Ask HN: What's the best feature you've built that no one uses?
Also curious to know the probable reasons for the failure.
Recently deprecated http://msdn.microsoft.com/en-us/library/ms537634(VS.85).aspx
Ironically now showing up in other browsers own extensions.
Reasons for failure:
* distrust of Microsoft inventions in the browser
* incomplete implementation strategy based on javascript instead of custom expression parser
* security issues because of using javascript instead of a custom expression parser
Always thought of them as a halfway point between pure CSS and Netscape's idea of implementing stylesheets in Javascript.
* Terrible performance - practically first rule for optimizing page rendering is "don't use CSS expressions"
http://code.google.com/speed/page-speed/docs/rendering.html http://developer.yahoo.net/blog/archives/2007/07/high_perfor...
* Only work on IE natch.
So, you have Javascript which was better performance and worked everywhere.
I think it's a little misleading to say it's the best feature no one ever used. CSS expressions were re-born as hardware-accelerated CSS3 transitions - with a much better, more focused implementation, and are pretty popular these days - especially for mobile.
Probable reason: some people would rather perform their own appendectomy than read a manual.
Seriously, I have often included features on strong user request (MUST! HAVE!) against my own better judgement, only to find that, surprise, they are not used. I don't necessarily regret it, there is a certain advantage to looking responsive regardless of the request.
Reason for failure? Google doesn't index search results...
I'd love to hear ideas for making something like this indexable. Generating a bunch of quality content around search results seems difficult (see: Mahalo), but maybe there's another approach I haven't thought of?
Just brainstorming here:
- maybe the hook could be: "judge books by their covers"
- the layout is pretty slick but constrains what you can do with the site. Have you considered scrolling down? It could be a differentiator though (your current layout that is).
- When you mouse over a book, have some type of overlay function: star the book, "have you read this book? what did you think?" (leave a comment), share on twitter / facebook (a necessary evil perhaps)
- when you click on a cover, it lightboxes in and has the things listed above
- make it like a reddit, but for covers
- people who searched for x also searched for y
It's basically all user-generated-content, and it actually does rank on Google for a few things, but it's the wrong kind of traffic: People aren't in buying mode, very few visitors actually click through to Amazon and buy something.
These are basically A/B tests.
1) Show the last 200 searches that people had made (or the most popular for the past day, week, etc.)
2) Instead of directly linking to Amazon upon clicking a book/product, perhaps you should fly out some hidden details for the item (not ajax if feasible).
3) You could also have a "Search by Author" and "Search by Topic" link that is provided when you click a book that provides more results with content.
4) Maybe allow people to create book shelves that they can explain the grouping of a set of books and why they go well together (etc!).
5) One last idea: Provide a list beneath that has a tiny bit more info - as they scroll left/right in the top list, have the bottom scroll up/down to show the range that is visible. This would help with you only linking/showing 20 or so results at a time if you let the list (text) show 100 or more.
Regarding Google indexing the site, was that toungue-in-cheek? If not, you may want to consider some basic changes to the structure - I think its less that Google doesn't like your site and more that there's very little there to index.
And maybe you could show a little more information about the book when you click on it. I'm not sure what data Amazon provides - there must be more than just the picture, no?
Like another commenter pointed out you will need to create static pages for the search engines to find & index. You can do this by creating them automatically from the last x number of searches, if you were able to aggregate some other sources into the engine you could create custom landing pages showing all results rather than people clicking straight off to Amazon & having no dedicated page for each book title.
How low has the traffic fallen? If you can't get search traffic, maybe there are some other traffic sources?
Usually horizontal scrolling fails miserably, but at bigbooksearch it "feels" natural.
I actually got the web slice accepted into the IE addons gallery (http://www.ieaddons.com) after waiting a couple months for approval. Then I made a small revision to the entry, and for some reason that made it go back into the "pending approval" queue. I sent several emails to the website support contact and got no answers. Months later, Amazon came out with their own extension. I finally got a response from a Microsoft rep after escalating with a friend, and they told me about how my extension can't be posted due to a partnership agreement. Pretty lame.
Made a site in college to sell a T-shirt for $1,000,000. Didn't work. But I did make $40 off an MIT professor who bet me I wouldn't try it.
I use it constantly everyday as I built it for myself. Judging from the download stat, not many people using it.
pop-global-mark (C-x C-SPC) moves you around a global mark stack, and C-u set-mark-command (C-u C-SPC) does this locally in a buffer. (I use this hundreds of times a day.)
The bookmark package (go to "Bookmarks" in the Emacs info pages) give you persistent bookmarks, with a fairly nice interface.
