Post History

## Two types of popularity contest

I'd like to see popularity contests posted, as a way of measuring how well they fit this community. I want to see what problems arise so we can decide how to mitigate them, and fine tune the rules. In discussing the good and bad points I recommend dividing popularity contests into two categories[^1], which sound similar but are likely to have differing outcomes.

- **Goal focused**  
These are challenges with a defined goal. Votes decide how well that goal was achieved.

- **Constraint focused**  
These are challenges with no defined goal, only restrictions. Votes simply indicate how much the community appreciates an answer.

At first it may seem that a challenge with a defined goal is more objective, but I would argue that it can effectively become _less_ objective, because voters can vote as they wish, disregarding the defined goal.

Similarly, it may seem that a challenge without a defined goal may be more vague and open ended, but I would expect that compensating for this by making the challenge rules very restrictive could lead to a challenge that is more focused. I would regard this as more objective than a challenge with a defined goal, because voters can ignore a defined goal, whereas an answer _must_ meet the challenge restrictions or risk deletion.

Here I'm assuming that the defined goal cannot be objectively measured, otherwise there's no need for the challenge to be a popularity contest - simply make the objective measure the score and votes are not needed.

### Types of constraint
Regardless of whether there is a defined goal, constraints can help drive creativity, and make a challenge not only more demanding but also more interesting. They can also make it more tedious and put people off. Judging what to restrict and how much will be difficult. For example, there may be restrictions on the inputs, outputs, or on the code itself.

When considering what to constrain, consider how this will affect which programming languages are able to compete.

Whatever the restrictions chosen, these _must_ be objective and measurable. The votes an answer receives are decided by human judgement, but whether it is a valid answer must be unambiguously defined, so there can be no difference of opinion over which answers are valid.

### Importance of deleting unconstrained answers
I believe that popularity contests with no defined goal, only harsh restrictions, could do well provided we are strict about deleting any answer that deviates even slightly from the restrictions. Seeing what is possible within those constraints is what makes an answer impressive, and being lenient detracts from this, resulting in a general art contest rather than a specific coding contest. Coding is a specific form of art, and the constraints make the contest more specific still. Allowing exceptions to the rules of the challenge would be like allowing string art in a painting contest. String art is fascinating, but it's a different thing from painting, and makes comparisons unfair.

This is particularly true of constraint focused challenges where the constraints are the sole feature of the challenge, but for the same reasons I also think we should be strict about enforcing the constraints of goal focused popularity contests too.

### Let's try goal focused popularity contests
I'd like to see challenges with a defined goal, as a test of my concerns, but I would expect them to be more difficult to write. I would also expect the answers to be difficult to judge. Do you vote for the answer which technically meets the defined goal but is uninteresting? Do you vote for the answer which adds more features on top but isn't any better at meeting the defined goal? How much does the resulting answer order reflect how well the defined goal was met?

### I'll try a constraint focused popularity contest
I have higher hopes for challenges with no defined goal, and I'm working on an idea for one which will be very restrictive but have no goal other than meeting the validity requirements. I'll link to it once it is posted. I think such a concrete example may make my point better than I can word it.

### Tagging
Personally, I believe that the fundamental difference between challenges with and without a defined goal would justify using two different tags, rather than using "popularity-contest" for both. I don't have a good idea of what those tag names should be. Suggestions welcome. In the meantime I'll probably stick with "popularity-contest". If necessary we can re-tag challenges later.

### Example of the difference between the two types
Consider two challenges:
1. Output an image of a tree
1. Output an image satisfying [measurable constraints]

The first challenge may result in answers capturing the structure, texture, and colours of a tree to an impressive degree, but the highest voted answer may be a kitten holding a leaf in its mouth. Since there is no well defined boundary between "tree" and "not a tree", it will be difficult to decide which answers to delete. That extreme example that isn't even trying to be a tree will probably be deleted, but wherever the line is drawn is likely to be considered unsuitable by some people.

The second challenge does not have this problem, because the measurable constraints can be checked objectively (ideally by a validator program provided by the challenge author). This way every non-deleted answer belongs in the challenge, with no differences of opinion, and votes only decide which one wins (and/or which one wins per language).


[^1]: By "categories" I just mean two different types of challenge, not creating new categories like ["Challenges"](https://codegolf.codidact.com/categories/49) and ["Q&A"](https://codegolf.codidact.com/categories/46).