That is an area that I need to improve. With a high focus on privacy, it makes it hard to track what users read without creating a user tracking system and that is something I really want to avoid doing.
The thread links use the CSS :visited selector to show the links in a different color to help identify the links you have previously clicked without the site knowing that information itself.
There is a history function for logged in members, but for user privacy it can be limited or disabled by the user via the Settings page and is also auto-purged periodically.
At the very least the site needs to show content that has been added since your last visit and that is something I plan to implement soon.
The problem is not really the methodology or implementation but the perception caused by it. If you can tell a user "we know you read this" then that implies you are tracking what they read which is something I try to avoid.
I plan to work on a solution, but if I do, users will be able to opt-in or opt-out depending on the implementation.
I really try to avoid using local storage because it delves into potential user privacy issues. I will look more into it as well as other options as I know having a list of already viewed results would be helpful for users.
My intention behind content diversity is not based on political interests, but rather a desire to offer a greater level of exposure for lesser known content and blog sites.
By "diversity" I mean that a developer who writes good articles on python whose blog only gets a few users a day will be treated the same as content from high traffic sites.
Unlike other news services that either have political leanings or are beholden to their advertisers, I want to provide an independent service where people can decide for themselves what they read or do not read.
The sun, daily mail, Fox News are “lesser known”? Not to mention tons of other stuff I’m too afraid to even checkout. Low quality and fake news is not diversity.
No, of course not, but if I only included lesser known content in the reader and search engine then few people would use it since they would have to look elsewhere to find content from the major news sources.
Major news sources will always get more overall exposure simply through quantity because they publish so many articles every single day. However, low traffic sites and blogs will get their content promoted at or near the top of the search results when it is indexed in the same manner that the high traffic sites do.
I understand there are sites you dislike which is why I added an Upgrade global filter feature where you can block sites, topics, keywords, etc. that you do not want to see. Other Upgraded users can do the same when they do not want to see content they dislike as well.
Promoting low traffic sites that actually publish content that’s worth while finding out about is admirable. But if your front page is an “open sewage” of the worst of USA and UK tabloids, I don’t think anyone who’s interested in quality content would bother to register or upgrade.
I posted it in the beta notes on the site, but forgot to add it to my comments here.
The Popular search results are currently supplemented by a variety of hand-picked source feeds because the site is new and there is not yet enough traffic history to generate an accurate list of popular content, but that will change once more people start using the site.
Of course people will not be happy with some of the sites selected if the target site has a political bias that they disagree with, but it is only temporary until traffic starts to grow more on the site.
I already have it built into the engine to ignore the supplemental sources once the activity level hits a certain level.
That said, it will still come down to the users. If a lot of users are viewing specific content, it will likely find its way to the home page.
On a side note, you can change your default home page via the Settings page to avoid the popular search results.
The site supports user subscriptions for advance reader and searching functionality, so my hope at least for now is that people will use the free version and at some point want higher limits or more functionality and then upgrade.
I have a lot more planned for it long term (everyone always says that, but it is true) such as an API service, but for now, I am focused on improving the reader service and adding more sites to the index.
I'm blind, so can't comment on the color scheme. However, screen readers depend on semantic elements like headings, list items, and so on to navigate the page. Currently the search results are not organized as any such entity, but are a few links that are related visually. The graphics are not marked as such, lack alt text, and an example how one of them reads is:
"9c064f09d669c4a7ec0da2bb54c16a…"
Setting aria-hidden might be the easiest way to deal with this particular issue.
"Popular" and "newest" are not marked as control elements like buttons, combo box, or a radio button so they look just like plain text to my screen reader. It is experience that suggests that hitting enter on them will produce any result.
The search button is not labeled and it reads just as "button". What you call a quick bar (a guess) is a line of buttons represented as links but with no text inside them and no apparent function.
PS: Great idea for the site btw. I'm bitching so much only because it seems as something that I might use and wish to have as comfortable as possible.
I have worked really hard to provide as many options as possible for people, but it is hard to anticipate everything people will want or need from it until they use it and tell you so feedback like yours helps.
I have added everything you wrote to my notes from this thread so I will see what I can do to improve them for you.
No results found is actively optimized against, which raises recall and reduces precision of search results Via synonyms, stop words, etc. the priority of a search engine is to deliver you results because those are actionable even if not fully relevant.
I agree and I like that meta search engines offer fuzzy search matching, because sometimes you know what you want but are not quite sure of the best keywords to use to find it.
I just wish more search engines offered an option or default setting where search results would only include exact matches.
1) Are you sure you cannot change the font and and site theme without logging in? I just tested it and it works fine for me in private and incognito modes.
I purposely designed it so a lot of the settings are available via the quick action bar for non-logged in users.
If you have no-script active or have disabled javascript, then the only option would be to login to change them via the Settings page.
Otherwise, you should be able to toggle those easily via the quick action bar above the search results.
2) Definitely. I probably should have already done that.
3) I am still thinking on this. My plan was to soon remove it for upgraded accounts. There is no user-specific tracking, but by anonymously tracking clicks in general it helps with ranking and generating "popular" search results order.
Most users have JS enabled, you could just track clicks making an Ajax get call with a click listener on all links before navigating if you don’t want to use something like GA.
The Settings page requires that you login, which only requires an email address.
The quick action bar on the search results page also has a smaller cog icon and other icons on it that will allow you to change several settings without logging in, but you must have javascript enabled to access them.
The thread links use the CSS :visited selector to show the links in a different color to help identify the links you have previously clicked without the site knowing that information itself.
There is a history function for logged in members, but for user privacy it can be limited or disabled by the user via the Settings page and is also auto-purged periodically.
At the very least the site needs to show content that has been added since your last visit and that is something I plan to implement soon.