"I don't recommend this option"


#1

What do you guys think about adding the ability to “not recommend” an option?

I like it due to:

  • More data to rank options
  • great opportunity to show a dropdown of the cons to get “why don’t you recommend this option”
  • You can see what people considered. IE out of the 8 options they picked this one and didn’t like those 2 due to X and Y.

#2

I was just thinking about this. It seems very useful, but VERY dangerous. There already seems to be some brigading in action in a few questions, this would compound it. I think the right solution is to steal an idea from Hacker News: only accounts over a certain karma threshold can downvote.

This would work very well with the idea I brought up on Slack a few days ago, to alter the color of a user’s icon border after a certain threshold. My motivation there was to make it easier to see who’s a newbie needing interaction, encouragement and feedback; but this makes the border change into a real functional badge, so it can be less subtle and could help encourage contribution.


#3

What if we don’t initially punish the ranking, but just allow them to do it to get more cons and con votes?


#4

If we do punish the ranking, we could weigh each downvote based on the number of options they downvote, as compared to the total number of options at the time - that way, if someone downvotes all options but their favorite, it will dilute their downvotes. If we want to be harsh, we could make it logarithmic.


#5

sure, we can alleviate the downsides by being clever, but do you think the upsides are worth it?


#6

I like:

downvoteValue = (weight - votes) / (1 + weight)
where
    votes = numOptionsDownvoted
    weight = numOptionsUpvoted + numOptionsTotal

And definitely think the upsides are worth it if abuse is factored out. This turns it into a tool against brigading.


#7

@johnlbevan or @actionless have any thoughts?


#8

I think the ability to “not recommend” an option would be fantastic.

For example, I’m thinking it would be very useful for questions with few data. So the few users who are contributing to the question can explicitly not recommend certain options based on their experience and this can help users see which option is more recommended.

Of course, it will be also be very useful, if not more, for questions with lots of contributors.

I also like that it would be a great opportunity to show a dropdown of the cons. This would help get more data for the cons.

That would be really helpful too.


#9

Another potential upside: it would allow hiding options below a certain threshold by default.

My first concern was with people effectively using it as two votes. Tom’s suggestion would work around it, but I see other issue arising from that - what do you display as the vote/recommendation count? If a vote is diluted to effectively be .1 of a vote, does it show 30.1 votes? Or is it obfuscated in some way via some sort of “Slant rating” (which in and of itself is not a bad thing)?

Also, I don’t particularly like the idea of dividing people into different classes of citizens (the borders thing). Especially in early days. Recognizing someone who has been a long time community member or pointing out someone’s a mod is a different thing, but I think it would discourage joining a community if it required arbitrary hoops to jump through before allowing someone to become a “true” part community due to elitism it could quite plausibly create. Joining a new community is hard all on its own. It may make sense at some point, but it’s something I’d rather avoid if possible.


#10

Functionality. The idea of having downvote functionality for an option is good - it makes the cons more valuable / helps to balance items which are popular but bad; this gives experts some means to correct this; which will be even more useful if you can filter to view votes of those you follow.

Affect on Score. I don’t like this negatively influencing Karma though; to me, having an option marked as bad is much more valuable than not having that option i.e. it shows that this option was considered and argues why it isn’t popular; that information’s useful. We’re not saying that someone’s given incorrect information; just that they’ve provided a valid option which isn’t recommended.
Equally, this helps avoid issues of time affected data; i.e. if I say the Nexus 7’s the best Android device out there, in 2 years I’ll start getting downvotes; I shouldn’t lose points for giving a good answer which has become stale. The alternative below may be a good option for avoiding that issue…

Food for thought: An alternative approach would be to allow people to rank the options; i.e. drag them in order of best to worst, leaving some on the side if the person doesn’t have an opinion (e.g. if they’ve not got first hand experience / feel their insufficiently informed to rank an option). This then gives more info than a simple yes/no, or even than a 0-5 star rating.
The only downside is it doesn’t say if something’s leagues better than another option, or just mildly better; but maybe that’s not a problem, or maybe a secondary mechanic could provide that information.
The big upside is this allows new options to easily bubble to the top; an item added 2 years after the most popular can quickly jump to the top of the list if a small number of people say it’s the best and no one rates it poorly.

