Suggestion: Improving post findability
I was reading a discussion about this on here earlier today, and I've already lost it! 😬 Sorry if this is just creating more noise, but we clearly need better ways to find content. The search feature will go a long way, but here are some other ideas:
Tag search. On any topic with tags, the tags should be clickable links to URLs like https://tildes.net/tag/elder+scrolls. This page would show all topics that use that tag, with sort and filter options. There should be a way (maybe built into the search form) to type any tag and jump straight to this page.
Recently viewed topics list. Reddit shows a sidebar listing the last n posts you viewed. It's admittedly a little creepy seeing your history displayed like that, but it's a useful way to jump back into conversations for follow-up later. The old Reddit design had a "clear" button to delete the history, but curiously that is no longer present in the redesign. (Privacy features like that should not be overlooked here.)
Saved topics. Another feature from Reddit. Every post has a "Save" link below it, that adds the post to your personal saved posts list, which can reached from your profile. Saved post lists are only visible to the users that own them.
Repost detection. I really like how Ask MetaFilter helps posters make sure their content is fresh before they publish it. The submit button under the new post form is labeled "Preview" and clicking it shows what the post will look like before publishing it. This gives posters the opportunity to proofread and ensure their text formatting is correct. More importantly, the site scans the content of the post and displays a list of five possible existing posts that match it:
The following previously-posted questions might be related to the question you're asking. Please take a look before posting to see if any of these answer your question.
This flow adds an additional click before you can actually post, but I think it's for the best. The slowdown politely nudges you toward considering the quality and originality of what you're about to say, without being overbearing. The main MetaFilter site also checks all URLs you enter to see if anyone has posted them before. Note that these tools don't prevent anyone from posting, they just empower users to avoid reposting and reinforce good posting behaviors.
Repost flagging. I have a half-baked idea about allowing users to flag topics as reposts, but I haven't seen this implemented before. This would be separate from voting. A user wishing to flag a topic would be asked to provide the URL of an existing topic it duplicates. This wouldn't affect the topic itself, other than to add a small banner to the top of the page: "n users flagged this as a repost of the following topics: [list of links]". Then anyone would have the ability to [agree], [disagree], or append a link to the list. Public consensus would affect the future of the topic... if enough others agree the topic is redundant, it could be auto-deleted or just algorithmically prioritized lower than non-reposts. If enough disagree, the flag could be auto-removed from the topic. The usernames of the flaggers should be public, and there should be a way to view both a user's frequency of flagging and whether consensus agreed with those flags. Accountability would be important for this sort of system.
Ability to subscribe to users. I saw the other feature request for a "friend" mechanic. I agree with the commenters who said it would be too much like a social network. However, I could see a use case for a "Subscribe" button on a user's profile page, just like the ones on group pages. This would cause all topics posted by that user to be included in your main page, even if they are in groups you aren't subscribed to.
I'd be interested to hear your feedback on these suggestions, as well as other ideas specific to increasing content visibility.
Thanks @balooga, good ideas and I appreciate you taking the time to write this up. I won't comment on every item, but some quick notes:
I'm hoping to have tag filtering (both "show only tag X" as well as "filter out tag Y") fairly soon.
The "recently viewed topics" could probably be tied in to the "mark new comments" feature, since by enabling that, the user is specifically opting in to allowing the site to track which threads they visit and when, so it could be used for a recently viewed list as well.
I like the thoughts about repost flagging, being able (or required) to attach it to the specific links it's a repost of would probably work well, and make it easier for people to find the previous discussions.
Tag searching would be a really cool feature. The "recently viewed topics" mechanic I would also love to see implemented, especially considering the big open space on the right of the website right now, could fit very snugly in there. Also agree with saving specific posts, another very useful feature.
I like the idea that before posting possible posts that have already been made are presented to the user, as long as the interface of this mechanic is not super intrusive but still catches the user's attention so they at least consider what they are posting has already been posted. Overall I really like this idea though.
I like the idea of users determining whether a topic has already been posted and can therefore collectively remove a post. This could always be manipulated, but I have faith that the users of tildes would use this mechanic properly to prevent repeated discussions. I am not sure how the post should be treated once it is determined to be repost however; deleting it seems to harsh however keeping it up makes the fact it has been identified as a repost redundant. Some middle ground may need to be established here.
Finally, I do like the ability to subscribe to fellow users, especially if you find a user who has the same interests as you or you simply find their opinions interesting. In terms of their posts being mingled in with all the normal posts on your main page, I think it would be better to simply have another tab that lists your subscribed users posts chronologically, essentially bypassing the need to click on each individual user that you are interested in. Instead all their posts are listed underneath one easy to access tab.
Excellent suggestions and ideas by the way, definitely want them to be implemented at some point in the future!
Flagging as a repost should as the others says add links to the old topics.
