Poll: Would you like tags implemented on HN
Regardless of whether there's any chance of it being implemented, I'm interested in the sense of the community. Would you like to see tags on HN?
Regardless of whether there's any chance of it being implemented, I'm interested in the sense of the community. Would you like to see tags on HN?
Keep a few to begin with. For example:
startups - news about or related to startups - IOW what this site was supposed to be about initially
gadgets
products - new products from existing non-startups
hacking - related to programming
business
advice
other - stuff that probably needs to be flagged unless enough people like the topic (cooking/food seems to be one seemingly OT topic that people like)
Except that things in the "other" category are not off topic, and shouldn't be flagged even if only a few people upvote them.
Per the guidelines, what's on topic is:
"Anything that good hackers would find interesting. That includes more than
hacking and startups. If you had to reduce it to a sentence, the answer might
be: anything that gratifies one's intellectual curiosity."
http://ycombinator.com/newsguidelines.htmlPersonally, I am opposed to limiting tagging to only a handful of pre-determined tags. But if that must be done, I'd prefer a much wider range of tags. The tags you list are far too limiting.
For instance, I am interested in Lisp, but not in Java. "Hacking" just doesn't do a very good job of filtering these.
I am interested in "gadgets", but don't want to see another iPhone or iPad story for as long as I live.
HN should allow a wide range of tags (ideally completely determined by the users or story submitters).
Someone might be interested in tablets in general, but wouldn't want to see stories on the ipad.
On the other hand, ipad2 is clearly a subcategory of ipad. And someone who is interested in the ipad will probably be interested in stories about the ipad2, even though he may not have explicitly looked for ipad2 tags but only for ipad tags.
Since "ipad" is actually a substring of "ipad2", this could be easily solved with a simple substring search in the tag.
A more problematic example is someone missing a story tagged "java" but not "programming" when searching for stories tagged "programming".
The solution to this problem is to have a tag hierarchy. Thus, the "java" tag would be explicitly made a subtag of the "programming" tag. Unfortunately, the problem with this solution is that it's not easily automatable (though I'm sure there's been a lot of research done in this area), so the hierarchy would probably have to be built by hand. And this, in turn, means that it probably won't be done for a lot of tags.
Perhaps a good compromise would be to require each story to be tagged with at least one predefined tag (from a wide selection of predefined tags that are already organized in to a tag hierarchy), and allow an arbitrary number of user-defined tags.
That way there'll be at least some minimal organization to the submitted stories.
In your example what you really don't want is for people to create distinct tags for ipad, ipad2, iOs, etc on accident. Making it easy for people to select from relevant existing tags means people will only add an ipad2 tag over using an existing ipad tag if that distinction is important and particularly relevant for their post.
Done right tags can be super useful without creating a mess or being overly difficult to implement. StackOverflow does a nice job on this front.
Look at Digg and its naïvely broad topics: they're nowhere near as useful as user-defined tags. A story on a TCP bug is in completely different interest areas to a story on Apple's Smart Cover, yet both are filed under "technology" in Digg.
== Categories and even there i doubt the "might work"
I have heard various people poo-pooh the idea of having tags. But none of their reasons seemed particularly compelling.
One objection is that it would "splinter the community". But this is far from a given, if the default is to keep the front page full of stories unrestricted by tags. Filtering by tags would be optional, and those who preferred not to do so wouldn't be forced to.
Another objection I've heard is that "it would be an admission of defeat". If the complainant had a better suggestion for how to handle the flood of information on HN, this might be a credible objection. But he didn't.
Keeping the status quo is just not desirable for a lot of HN users. Why not offer them the option of a better user experience and at the same time make it a lot easier to search HN for subjects of interest?
http://metaoptimize.com/projects/autotag/hackernews/
For example, here is the topic "conversion":
http://metaoptimize.com/projects/autotag/hackernews/term/7c/...
