You would need to have a certain number of people to participate to create any statistical significance. Any statisticians around who can enlighten us with what should be the minimum population needed to get as relevant as possible poll results?
Only have some training/experience in statistics and polling (from a sociology and marketing standpoint), but do have a decent understanding, so I'll give it a shot. There's already a few erroneous assumptions about how polling or how statistical information functions in this thread and there's no easy answer to your question, so I'll just try and give a very brief (very simplified) rundown, which may have some errors (it's also 1am here :P).
The number of people required for a poll is based on (though not always limited to) population size and desired error rate/confidence interval and confidence level. Error rate is margin of confidence (expressed as a percentage) and is considered repeatable via probability/confidence level ('x' times correct out of 'y, or as a percentage', is sort of like standard deviation). And it means when you get a result for a question, it is likely to be correct within that error margin (up or down) from that number, as frequently as the probability. E.g. An answer for a question that gives a result of 50%, may have a range of 47% to 53% in a survey if it had a 3% error margin, and would be stated to be correct 90% of the time, or 9 times out of 10.
Determining confidence level is complicated and hard to apply here, but if you want to make it more accurate, really what you'd want to do is run the poll multiple times (not necessarily on the same people) and ensure the population is as diverse as possible (so, not only founders, who are self selected in the first place). And make it as accessible as possible (not buried in a specific topic, on page 6, since there's even more self selection there).
You'd also need to decide on a margin of error that's acceptable. Frequent increments used are 10%, 5% and 3%. But, the thing is, the required response rates scale with population, so it's not a simple ratio, and the confidence interval is also effected by how divided the answers are (e.g. extreme answers are more confident than 51/49 splits). So, if there are 1000 Founders, you assume a confidence level of 95%, and you randomly selected 500 of them to send the poll to, and 88 answered (decent response rate is between 10 and 20%), with 50% selecting one of the answers to your question, you could reasonably say the range would be 40-60% of Founders believe/feel/agree with that answer. If there are only 100 Founders, with only 50 people responding, with 50% selecting one answer, that would also give a 40-60% range.
On top of that, one of the key things is to ensure the question is correctly formed and the answers allow for a representation of opinion. The current poll doesn't do either of those, I'm afraid. The question "Planet Design / Art Direction" is too vague and combines multiple broad topics into one, which means each person responding is going to interpret it differently (even if they've read the context of the whole thread). Answers are inconsistent in word usage (love versus like/dislike), there's no neutral response (neither like or dislike) and no option not to answer (which is actually useful for a number of reasons, like gauging how much this topic matters to the population, or uncertainty, etc). Though, at least the multiple vote option is disabled.
Not sure if that helps at all, but there's some (remember, very simplified) info.
All that said, even the worst formed polls may be more accurate than reading through a forum and trying to perceive a response/feel out the community directly, simply because the human mind is so full of cognitive bias and glitches that there are still, for example, people out there who genuinely believe the world is flat.
StarControl Forums
Stardock Forums
Please??