I've written stuff in elisp only to find that it already existed in Emacs, too. Now, I usually check http://emacswiki.org first.
Global mark (or local mark) doesn't work on buffers that have been killed. When I jump back to a quick bookmark, I want its file loaded back into memory if its buffer has been killed.
Global mark doesn't work consistently. It seems to only remember the last mark set in a buffer, so it has an one mark per buffer behavior. The flow seems to use global mark to jump back to a buffer, then use local marks to jump to different points in the buffer. I want just one quick bookmark system, no global or local distinction, to remember multiple points in a buffer and multiple buffers.
Mark is kind of overloaded for different purposes. Its primary use is for marking regions (for copy/cut/etc). I turn on transient-mark-mode to better highlight regions. Whenever I made a mark for quick bookmarking, the region is highlighted and have to be canceled; it's very annoying.
The Bookmark package requires assigning a name for each bookmark, entering its name to jump back to the bookmark, and deleting the name when it's not needed. It is good for permanent bookmarks but very cumbersome for quick bookmarking. I just want to set anonymous quick bookmarks and jump back to them quickly. I don't want to manage the names.
The good thing about Emacs is that if you don't like the way things are working, you can roll your own. So I did.
Then again, even when implementing something that (it turns out) already exists, solving the problems that come up along the way almost always teaches one something.
And with Emacs extensions, even if you're the only one who finds it useful, it's still useful. :)
BonjourFoxy[1] adds Bonjour support to Firefox and is predominately used to find websites advertised by printers, network cameras and so on, but it can also be used by other extensions to advertise and discover network services.
Currently the only extension I'm aware of that makes use of this is iGiro[2] which combined with an iPhone app let's you scan bill numbers straight into Firefox's text field.
Now admittedly I haven't got any killer P2P extensions in mind, but surely some other folks out there do —?
Turns out we had an audience whose blogs were composed of copying lyrics and changing colours. You can guess not many really used it.
Oh well. SP went out of business a while after so I never had a chance to do what the audience really wanted (a like button)
Based on the terribly original idea of writing about your travel experiences. A couple good features (imo):
- Inherent credibility. I figured if you ask someone you meet on the street where's a good place to eat or stay - being a complete stranger - the chance of them giving you good information is about the same as flipping a coin: 50-50. So everyone starts with 50% credibility. As you write reviews and people are in agreement (via like/don't like or comments) your credibility goes up (influenced of course by the credibility of the commentor).
- Built in messaging and chat. Added some "bots" to make it look like there were actually more users at one point. The idea was to implement a parser so that you could ask someone in chat: "where's a cheap place to stay in guam?" the app would parse the query and a bot could answer with the highest rated hostel in that location for example.
- Parse search terms with ability to search by category. So if you search for "a bed to crash on" while you're in manhattan, the app prompts if you'd like to search the "lodging" category.
Reason for failure:
- Other people did it better (tripadvisor, virtual tourist, etc)
- SHOCKWAVE (although this was 2005)
Basically it is a timer for jobseekers that runs in the background, to push them to spend a longer amount of time hunting for a job than they normally would (even if it isn't on our site). People have always known that the biggest difficulty with job-hunting is spending enough time on it, but it is really demonstrated by the this 2009 NYTimes interactive graphic: http://www.nytimes.com/interactive/2009/07/31/business/20080...
Despite some bloggers writing about it (http://helpmegethired.wordpress.com/2009/06/), it never took off. I'd say the failure came from the execution. It needed to be dead-simple to turn on (w/o requiring an account), and always present yet unintrusive. Maybe the next version...
Why did it go unused (I wouldn't say it totally failed)? The biggest reason is probably because no one wants to use your site as much as you want them to. If they couldn't find what they wanted on my site right away, why would they stick around when they could just go look elsewhere?
That's why I ended up building the "Around the Web" feature, where if a rental couldn't be found on RMSR, it will aggregate and map properties from other sites (around the web). This feature was mildly more successful.
Quick and simple to make, thought I could make some money on referral fees, but went nowhere.
Failure, I suppose: browsers can do the same thing, with a little setup; and, more importantly, people seem to have a preference to navigate to each of these sites and search them directly.
Would be fairly good to be able to search Wikipedia by typing w: in the URL bar.
Preferences -> Basics -> Manage (Next to Default Search)
nobody searches with capital letters, so hitting the shift key to make a ":" is a pretty big cost. Just use "g <search>", and detect prefixes you know; it's much faster and less awkward to use.