You can search for tags, and on tag pages you see the most related tags and the most related articles. So you can browse, effectively.
Quality of the autotagger has improved since I ran this demo a few months ago. It's also only through articles through 2009. But it demos the tech and the concept. To what extent are people interested in seeing this updated?
I'm working on a site based on news.arc that has tags. They can be followed, and used to sort stories. http://hubski.com
However, hubski is a general news site. Since HN has a focus to begin with, I am not sure how helpful it would be. I would be interested in a custom feed that follows people on HN though. -I have that implemented on my version.
There are some people on Hacker News, where I would like to keep tabs on everything they submit, whether it hits the front page or not.
HN has focus... in theory.
But even according to the guidelines, what's on topic is:
"Anything that good hackers would find interesting. That includes more than
hacking and startups. If you had to reduce it to a sentence, the answer might
be: anything that gratifies one's intellectual curiosity."
In practice that allows for a very wide range of stories, even when those submitting and voting on the stories bother to read or follow the guidelines.And that leads to the very problem tags are intended to solve.
The tough part is getting tags right. I went the organic route, as I assumed that I would either overlook topics that might be needed, or create redundancies. My site is still too small to see if it works well. I'm not a big fan of the subreddit approach, as I think it divides submissions and attention to a large extent.
But it might be at least partially solvable by offering the user a choice between existing tags before allowing him to type in his own phrasing.
Or, once the user has typed in his own phrasing, a search could be done on existing tags and the user could be presented with a list of matching tags to pick from.
http://www.readwriteweb.com/archives/favikis_social_bookmark...
I can envision a lot arguments like that taking over the discussion.
Slashdot, for instance, has a restricted tagging system for their comments. Comments can be tagged as "Interesting", "Insightful", or "Funny".
I've heard some people on Slashdot complain about the "Funny" comments, but it's quite easy to filter those out via Slashdot's interface. I'd say that was pretty useful.
We should cross that bridge when we get to it. If there is no better alternative to tagging now, this objection does not hold much weight.
How long are we to hold off implementing improvements in hopes that a better solution is found? A year? Ten years? A hundred years?
Remember, there are no guarantees a better solution will ever be found.
And your argument could be used to hold off implementing any feature, not just tags. After all, a better solution might someday be found.
We should make reasonable, incremental improvements like tags instead of waiting for someone to find the holy grail.
I'd rather see an un-vote option.
Commercial Software Vendor User Group Conference:
- in Las Vegas or New Orleans
- Monday thru Wednesday
- vendor presents infomercials disguised as "flights"
- users decide list of enhancements needed to package
- vendor says yes and puts them on enhancement schedule
- enhancements never get done anyway
- everyone gets drunk and gambles or plays golf
Hacker News User Group Conference: - in Mountain View
- on a Saturday
- users get to meet and associate faces with names
- users network and share what they're working on
- users present list of hn enhancements to pg
- pg says no there isn't time
- users get to see Trevor's latest robot
- everyone gets drunk and discusses their favorite algorithmHN users are more likely to meet up given an excuse.
Conferences are normally centered around a series of core speakers- That's what Startup School is all about. The only problem is that it's dramatically limited in attendance.
I would also suggest the application of default tags, such as "n00b" when the article is posted by someone with less than (total_age_of_hn / 5), "anonymous" (with strong flag bias) when the account is less than a day old, "oldie" when it's a dupe that was reposted.
Anyone would suggest more?
I'd like to be able to ignore gadget porn (here's this new iPad cover!) by default, and weigh the programming/science/DIY hacking/design topics higher.
Might be too much load on the server, though, as it will actually mean that every user has his own front page.
If I go to a site like http://answers.onstartups.com, I have trouble scanning the page because of all the visual elements like the tags, the background image, the blue text, the side bar advertising, and so forth. I realize not everyone is as visually distracted by those things as I am, but I really appreciate not having any of those page elements on HN.