RANKED

  1. Notepad++
  2. Programmers’ Notepad
  3. Windows Notepad

N/A (not ranked due to lack of opinion / knowledge)

  • TextPad
  • Vim
  • Sublime Text
  • Emacs

Humanising Slant
#11

i think it will be very useful. especially i see it like a full contrary to ‘recommend’. ie it will be nice to have two columns for each rank item, in one column will be number of recommends and pros sorted by number of votes and in other column it will be number of unrecommends and sorted cons

and as @tejon proposed it can be some new sorting of ranking items (while they will have both positive and negative rank) but also it will give some new ways for analysis/sorting of items – for example similar to way how comments sorted in reddit when u’re select to sort them “controversial”


#12

I think this would provide the most useful and accurate data. The downside would be UI complexity from what I initially imagine the UX to be like. The current “core” flow of Slant is “I recommend this option, here is why”. It’s a single button press that’s pretty intuitive, followed by the dropdown that our data shows is very effective getting people to contribute. The rank options UI would require a pretty significant rework of that flow. Not saying we can’t do it, just it would be a big project and would replace the core flow of Slant. Perhaps we could work it in as some sort of advanced voting UI, maybe in the compare view.

That would be so damn useful.


#13

After going through a lot of content, I’ve noticed the importance of allowing users to either not recommend options or to let options get buried over time if there isn’t a constant stream of recommendations (something like reddit’s “hot” filter).

I’m thinking about the content a year from now and wondering how new options will compete with well known options that have already gotten a huge amount of support. In order for new options to move to the top, each new product must have more votes than the one before it. The purpose of Slant is to recommend the best. If old options don’t eventually make way for new options as technology progresses, we’ll just be recommending what has been most popular over time - not necessarily the best thing on the market at this moment.

Another way to approach this problem is to rank the best options relative to each other based on concrete qualities shared by each group. Maybe each pro and con could be linked to a more general attribute of the option, and each vote on pros and cons will influence the option’s rating in that attribute. Some of the attributes that options could be compared by could include privacy, battery efficiency, cost, user friendliness, speed, stability, compatibility, and so on.

To summarize:

  • The ability to not recommend an option is important for relevant material
  • Options should be compared on shared attributes instead of random pros and cons.
  • If options are compared based on shared attributes, it will be easier to tell the difference between quality and popularity

There’s still a ton I need to learn about Slant, but these are my thoughts from what I’ve seen. :smiley:


#14

Thoughts on this for the button design?


#15

Looks nice, will there be a toast notification that says “not recommend” or something along those lines?


#16

Will the recommend count include downvotes, or will they be strictly used for list ranking?

For an indicator I’d suggest inverting the arrow button (white arrow on red background) and either putting a red slash across the number if it’s recommend count only, or turning the number red if the count will include downvotes.


#17

Do you mean a tooltip? Like something that appears on hover? If so, yes.

We have no idea how people will use this, so we kinda just want to put the button there for now to trigger the "why don’t you like this option’ dropdown to get more cons. Once we take a step back we will figure out if we want to show how many people didn’t rec something or/and use it in the rankings.

Do you think it’s intuitive at all that the count shown is just for recs, and isn;t impacted by downvotes?


#18

I think if there’s a “Recommend” button, the opposite should be a button of equal weight. A downarrow is vague. “Discourage” or “Dissuade”; something like that. To emphasise that it’s not a popularity contest.

I think it’s pretty easy to misinterpret that, unfortunately. Or, at the very least, be uncertain.


#19

“Recommend” and “Disparage” with separate counts?


#20

Eh I had to Google “Disparage” to double check the definition, so lets rule that one out as too clever.

The design constraints are the more difficult ones here than what word to use. The current design somewhat requires the rec button width to be constrained to the image thumbnail width. This is why the don’t rec button was iconised, it’s mainly a space thing as well as we want to make “recommending” a simple/clear primary action and giving equal space to "don’t rec’ takes away from that.

the other option is a "do you rec this? {Yes}/{No} with “3/5 people rec this” but I don’t like the visual feel of it. The current design has a nice & simple primary action that I don’t want to ruin with this feature.