[HN Gopher] What I learned from running a concierge search engine
___________________________________________________________________
What I learned from running a concierge search engine
Author : researchers
Score : 43 points
Date : 2022-03-28 16:59 UTC (6 hours ago)
(HTM) web link (re-search.xyz)
(TXT) w3m dump (re-search.xyz)
| sk55 wrote:
| Very cool! This reminds me of crowdsourcing research done by
| Aniket Kittur as part of CMU's social computing lab.
|
| Here's the relevant research paper: https://joe.cat/CHI-ka/
|
| Looks like a very similar process you landed on except their
| approach has many workers crowdsourcing the answer together.
| data_spy wrote:
| There is a legitimate start-up that basically is outsourcing
| tasks to a digital personal assistant but they log all questions
| and their researched answers. It make take a while but they will
| have a moat within a few years with a really robust repository.
| brandonhorst wrote:
| The problem is that this then becomes a cache invalidation
| problem. Many of the examples presented will be outdated in 1
| year, and completely wrong in 5. How can you know which answers
| need to turnover and when?
| researchers wrote:
| That's definitely one way to do it. Our approach is not to
| literally run a concierge search engine, but rather build up
| intuitions on how humans effectively find knowledge on the Web
| and then build tools to make that easier.
|
| Is that the end-game of the startup you mentioned?
| malfist wrote:
| I think this is a brilliant idea. Couldn't figure out how to sign
| up for the service though
| researchers wrote:
| Unfortunately we are not running the search concierge service
| indefinitely. But we will translate the findings into a search
| tool that makes it easier to answer complex questions. Sign up
| for our mailing list or shoot an email to
| outresearching@gmail.com . We will let you know when its ready.
| hudsonz wrote:
| Are you me? We created a very similar service prior to building
| our e-commerce search (shopdeft.com). The results have been
| fascinating!
|
| https://concierge.shopdeft.com/
| groby_b wrote:
| Having worked with personal assistants virtual and IRL: There's
| only a limited scope for that. Many of the more complex questions
| can't be answered without a large amount of context - something
| the concierge won't have until they've worked closely with you
| for a few months.
|
| There's nothing wrong with targeting the market that is OK with
| context-less (or context-poor) search, but it's worth finding out
| how strongly you're positioning yourself against the somewhat
| overcrowded "virtual assistant" market, and how you'll
| differentiate.
| vineyardmike wrote:
| Really cool business idea, but i came across this line, and i
| think the conclusion is wrong...
|
| > While I can't share all the queries, they tended to be starting
| points in a larger journey. People asked for product
| recommendations to start a new hobby, or for evidence to support
| a career transition.
|
| The conclusion the author mentioned is:
|
| > Since these queries had no singular answer, no single document
| would suffice and no algorithm could perfectly rank the content
| to deliver a tidy answer.
|
| I agree - BUT - I think outsourcing this is the wrong approach
| for the people asking the question. If you're starting a new
| hobby, you probably want to gain the context from researching the
| topic. They used the example of kayaking. In the process of
| researching a good kayak, you'll likely come across domain
| experts and their blogs, talks, etc that can share the WHY as
| well as the answer. This may lead you to gain new insights for
| your hobby. The other example is starting a career, and i think
| the same conclusion applies that if you want to become an X, you
| should ideally learn other people's view on X and the day to day
| tasks, and growth opportunities, and what makes a good X, not
| just "steps to become an X".
|
| Maybe sharing the citations and research is the answer needed to
| these questions - like a real research paper.
|
| Things like "romantic getaways near SF" or "traveling to hawaii
| with kids" are things where you just want an answer, not the best
| answer seem like a better fit.
| paxys wrote:
| I'm struggling to see how any of what the author mentioned is
| actually automatable. If you can invent an AGI, sure, but
| otherwise the problem can only be solved by human intellect, and
| good luck making that a business for $10/month.
| wolverine876 wrote:
| Related - I would pay for a search engine of authoritative data.
| I mean one that only uses the most authoritative sources, such as
| reputable research, leading journalism (NY Times, etc. - and the
| news, not the editorial side), reputable reference sources
| (Britannica, Mathworld, etc.), textbooks, MDN, Microsoft
| knowledge base articles, etc.
|
| That would save me so much time - and maybe change the world. In
| fact, it really could be and should be free. Beyond possible
| access to DRM'd info, such as the textbooks, it's just search
| engine for a limited curated set of sources.
| WinterMount223 wrote:
| I wouldn't feel full or satisfied with such research done by
| other on open ended questions such as What kind of inflatable
| kayak should I buy? Maybe after researching, I realize that
| really I should get a rigid kayak instead. Such research is
| iterative, interactive, and some of my preferences or personal
| knowledge can't be easily externalized.
| verve_rat wrote:
| Sure, but then you submit the next question to the concierge
| service. There might be follow-ups and tangents, but you pay a
| flat fee under this model, why wouldn't you have them do some
| follow-up for you?
| drewcoo wrote:
| So when this person delegates something, we know it's not
| important?
|
| If that's really the case, it's probably better not to tell
| anyone.
| dzink wrote:
| Tell that to librarians before the internet.
| researchers wrote:
| In our sample, we found that people really cared about those
| topics but they were not urgent. Hence the reference to the
| Eisenhower Matrix.
|
| The current state of search tools doesn't make it easy to
| answer complex questions so complex questions go unanswered. By
| lowering the activation energy to do an activity, the market
| for that activity increases. That doesn't mean that activity
| didn't ever matter.
| researchers wrote:
| Hi HN! So to clarify to everyone, the idea is that what a
| concierge search engine would do, a machine could do as well.
|
| sk55 shared this very interesting link [0] which outlines how a
| complex question could be broken down into micro-tasks and
| assembled into an answer. The interesting thing here is that each
| of those steps is now within the realm of automation by
| Transformer-based models. With the addition of a UI that
| encourages exploration/refinement of the query, it could be a
| good way to quickly gain an understanding about a broad topic.
|
| This is one of the many theses we are exploring. Another one is
| that "A Google Replacement Will Not Look Like Google" [1]
|
| [0] https://joe.cat/CHI-ka/ [1] https://re-
| search.xyz/writing/mapping-the-new-world-towards-...
___________________________________________________________________
(page generated 2022-03-28 23:01 UTC)