Really not sure what tags would achieve over Google and/or http://searchyc.com/
Oh, and even if there is a GreaseMonkey script, displaying reminders is still a good idea.
And how would you like to see that technology applied to HN?
The game of 20 questions demonstrates that there are about a million 'concepts' that are our shared human consciousness. Although a early attempts to catalog these in the 1980s are widely regarded as failures, we've had 30 years of progress in software and hardware since then so what was once impossible may soon be easy.
I'm aware of the existing solutions such as the ruby gem or http://api.ihackernews.com/ but functionality is fairly limited, and it would be nice not to have to resort to just scraping it.
But more so I'd like the ability to curate and follow lists of people that post comments and articles that I am interested in. It seems to me that this idea from twitter could be used to stop Eternal September problems on sites like HN and reddit (the subreddit system is not fully effective.)
There's also http://beta.fedang.com/
I agree that /r/ doesn't quite do it. Some of the communities are just too small to pick up on some very good submissions.
I think last.fm has used this kind of model successfully.
Weights can combat the chaos from users entering anything, but allows for evolution.
Maybe limit the maximum reply depth? (to 4 or something well low)?
Even ordinary web forums are atrociously primitive in comparison, and HN comments are even more primitive than that. The simplicity might arguably be an advantage for very short threads. But for long threads the lack of features can be a nightmare.
At the very least I'd like the HN interface to flag which comments are new since the last time I refreshed the page. That way I don't have to manually reread many dozens of comments in the longer threads while trying to remember which comments I've already read.. and do this over and over and over again as I reload the page.
UPDATE: It's called HN Unread Comments and you can find it at https://chrome.google.com/webstore/detail/fpndmkcfggkffpablc...
I don't think it works on different browsers and it can get annoying when you use multiple machines because the read comments don't transfer, but otherwise it is a very nice tool.
E.g.: Show HN, Ask HN, etc.
http://www.google.com/search?q=site%3Anews.ycombinator.com
As mentioned elsewhere in this thread, there's also http://SearchYC.com
It automatically generated a fairly sensible set of tags from context. Then the user could adjust. They were presented as a comma-separated list -- just edit the list, free-form.
I had thousands of bookmarks, spanning many months, in the thing. And by typing a list of words/tags, I could invariably find what I was looking for in about 3 seconds. I believe it also did partial matching against all tags beginning with a search value; one didn't need to remember the exact tag to generate a match.
I still miss it.
UI's like Gmail's labels are better than a hierarchical, single-location structure. But they are still miles away from the above.
OT: I also want tagging in my file system. My understanding, purely from reading, is that BeOS was one of the few/only to really do this. Anyone have pointers to a current, maintained file system that does this?
--
EDIT: I think I mis-remembered, and that the program presented and used a whitespace separated list. Whitespace separation was, in my opinion, easier to use than e.g. comma separation.
The closest thing I'm aware of is KDE 4, but it's not using a special file system. I haven't used KDE 4 yet but it sounds like it's integrated into the overall desktop system, including Dolphin, the file manager.
It uses NEPOMUK (Networked Environment for Personal, Ontology-based Management of Unified Knowledge). Quick description from wiki: "NEPOMUK-KDE is featured as one of the newer technologies in KDE SC 4. It uses the RDF store Soprano and, on a technical level, allows associating metadata to various items present on a normal user's desktop such as files, bookmarks, e-mails, and calendar entries. Metadata can be arbitrary RDF; as of KDE 4, tagging is the most user-visible metadata application."
The KDE site for NEPOMUK looks like it has some good information: http://nepomuk.kde.org/
General NEPOMUK specifications: http://nepomuk.semanticdesktop.org/xwiki/bin/view/Main1/Deli...
Several of the problems mentioned caused by variation of the same tag, it seems to me, can be solved with related tags. So if I am searching for "string methods" http://ting-1.appspot.com/rt?rt=string%20methods, related tags also shows "python string methods".