively between Kirk and Picard. A few still choose to waste their vote though. And Riker, of course, didn’t make it through primaries. He was beaten by Picard of the NextGen party so he doesn’t even get to be on the ballot. The end result would probably be:
- 48% Kirk
- 46% Picard
- 4% Sisko
- 2% Janeway
If the people who voted for Sisko and Janeway had chosen one of the top two instead, the election would probably have a different result. We see spoilers like this pretty frequently.
In Approval Voting, let’s assume half the people marked their top two choices while half only marked one. This leaves us with a victory for Riker:
- 50% Riker
- 27% Picard
- 26% Kirk
- 24% Sisko
- 23% Janeway
In IRV, the first round has no clear winner (first choice > 50%) so the first candidate eliminated would be the one with the fewest first-place votes — Riker. But there’s still no clear winner, so they eliminate the next lowest — Janeway. There’s still no clear winner so they eliminate Sisko. Then the result is highly erratic based on the random rankings people placed farther down the list, but either Kirk or Picard wins. It’s extremely unpredictable, except we know for certain that Riker can’t win in an IRV system.
In Condorcet (any Condorcet-compatible method), it’s pretty straightforward since there are no loops in the top tier results. When paired 1-to-1, Riker beats every other candidate by a 3-to-1 margin, so he wins and everyone gets their second choice. Here are the numbers, assuming for simplicity’s sake there were only four unique ballot results:
Ballots:
- 27% Picard>Riker>Kirk>Sisko>Janeway
- 26% Kirk>Riker>Sisko>Janeway>Picard
- 24% Sisko>Riker>Janeway>Picard>Kirk
- 23% Janeway>Riker>Picard>Kirk>Sisko
Pairwise results:
- 73-to-27 Riker vs Picard
- 74-to-26 Riker vs Kirk
- 76-to-24 Riker vs Sisko
- 77-to-23 Riker vs Janeway
- 74-to-26 Picard vs Kirk
- 50-to-50 Picard vs Sisko
- 53-to-47 Kirk vs Sisko
- 53-to-47 Kirk vs Janeway
- 77-to-23 Sisko vs Janeway
- 73-to-27 Janeway vs Picard
The main criticism of Condorcet is that it tends to choose centrists. With lots of people pulling in many different directions, the average tends to fall in the middle. But that might not be such a bad thing, especially considering how polarized things have been lately. At least it’s a big step up from the first-past-the-post system, which tends to elect the most extreme candidates.
TK, you are truly a Renaissance woman. Thank you for your very interesting replies, which are quite clear. I do not want to continue to disrupt this thread (although it is about polling, after all), so I will PM you a reply.
Chuck
Voted for white and army green!!
1) White
2) None of the above
Just do TK Purple.
……………….Then Blue.
………………………Might as well do ‘Army Green’ too cause it looks so good.
Then call it a day. 
White ’looks’ pretty good but to much work keeping it clean…… 
ymmv
I'm voting Kirk then Picard
But have to admit that I have no idea what the hell you're talking about or why .
imagine this ....Brushed aluminum just like what was originally offered .the very thing people got excited about in the first place .how easy is it not to do a thing and NOT anodize them .
Somehow this choice always eludes people .
\ How much easier is it to polish up without first having to disassemble and strip.
I vote bare naked
You’re right! I’ve been cooped up with bronchitis, and my delirium got the best of me. Post deleted (I’ll send it to TK as a PM). All’s well now. No, really.
Chuck
CLB, I think you set a World’s Record editing this post. 
What was it, 7 or 8 times?? I tried to keep up but got lost. 
Hope you get to feeling better soon!! Take care….
… and I missed it. Oops.
Apologies for the long digression earlier. I have a bad habit of rambling about stuff I’m interested in.
The TL;DR version of all that is: This color poll is flawed — almost as flawed as the polling methods used for big important stuff determining the fate of entire populations. We really should fix that. But fortunately, there are no lives on the line in this poll; it’s probably okay to have an imperfect method for choosing what color(s) the flashlight should be.
Also, the current totals with 222 votes are:
- 134 blue
- 133 purple
- 90 green
- 87 white