Search chatbots? Pah, this startup's trying on Yahoo's old outfit of web directories

Web search, long dominated by Google, is in play again, at least among incumbents and entrepreneurs if not frustrated web searchers. It's not just that Google Search has by some accounts got worse, though Google maintains its results are still better than other search engines. It's that both Google and Microsoft are betting …

[Doctor Syntax]

"For years, the fields of information science, information retrieval, and human-computer interaction have studied how people make use of search systems and have investigated ways to improve search interfaces. The integration of generative AI chat components is a major development that may profoundly change the way users interact with search systems."

The people being studied can only make use of what's there and can't make use of what's not there. If what's there doesn't offer decent filtering facilities which respect logical operators such research can't show people making use of them. If the research shows that what people are using tries and fails to double guess what they want then obviously it will point to using AI to make even worse guesses.

[Michael Wojcik]

While this is true, there's plenty of human-subjects research being done with uncommon and experimental information-retrieval systems. It's not like people in those disciplines (and we should throw in library science, too, plus cognate research in fields such as cognitive science and psychology, and interdisciplinary fields like digital humanities) haven't ever tried anything that's not public and widely-used.

I've yet to see a mechanical information-retrieval system that, for precision and recall, beats a good reference-desk librarian. Mechanical systems are great at speed, scale, and breadth, which makes them ideal for large numbers of easily-answered queries, where either accuracy is not particularly important, or the correct answers are so widely available that the system has a very high probability of returning them. But trained humans still do better on the hard stuff.

And no, LLMs are not going to replace those soon. That's a really difficult fine-tuning problem. Nothing I've seen published on fine-tuning so far has any hope of coming close.

[Roj Blake]

Information Retrieval

Wasn't Information Retrieval Michael Palin's department in Brazil?

[Mike 137]

"Google maintains its results are still better than other search engines"

Or more realistically, they think other search engines are even worse. Do they include all the engines that are effectively proxies for Goooooooooooooooogle, or only the independents? And BTW, how do they quantify search engine quality?

One of my most important criteria is whether a query returns relevant results, and another is whether these are buried or not in irrelevant garbage. On both counts Gooooooooooooogle scores pretty poorly. For example, when I specify a search phrase in quotes (supposedly indicating I want the phrase as a whole searched for) I do not want masses of results based on individual words dragged from the phrase, and when I search for a specific word I do not want results based on all possible grammatical variants of that word. I want what I bloody well asked for (or to be told it can't be found.)

[Doctor Syntax]

Re: "Google maintains its results are still better than other search engines"

Google's metric seems to be simply the numbers of hits. "Nothing found" might be the true result but by Google's metric it's not an acceptable one.

[DS999]

Re: "Google maintains its results are still better than other search engines"

Which is why when you make a search that's too narrow Google will "fix" it for you by ignoring terms you included, or including results with terms you told it to exclude. Because they get paid by clients for specific search terms, even if they are served to people who find them useless.

[Martin Gregorie]

Re: "Google maintains its results are still better than other search engines"

I haven't used Google for over a decade: basically ever since Duck Duck Go launched.

[DS999]

Re: "Google maintains its results are still better than other search engines"

I use DDG 95-99% of the time. In the rare cases when I can't find what I'm looking for with DDG I will retry the search with Google. More often than not I can't find it there either.

The difference is really stark in the "shopping" category. DDG has everything neatly organized, with a lot of results shown per screen. Yes, the "ads" are shown first, but are clearly marked as such. By contrast on Google the images are larger so you get far fewer per screen, and as far as I can tell it is 100% "ads" i.e. people paying for placement.

[EricB123]

Re: "Google maintains its results are still better than other search engines"

The only time I don't think DDG is better than Google is for, err, um, porn. Maybe that's where Google got it's "quality" metric from.

This is just a hypothesis in a scientific experiment, mind you.

