[tvural]

"Fix those two problems, and a new search engine could be better than Google. Whether anyone would notice is questionable."

Yeah, I don't think many people will care about the difference between good and perfect. You might be able to find a niche in search that Google is ignoring, but you would have a hard time expanding from there into general search.

In that sense Google is a bet against technology - you would invest in Google if you believe the Web's going to stay the same for a long time and nothing will replace it.

[Animats]

The next direction in search seems to be direct question-answering. Echo (Amazon) is interesting in that, being voice only, it can't punt to a screen of search results. It has to answer the question.

[19eightyfour]

The next direction in search is search that machines can use.

Without needing humans to tell them what the results mean.

[arekkas]

This isn't task of the search engine, it's the task of the machine that uses it. If the machine doesn't understand meaning there is no way the search engine can teach it that meaning.

So no, this will not be the next direction - unless there is a universal model of machines that understand things. But there isn't, and there won't be for at least another 10 years.

[19eightyfour]

That perspective assumes a result-consumers-have-strong-ai version. That's not what i think will happen and it's not the only thing that can happen because i think that misses some pieces of the picture.

IMHO, the other pieces of the picture, and what i see happening, is search engines or a service, will index entities and relationships ( in the ontology sense ), and return results enriched with that data through something like an extended microdata vocabulary. There will be an API where machine consumers can query entities and relationships, and see those attributed to their sources in the webpage. This prediction is nothing new and has been consistently foretold by ai and information retrieval augurs for decades. The difference is that I see this moving beyond the realm of expert systems in large corporations and into the realm of being an API generally accessible to anyone.

I think it's very possible that, as you say, a current incumbent search engine may not consider the provision of such a service its job. Which makes things very interesting for potential new entrants in the semantic search market.

Other business types could surely supply these entities and relationships however who better to be involved in the supply chain than search engines because they hold a hose containing nearly all the world's information, and in the other hand a market hungry for all the world's information.

But this type of service could be niche because most humans are not going to care about getting ontology data in their search results. So even though parsing out entities and relationships could improve the usefulness and accessibility of information there might not be enough universal demand for someone like Google to really care about it.

With these caveats, I think the trend from information, to knowledge, is a very natural and already apparent progression for search engines. The type of API described here is possible today given the right incentives.

[bambax]

The question could be "find the page that has the following words".