[HN Gopher] Show HN: An experimental, people-powered search engine
___________________________________________________________________
Show HN: An experimental, people-powered search engine
Author : zeeshanqureshi
Score : 97 points
Date : 2021-05-28 15:42 UTC (7 hours ago)
(HTM) web link (ninfex.com)
(TXT) w3m dump (ninfex.com)
| zeeshanqureshi wrote:
| Hi HN, I'm the creator of Ninfex.
|
| I've been unhappy with the inconsistent quality of web search
| results for quite some time now. Wanting to do something about
| this for myself, I started experimenting with the way I save my
| notes/bookmarks.
|
| In all of my trials, two things seemed to work more than all
| others and proved to be useful in the long term. One: saving my
| good search result URLs. Two: saving links to discussions on
| those URLs from Reddit, HN, Lobsters, etc. because in my opinion,
| community feedback is a (relatively) better proxy for URL
| quality.
|
| This combined with the Apple notes search feature became a very
| simple but effective personal index. For the main topics that
| interest me, I often found myself searching my index before
| searching the web. Once this setup started working well for me, I
| thought it was probably time to add a feature which lets users
| vote on URLs and host a version of this online, so others can
| both contribute to and benefit from it.
|
| With that in mind, I started speaking to more people who were
| unhappy with the current state of search. Turns out the most
| common workaround people use is restricting results to
| communities (mostly reddit) either by just appending the name or
| using site: operator (example - bone conduction headphones
| reddit). This was further confirmation that the so-called "power
| users" had already moved on to relying on using communities as a
| proxy for their search results.
|
| And so I built a basic version of Ninfex with all those features:
| community-curated index, votable urls, forum links and search.
| It's an early stage prototype and I'm slowly populating the index
| with URLs from my personal wiki.
|
| I look forward to your feedback and I'm open to all kinds of
| suggestions.
| Grimm1 wrote:
| This is awesome, we were doing the same thing for a bit at
| first but ultimately moved away from it in favor of focusing on
| a product more in our wheel house, but I think this has a lot
| of potential to be very big. Very much rooting for you!
|
| One thought we had that was different was to let users create
| communities but auto populate the communities with links that
| we've crawled, to solve the link gaming issue. And then let
| them choose how things would be ranked and curate the links we
| auto populated.
| duncanwerner wrote:
| This looks very much like the old site del.icio.us (which was
| bought by yahoo and discarded). Don't bother going there but
| see [wikipedia][1].
|
| The trick there was that it was originally a "personal
| bookmarking" service, so there was a selfish reason for
| individual users to submit + curate links. Search got layered
| on top later.
|
| [1]: https://en.wikipedia.org/wiki/Delicious_(website)
| zeeshanqureshi wrote:
| Just read through the features of del.icio.us on wikipedia,
| you are right about the overlaps.
|
| I suspect that coupling community + search from the start,
| might result in different outcomes (both positive and
| negative) as opposed to say just community.
|
| Take reddit for example, it's all about the community, the
| search is abysmal. But it's fun to imagine what if reddit was
| mainly about search and the community was a bonus. Surely a
| balance needs to be struck.
| searchnit wrote:
| Interesting project.
|
| One UI nit:
|
| I clicked on Mathematics and the top submission at the time
| was: https://ninfex.com/item?id=sVfpxepq4rNQ
|
| The search UI says: "5 days ago."
|
| I thought it sounded odd since I missed the discussion here 5
| days ago and clicked through to discover the link leads to an
| HN discussion from 2014.
|
| So, that "5 days ago" refers to when it was "submitted" to
| Ninfex instead of any relevant time information on the actual
| submitted material.
|
| Felt a bit misled, and finding relevant current information
| over higher ranked older material is a painpoint for me in
| other search products.
|
| Happens when I am searching for information on a library and
| the most popular hits refer to information from years ago and
| the library from multiple major versions ago. Sometimes the
| answers are incompatible with the current state of the library
| and its API.
|
| These are different issues, but they feel orthogonal.
| zeeshanqureshi wrote:
| These are a valid issues that I will certainly resolve,
| thanks for pointing them out.
| kieckerjan wrote:
| I like the idea very much. First question that comes to mind
| though is: how gameable is this system?
| zeeshanqureshi wrote:
| Thanks for the feedback. Honestly, I'm not very sure myself
| and I'll have to wait and see. The votes could be gamed
| perhaps (multiple accounts etc.), but I could restrict voting
| capabilities to say, email/phone verified accounts, aged
| accounts or minimum karma (or a combination of all those).
| Like I said, we'll have to see how it evolves.
| meowster wrote:
| Aged accounts won't stop anything, and Karma probably won't
| stop anything (afterall a person can buy karma-farmed
| reddit accounts), and I'm not going to give a phone number
| to try the service out. I hope you figure it out, and I
| wish you success.
| puttycat wrote:
| Maybe it is just me, but "people-powered search engine" made me
| think the search queries are sent to humans who look for an
| answer for you, like an operator.
| zeeshanqureshi wrote:
| People-powered as in user submissions, votes.
|
| Sorry to disappoint you :)
| ravenstine wrote:
| Remember ChaCha? It was basically that, but you could also do
| queries via SMS.