However there should be a secondary mechanism for determining if the thread in question still deserves to live, such as if the old threads never got a good answer or there has been new developments. In this case, the old threads would also link back to the new one.
I put up a new official invite request thread on reddit this morning so most everyone who can probably respond to this post in depth is probably going to be busy the next few days (myself included). But I will try to address some of the issues in a quick and dirty fashion (if you don't mind).
Tag search - My understanding is that tags are going to be an integral part of search (a basic version of which is already in development). However I do really like your idea of the permalink style multi-tag pages (similar to multireddits) method so I added it to the issue tracker as a "suggestion".
Recently viewed topics list - Good idea. Added to the issue tracker as a "suggestion".
Saved topics - That is likely coming as some point too and is already in the issue tracker as a "suggestion".
Repost detection - That is for sure coming eventually using canonical link comparisons and custom parsing for sites that don't support canonical links (e.g. YouTube). How reposts will be handled isn't exactly decided yet AFAIK. But @deimos has mentioned potentially adding a system like reddit's "other discussions" feature.
Repost flagging - How reports are going to be handled hasn't been decided yet either AFAIK. However adding the ability to report reposts is not a bad idea once the report function is in place, IMO.
Ability to subscribe to users - This is already in the issues tracker as a "feature" (for sure coming eventually) as well.
Sorry for the quick and dirty reply but hopefully this helps.
p.s. Be sure to check the ~ development board to get a sense for what else has already been suggested and what features are coming down the pipeline in the future.
How do you plan to manage the tag database? If tagging is unrestricted as it is now, an enormous amount of tags will accumulate in a short time, with a lot of misspellings, redundancy, irrelevancy, etc. It would be a nightmare to clean it up again and again, Leaving things be, however, would degrade the searchability via tags.
I believe part of @Demios' plan is to allow people with higher reputations to correct tags. These people will be able to correct tags to the canonical variant, which should prevent most of the misspellings and redundancy.
As I said, that idea doesn't scale well, and it can be quite arguable in what way to merge tags. Missing tags would also break existing saved searches as suggested above, and eventually scripted ones.
A synonymous tag database (e.g. ‘rnb’ = ‘r&b’ = ‘rhythm n blues’ = ‘rhythm and blues’, etc.), tag standards/guidelines/conventions for trusted users to follow, redirects for synonymous/missing/merged tags, etc. Management of tags can still be made to work without being unwieldy at scale and without breaking saved searches.
You mean only trusted users will be able to generate new tags or are referring to the review process?
Manageably scaling a tag db is naturally possible, but tildes isn't specialized site like stackoverflow, so preparatory work on design and presentation is particularly important. Nothing in regards to the planned system has been documented, however, that's why I'm asking.
I was referring to the the review process but basic guideline can also be provided to users who submit content so their tags are more likely to conform to the standards. Also keep in mind that the way tags work now is not how they will always be. When submitting content the plan is to have most of the metadata automatically filled out for users using specialized scrapers and public database APIs (e.g. thetvdb for ~tv posts, themoviedb for ~movies posts, etc). What's in place for tags now is just a very rudimentary implementation to help give @deimos a feel for potential problems and how best to design the system going forward.
Another thing to remember is that ultimately tag editing will be tied to the trust system as well, so trusted users will be incentivized to keep the tags well maintained and rewarded for doing so. Whereas abusers of the tag system will lose trust and may eventually lose the ability to edit tags at all.
And nothing is written about the tag system because not much is written about much of anything yet. The site is still in Alpha and has only been running just over a month now. It'll come eventually though. :)
Well, it wasn't a judgement on what you should've done by now, but a mere description of the situation and thus the cause for my questions.
Anyway, automated, db-based pre-selection will certainly reduce a lot of noise, but it is the accumulation of missing, mislabeled or orphaned tags and groups of tags I'm talking about.
Basically, I was wondering whether you're going to establish a system which doesn't require the tags of every submission are checked by at least one user--for instance, by limiting the selection to existing tags--and what tools there will be in order to help with the (re-)organization of tags and the semantic links between them (synonymous lists and such).
edit: grammar
You would have to ask @deimos about the technical aspects of the plan for tags and tag management but I assume there will eventually be a healthy dose of automation (pre and post user input) in the process. He is the creator of automoderator, after all.
And from a human perspective, all I can say is that I regularly work with giant corporate databases that are often unsanitized and/or completely broken (e.g. CSV files with commas littered throughout all the ‘user input’ fields) doing datavis and analysis for consulting jobs. Manually sanitizing, cleansing and restructuring them to get rid of the noise and user input errors at scale, while admittedly a PITA, is more tedious than it is difficult or time consuming, especially thanks to regex and specialized tooling designed for just that purpose.
If there was an intermediate “preview” step before a topic is published, like the one I described above, it could be used to suggest preferred variants of tags similar to what the user entered.