[DS999]

Re: "Google maintains its results are still better than other search engines"

I would bet Google is collecting a LOT of money from buyers of porn related search terms, and in order to justify those huge payments those paying it would have to provide exactly what you're asking for. They can't show you their ads if you don't stick around on their site, after all.

DDG/Bing isn't collecting those giant wads of cash and must rely on the traditional methods to determine the site. When probably every porn site has hidden text for every possible search term from milfs to midgets to things we don't want to know about because we could never unsee them, it is gonna be really difficult for it to figure out how to index which site provides the best content for a given search.

[Martin 47]

Re: "Google maintains its results are still better than other search engines"

DuckDuckGo’s browser seems pretty good too.

[Roopee]

Chopping's is the first intelligent analysis of web search, and the economy that has evolved around it, that I have come across - thank you El Reg for an excellent article!

[katrinab]

Google was not a “first mover” in search. They were actually pretty late to the game.

So they didn’t have first-mover advantage. What they did have was the advantage of seeing where everyone else was doing wrong, and they fixed that and mostly cornered the market, except in China, Japan, and Russia.

We are at the stage now where Google has got so bad that someone else could do that again.

[DS999]

Google's founders figured out the optimal formula to index sites

That's why they beat out everyone else. If it wasn't for it being so easy to game that optimal formula they'd still be using it, and now that the PageRank patent has expired everyone else would be able to use it too. They'd all provide great results.

Unfortunately PageRank only works if no one knows that's how you're indexing sites. Once they knew that, they could alter the sites to game the algorithm and improve their site's rank. It has been a game of whack a mole ever since, with Google trying to tweak their algorithms to work around the SEOs and the SEOs striking back.

That's probably the biggest reason that DDG is basically as good as Google despite a fraction of its resources. The SEOs have made web search so useless that pretty much everyone is at the same mediocre quality, and that's unlikely to ever get any better (though I'm sure before long someone will be marketing "AI based search" that's supposed to be way better by working around SEO optimizations in real time, and instantly be given a multi billion dollar valuation by the VCs)

[Anonymous Coward]

> Suddenly there will be no incentive to produce content

Finally. SEO, die b*tch!

Would only the true enthusiasts keep producing content as in the good old days?

[alcachofas]

Interesting piece.

I’m not at all sure that this guy’s approach is going to be a sustainable and widely used one but it’s good to see people attacking the problem from a different angle. There’s way too much groupthink right now - we get bursts of innovation and then just a truckload of people all doing the same thing.

[Bebu]

Most sensible thing...

《El Toco uses machine learning on the backend to classify the pages it crawls and label them so they can be filtered more efficiently.》

This has to be the most sensible thing anyone has uttered concerning AI/LLM and by a bloody economist. :)

Really going back to an acquisitions librarian preparing catalogue index cards for the library's resources.

Back when you could actually efficiently locate quite specific information using the card indices and a bit of human intelligence.

Then the only "conversations", at least in a smaller quiet satellite university library, might be an attempt to "chat up" the librarian which were I recall quite attractive contrary to the usual stereotype (today this would drop you into so many cauldrons of hot water.:)

I recently searched for "how does a floating moon light work?" The dogs' breakfast of results omitting those offering to sell one were utterly useless. When you ask how to make one the arrant nonsense returned is mind numbing - admittedly some were clearly the product of ChatGPT's etc total lack of a handle on physical reality.

A decent, documented query syntax in front of a database of the filtered and fine grained classified crawl results with an efficient query processing engine would save me a lot of wasted effort.

I shall follow with interest the future endeavours of the "The Touch" search engine. (Nice choice.)

[EricB123]

Dogpile?

Does anyone else remember a site called Dogpile, which basically returned a poorly organized summation of all the major search engines of the time?

Honestly, I'm not sure where I'm going with this. It just popped into my head while reading this article.

[Roj Blake]

Re: Dogpile?

There was also an app called Copernic that collated results from multiple search engines.