|
| https://web.archive.org/web/20100428182234/http://www.chacha...
|
| Now it's a generic Chinese domain parking page:
|
| https://chacha.com
| kevincox wrote:
| While I love the concept it seems somewhat flawed. If every URL
| needs to be manually submitted it is far easier for spammers than
| actual users. Especially if you need a handful of form URLs, is
| this intended for readers or authors? Because a reader likely
| just wants to see it or post it on their favourite form, I can't
| see them going around to a handful. If it is expected that the
| author does it than that is somewhat encouraging spamming forms.
|
| That being said I like the idea, maybe submission could be made
| easier with a browser extension? Or you could crawl popular form
| feeds and become a sort of form aggregator as a form of crawling.
|
| On the other hand how is this different than Reddit? You are
| basically aggregating links, except instead of providing a form
| you are linking to multiple forms. I'm not sure if that is better
| or just an inconvenience. Of course maybe just providing a
| functional search is enough to distinguish you from Reddit.
| ksangeelee wrote:
| I wrote a browser extension that could be used or adapted for
| this purpose.
|
| https://addons.mozilla.org/en-US/firefox/addon/send-tab-url/
|
| The description contains a link to the source code.
| zeeshanqureshi wrote:
| Thanks for the feedback. The forum URLs aren't mandatory. But
| they do add value to search results that are either pages or
| blogs. It is hard to illustrate that value with example search
| queries. I hope that changes with time and a larger, more
| diverse index.
| akomtu wrote:
| The idea has been floating in the air for quite a while and I
| agree that something like this is really needed. What I'd
| personally want to see is a network of anon users hand picking
| interesting pages with a search and a news feed on top of that.
| I'd hand pick all these users and kick out anyone who clutters my
| feed with junk pages. Vouching for pages should be as easy as
| clicking the FB like button (which is blocked by my uBO), ideally
| without JS. Later, when this thing earns some trust, I'd install
| a browser extension. It should be easy to change the anon userids
| while optionally retaining the network I've been subscribed too.
| Saving the network can be as simple as copying userids in my
| list, maybe with some annotations. It should be easy to share my
| id and discover others, e.g. I came across an interesting page on
| Wikipedia, saw that mr4123 liked it and added that id to my list.
| I may want to have multiple userids with different type of pages
| that I wouldn't want to mix together. I'd pay $5/mo for that,
| maybe more if I notice the network brings interesting stuff I
| wouldn't find myself. From a higher level, such service has sound
| economic value: instead of hundred people spending an hour to
| find something, the service lets everyone except one to save that
| one hour. That's why people come to HN, after all.
| ibraheemdev wrote:
| This is a very promising idea, thanks for making this! I was just
| wondering if you were planning on making the site open source, or
| will this a commercial, closed-source product, or somewhere in
| between? Not pressuring you into open sourcing it - it can be a
| lot of extra work - just wanted to know your eventual plan.
| zeeshanqureshi wrote:
| Thank you for the feedback. I haven't really thought about it
| yet. Too early to tell if this has legs.
| altcognito wrote:
| Hey, it's dmoz? I like this user focused version, not good for
| search, but good for browsing. An adjunct to Wikipedia. Probably
| need a lot of guidelines as to what is acceptable.
| mech422 wrote:
| First thing I thought too... "Someone's rebuilding DMOZ?"
| zeeshanqureshi wrote:
| Dmoz and other directories of old were (rigid?) pre-defined
| category based classifications of urls. I don't intend to
| replicate that.
|
| Though you may be right about certain features being similar.
| For now, everything that anyone finds interesting is
| acceptable.
| throwamon wrote:
| This is very similar to something I envision but am too
| lazy/unknowledgeable to implement, so I wish you lots of success.
|
| What about adding a functionality to Ninfex so it automatically
| includes links to posts on well-known communities such as HN and
| Reddit that link the URL submitted by the user? There's a Chrome
| extension called Kiwi Conversations that does something like
| this:
|
| https://chrome.google.com/webstore/detail/kiwi-conversations...
| atweiden wrote:
| Great idea. Reminds me of a previous HN comment [1]:
| How to build the next Google: all good results these days are
| within communities, and Google search has become useless
| for most of these searches. So don't build a
| search engine: build a "rotten tomatoes for X" where the
| sources for each X are "the top N
| subreddits/communities/editorial-sites/forums for X".
|
| I'd be curious to know how you plan on covering costs.
|
| [1]: https://news.ycombinator.com/item?id=24714546
| geocrasher wrote:
| Reminds me of DMOZ. https://dmoz-odp.org/
|
| What is it that makes this a _search_ engine, though?
| quickthrower2 wrote:
| I think it's pretty much the same concept. Blast from the past
| CA0DA wrote:
| similar to https://diff.blog or https://wiby.me ?
|
| Edit: fixed url
| zeeshanqureshi wrote:
| The second one is a parked domain. Diff.blog looks interesting
| though, thanks for sharing. I might be adding it to my daily
| list of sites to visit.
| CA0DA wrote:
| Sorry, mis-typed: https://wiby.me
___________________________________________________________________
(page generated 2021-05-28 23:00 UTC)