Google results seem to be getting progressively worse, and even if they aren’t, there are too many relevant websites to accurately display on a single search results page. It seems to me that a return to the original Yahoo model might actually have a place in 2019.
Is part of what makes google less useful their attempt to "curate"? It's worth thinking about what constitutes curation.
I would say that adapting results towards expected customer preferences (eg giving pet owners and programmers different results for "getting started with python") is curation.
Perhaps more controversially, so too is the pruning of results at all – if most of what I'm searching for is spam/malware, then so be it.
Search engines are already very curated, IMO. There's room to differentiate by choosing NOT to curate. I would be curious to use a search engine that used nothing but PageRank circa ~2006.
Pedantically, any algorithm/ranking/weighting system is unavoidably curatory by design.
But to answer what I think you mean, yes, I think there's room for opinionated search engines for many values of "opinionated". Another way to think of these might be "contexts" that you apply overtop of a search engine, and perform many searches within that context.
For example, when I'm searching for front end things, I wish google were smart enough to time-scope my searches dynamically using knowledge I don't have. I search for Mongoose operations and get top results from 2011 saying X doesn't exist, but it was added in 2016. I would like to see only results from when the thing I wanted existed. But only ever searching the last two years all the time costs results that would be helpful in other times.
When searching for device repair instructions, I wish the engine were able to adopt my estimations about sites that aggregate forum replies or repair instructions in automated ways to gather clicks, and either banish them or loudly mark them.
I'm not sure how to do this with major traction, or without being eaten by google, but "ecosystem of many opinionated and accurate search contexts" feels like it would add value to the world.
Curation is unlikely to be helpful when it comes to a search engine of any decent size, in my opinion.
However, I agree Google's results are getting worse, but I think that is more a result of the way they weight results, and personalize them.
Covering personalization - I've yet to see it work effectively for a large audience. It does drive certain markets, but by whales rather than everyone. So most people get things badly, and a few are driven towards a certain goalpoint, in Google's case adverts, where they tend to make clicks or purchases of the intended partners.
On weighting results - Google seems to have stopped weighting results for relevance a long time ago. They seem to prioritise things differently now. Exactly how I'm not sure, but it does seem to include Recent Events, using technology Google has a vested interest in, and Partners, and all those get addressed before relevancy.
If you can solve the relevancy problem (which is huge, from above, Recent does not always mean Relevant, but it can), then you can certainly attract the technological crowd.
Google seems to have aimed their search engine at popular events, news and social media.
- If there are too many relevant websites to display on one page, search vs. manual directory does not really make a difference.
- The curated model was long kept up by DMOZ, which had an army of volunteer editors but eventually shut down. Some editors took over and built curlie.org, but there is no innovation since DMOZ (which itself had no innovation)
- Curation is a big topic nowadays. Pinterest for example is based on that idea. However, things might be a bit more decentralised these days. I am just thinking about the awsome-* lists on GitHub.
- Manual directories may fail build a "searchable" taxonomy. I am searching maybe for a technical support company specialised in that product, located in this city, offering low rates. This is an easy example, but there are tougher ones. General purpose information retrieval (like Google does) might be a better fit for such problems than prebuilt taxonomies.
- Based on the last two points, curation might be more successful in small topics rather than in a Yahoo-like all-topics setting.
Pinterest is a completely parasitic website that doesn't provide backlinks to content, and hides results from google search. No attribution and no original content. That is completely disgusting to me. There is no way to get to the original post. It honestly wouldn't surprise me if they were hotlinking too.
For sure. I think there is both demand and it's possible to scale it. We have examples of amazingly successful group curation at scale with wikipedia, stackoverflow etc. We also have forum platforms that successfully curate information, Reddit, HN etc (which many people use to find better recommendations than Google).
However, the value in curation is only on a particular subset of queries. Those where you want to discover information and have some introductory terms to go on ("Motorbike Safety", "Analytics SAAS", "Buying a Fridge" etc). What youre looking for is an expert in a niche area to point you in the right direction. Curation is very beneficial here.
The second type of query, where you are looking for a specific piece of information (specific business, specific book review, weird fact) etc Google already does fine on. And curation wont help (nor will it scale on such precise info).
I'd say so. Google is a joke, it doesn't give relevant results anymore and limits itself solely to 'pop culture' results. There needs to be 'contextual search', i.e look at your previous queries and factor that into your current search.
I don't work at Google, but considering the amount of research scientists on the search team and the troves of data they have on users, I think they already use contextual data quite heavily.
Anecdotally, if you type a google search for something along the lines of spring break in Florida, then search Google Flights, it auto-populates the dates of Spring Break into the search.
I think there is room for 30 - 50 curated niche search engines - done similar but different then the old dmoz and yahoo.
In so many ways google is failing many niches that it once served well. I have been doodling ideas on how to do some of those niches better, like adult.
Curated and Niche. Google is still great for general searches in my opinion but when I need a drill down level of search that can only happen with curation, we have a gap there.
I would say that adapting results towards expected customer preferences (eg giving pet owners and programmers different results for "getting started with python") is curation.
Perhaps more controversially, so too is the pruning of results at all – if most of what I'm searching for is spam/malware, then so be it.
Search engines are already very curated, IMO. There's room to differentiate by choosing NOT to curate. I would be curious to use a search engine that used nothing but PageRank circa ~2006.
Pedantically, any algorithm/ranking/weighting system is unavoidably curatory by design.
But to answer what I think you mean, yes, I think there's room for opinionated search engines for many values of "opinionated". Another way to think of these might be "contexts" that you apply overtop of a search engine, and perform many searches within that context.
For example, when I'm searching for front end things, I wish google were smart enough to time-scope my searches dynamically using knowledge I don't have. I search for Mongoose operations and get top results from 2011 saying X doesn't exist, but it was added in 2016. I would like to see only results from when the thing I wanted existed. But only ever searching the last two years all the time costs results that would be helpful in other times.
When searching for device repair instructions, I wish the engine were able to adopt my estimations about sites that aggregate forum replies or repair instructions in automated ways to gather clicks, and either banish them or loudly mark them.
I'm not sure how to do this with major traction, or without being eaten by google, but "ecosystem of many opinionated and accurate search contexts" feels like it would add value to the world.