Fixing our data structure

We’ve had an idea to make Slant an order of magnitude more useful. Before I get into the idea some context.

In my current conceptual analysis “comparison data” can have the following attributes:

  • Comparability or uniquness
  • Qualitative (needs descriptions) or Quantitative (just numbers) or Boolean (yes/no)

This is manifested into the following types of data that have the above attributes:

  • Comparable Qualitative Data - Descriptions of build quality across laptops
  • Comparable Quantitative Data - Price
  • Unique Qualitative Data - Features like “any do moment” that are completely unique to a particular option.
  • Unique Quantitative/Boolean Data - Outlying specifications like the cheapest or the lightest or the only option with github support.

On Slant we have two data types.

  • Pros/Cons - Designed for qualitative data
  • Specs - Designed for quantitative and boolean data

Obviously we’re missing some nuance in our data types.

We currently do a passable job with:

  • Comparable Quantitative Data - Specs!
  • Unique Qualitative Data - Pros/Cons are great for these.
  • Unique Quantitative/Boolean Data - These outlying specs make decent Pros/Cons. A better system would tie the pro/con to the quantitative spec so you can confirm that none of the other options have that attribute. Our system breaks pretty hard when the data changes too (another option adds Github support). That said, our current way of handling this is passable IMO.

We currently suck at Comparable Qualitative Data. Our pros/cons aren’t directly comparable across options, so we’re doing a terrible job with this one. Try going to “best 2d game engines” and compare the editors of all the different engines. Try going to “programming keyboards” and figure out which has the best build quality. Super important to help people make a decision and we make it very hard.

Some thoughts about this:

  • If Specs taught me anything, it’s that properly structuring the data cleans it up a lot. By moving all the “Is cross platform” Pros into the spec table it made the Pros/cons so much better. The same is going to happen when we fix comparable Pros. All the comparable pros will be places together and this will let the unique pros really shine.

  • “Comparable Pros” will be compared against a set of attributes that belong to the question. For example “Best 2d game engine” would have:

  • Editor

  • Performance

  • Support

  • Ease of use

  • Doing ^ would theoretically let us award options with “Best editor” algorithmically. This would stop all the silly “really easy to use!” titles that are effectively randomly added by users. However if we do this we’re going to need a better way to rank Pros than just vote counts.

  • This is most likely going to be achieved by tagging Pros/Cons with the question attributes and then grouping them.

4 Likes

Great explanation! I absolutely agree on what areas of where the site is currently strong and effective. In addition to that, I feel it’s also showing the site’s improvement (in several topics, but not all) the more it matures. I believe that this is key in how much it can potentially impact the site overall, good or bad. I don’t mean to be so blunt, or rather stating the obvious, or even worse as simply way off topic on what’s being addressed.

These concerns in particular being raised is a lot to do with what caught my interest in the site initially and liked how well and unique it approached the purpose for these type of sites. Both making it separate in the rules and format in how it rates the topic being asked (pros and cons, use of comments, and the format of the topic in how it’s stated, i.e. what are the best … for … ).

I suggest even making this into a main category itself, and creating fixed or static discussions on each part of the structure being outlined. For example, feedback and input on details on the strong points, such as the

and

Just for starters. But as I meant by being static or fixed, these will be created only by @Stuart, for an example.

What do you think?

I was wondering about another idea I’ve wanted to mention and thought it might be a good time to bring up if you agree it can help think about this topic. Since it’s very different to my previous post, I felt it was best to separate.

I am curious to know more, if possible, particular to the total number of users that corresponds to how active they are. One example, could be a chart of number of users and how that compares to how active they are, which includes the common span of activity. Are more users active when they firs join, then gradually become less active, or are more users generally active the same as the others, whereas the factor of how long they have been a member isn’t relative to their activity (participation).

I think this can be relative in understanding certain areas of the data structure in regards to what to modify and how to go about it, as well whether some of it should be modified at all. the data structure is statistical data on the user base.

So, basically add tags to questions for grouping up Pros/Cons among options so we can easily compare and rank them?

Suggestion status: approved.

Next status: in construction.

:stuck_out_tongue:

That’s exactly right. Just need to figure out how the UI for that would work.

Status: In construction.

1 Like

That looks fan-tas-tic! :slight_smile: :stuck_out_tongue_closed_eyes:

1 Like