I think I should have just released that one feature as an MVP.
http://college.mychances.net/tools/college-choice-matrix.php
This tool is based on ~25,000 college applications and shows you the relative preference for each college. It's not based on direct matchups but instead based on an Elo point system, so that all colleges can be ranked along one dimension (essentially, revealed student preference).
It's rarely used because it's really hard to find on the site. It's also tough to figure out how to choose the schools that show up, instead of just looking at the defaults. It's also possible that college applicants care way less about this sort of thing than I, a processor and curator of their data, do.
I would have loved a tool that that would let me enter a GPA and SAT score and then using actual data could tell me that 500 students with comparable stats (like +/- 0.1 for GPA and +/- 50 for SAT) applied to school X last year and 380 were admitted.
But I like the feature because its based on AppEngine and XMPP, so it doesn't have any costs and will probably be available forever. (forever on the internet means sth. like 2-5 years)
I did have a website that pulled multiple RSS feeds (thousands) and then parsed each entry matching it to others that it had seen before. It could then show you what was popular at any moment on the web, with the ability to see what the article was linked to, and what they were linked to. It showed some interesting things like how a website was effected by something happening on the world stage.
The idea was you could find what was hot on the web at any one time. Nobody really showed any interest in it though.
Reason... I didnt want to pony up the money to put a live version on the web due to not knowing how the producers of the RSS feeds would fee. So there was never anything more then a demo concept.
If I understand what you are saying then I don't think there would be any. It sounds like a neat project, where can we take a look?
Should be how they would feel about it.
Its somewhat discontinued. I may spin it up again someday, as everything is there minus the database with data. If I ever do, I will post it on HN for people to have a look.
EDIT - I will have a go at spinning it up tonight actually. It should be pretty easy to do in theory.
Chicken/egg problem with reaching critical mass. My plan was/is to grow it organically with long tail keywords once Google starts indexing all the pages. But they haven't done that yet because the site doesn't have enough incoming links to warrant indexing 1000 pages (another chicken/egg problem :)
1. The entire http://www.choip.me website, I built it because I was sick of seeing ads for twitter scams on other tweet longer sites, so I just threw together a Twitter + Disqus mashup on appengine to do the same thing, and provide a better way to take discussions "offline" on Twitter. Not bad for basically a weekend project I think.
2. The custom search portals on www.unscatter.com. Of course I'm still working on it, the interface for creating them is very rough. But the functionality is there. A couple examples of what can be done with it are the Devsearch - http://www.unscatter.com/search/unscatter/devsearch/ and Conservation Search - http://www.unscatter.com/search/unscatter/conservationsearch...
They initially loved it, but when they found out it couldn't be integrated into their Wordpress blog, they dismissed the completed project. So I guess the reason for its failure was lack of input.
The project didn't take that long to create and was made to be more of an advertising tool for my company.
http://crossfitter.ca (although you can't see that much unless you belong to a signed up gym)
I always wanted a site like this to discuss software process things without the hardcore evangelists using buzzwords. It didn't take too long to throw this up so it wasn't a massive waste but I actually thought it would start growing once I shared it with my developer/agile network. Not much interest, apparently. I'll probably have to kill it soon so it doesn't sit there mocking me.
http://jetpackgallery.mozillalabs.com/jetpacks/55
I'd say it didn't take off because it requires the user to have firefox, install jetpack, use igoogle, and be annoyed by all that wasted space etc :)
I wouldn't say it failed. I use it at work and home, and I love it :)
Freelance work btw, I don't run a affiliate marketing site myself.
I could publish it but I would need to modify lots of thing to make it intuitive for new users, write docs, and basically just to make it "public".
None of those sound like prerequisites for publishing something. Is the functionality there, and is the tool modestly resistant to things like SQL injection? If so, why not put a link out there (like, say, in this thread)?
The take-up is probably limited by the fact that my documentation is erring towards the minimal.
That and the erroneous dependencies I forgot to remove until last week. :)
I thought this would be a competitor for http://www.fmylife.com/
Not just any feature, the entire site is something that none uses
Only show me Tweets that don't include:
Photos URLs RTs Hash Tags
Lesson learnt !
probably it's not visible enough on Google or it's just me who finds song similarities interesting :)
Most of them don't have an idea what the data means. It's over their heads, so we missed the mark audience wise.
This is probably our failure, not theirs. We can't expect our audiences to understand something the way we understand it.
http://jetpackgallery.mozillalabs.com/jetpacks/55
I'd say it didn't take off because it requires the user to have firefox, install jetpack, use igoogle, and be annoyed by all that wasted space etc :)
I wouldn't say it failed. I use it at work and home, and I love it :)
There just isn't much to it. I'll add a bookmarklet when I get around to it.