There was a brilliant article by an occasional contributor about monopolies and how they aren't always bad for the consumer.

In particular, Google was mentioned.

The TL;DR was that having a monopoly means a company is forced to keep innovating, or it will lose out to that do.

Google chose not to innovate (which is synonymous with invest) and decided to milk the goose.

"Not a tear, no not I" as Bono once sang.

Yup - we had a speech on this a few years ago: We need to maintain our momentum else we might as well close our doors and look for other work. So it's not just the tech industry: It's all industry.

The problem with monopolies is their competition is so far behind, it's easy to become complacent, and with that you lose momentum, and face stagnation. Meanwhile the competition has a clear path to follow, and can more easily build momentum and so will quickly catch up and pass the leader.

Flip side: If there's close competition, there's a heck of a lot of motivation to keep innovating.

The problem for the monopoly or any big business is that all those small companies snapping at their heels have a lot less to loose. Taking risks on new innovations is easier when you are small. If you have 50,000 employees going bust and starting again is not as easy as if you have 50.

Governments deciding what technology they want to succeed is even worse. It distorts the process of innovation and industries fail, just look at the European and US car industries as consumers' desire to buy EVs to meet ZEV targets has failed to materialise.

"The problem with monopolies is their competition is so far behind, it's easy to become complacent, and with that you lose momentum, and face stagnation."

Stagnation would have been OK. It's the enshittification that comes with prioritising vendors' self-interest over customers' requirements that's the problem.

Google did innovate... to improve short term revenue. They innovated so far in that direction that they created an opportunity for others focused on search result quality to gain users. Users are the product, not search results. When Perplexity has sufficient users to sell they will have to make choices about how much quality to sacrifice to get the ad revenue required to satisfy investors. Perhaps they could increase ad revenue by trimming down the LLM generated answer while leaving the links to source material in place. When that is not enough they will follow Google's example and we will be looking for Perplexity's replacement.

...using your Google account!

Oh, it will give me what it "thinks" is the answer I'm looking for and - very kind of it - it's sources. Nope. When I search I'm not looking for an AI double-guessing engine. Search has already become a welter of double-guessing engines. It's not a summary I'm looking for, just the sources. Judging something like that to be better than Google would only be "better" in respect of added enshittification.

This may be a novel idea (spoiler alert, it isn't) but how about a search engine that just takes search terms with the usual operators of and, or and not, and gives the results that fit including the null result if nothing fits. Just like Altavista used to AFAICR.

OK, a quick test: "coal pits mentioned in wakefield manorial rolls"

Yes, it comes back with a short summary and a few snippets. At first sight the snippets look as if it's been at work summarising things until I look at the sources.

What would you expect the sources to be for that prompt? A good selection of WMRs are on archive.org so there's no problem getting that answers is there? Except the sources are just secondary publications, True at least one of them involves material not on archive.org (AFAIK 1402 isn't a published roll) but clearly the snippets are either just bits of human summaries or, even worse, AI summarised bits of human summaries.

Maybe I'll go back and try a prompt that can only be answered from real sources.

I did that:

Prompt: "how many fines were issued in published wakefield manorial court rolls for digging coal but not selling it, digging and selling it and just selling it"

Answer: "Based on the available search results, it is not possible to provide a precise count of fines issued in the published Wakefield Manorial Court Rolls for the specific categories of digging coal but not selling it, digging and selling it, and just selling it. However, the search results do provide some relevant information about coal-related fines in the Wakefield "

It's perfectly possible. It's just a matter of going through and keeping 3 tallies.

How much money did you pay The Register to carry your advertisement as an "article"?

I asked it to write a biography of me, and it pulled together various bits of information to come up with something not bad at all. But you can ask ChatGPT and other LLMs to provide citations for their claims, which is an easy way to fact check. However I bookmarked this particular one...

Stuff it - at this point, maybe we should go back to the 90's and have a curated Searchable directory, A la Yahoo.

Or if we want to open source it, maybe it's time for the re-emergence of Web rings...

In the good old times, google would let you hunter and gather all over the internet, so you could prepare your own meal.

With time it increasingly pushes on already prepared stuff you are not interested in.

Now with AI, the "solution" is to have a machine eat and digest the food for you, leaving it ready to be shat directly in your mouth... Is this progress?

I searched for me with Perfidity or whatever it's called. I've had an exciting life seemingly and my passion for off road quad bike customisation is monumental. PMSL.

"how is perplexity's free search funded"

"Perplexity's free search is currently funded primarily through investor capital. The company has raised significant funding, including $165 million as of 2024, valuing it at over $1 billion. This investor funding allows Perplexity to offer its free search product while developing revenue streams."

Otherwise, the results suggests the usual hopeless hopes: Premium subs and Advertising. And then they will go the way the other search engines have gone, with the results sorted in descending order of revenue, not relevance.

The problem with AI search is that when I use a search engine I'm not looking for it to give me an answer - most of what I search does not have a single, definitive answer. I don't want a summary of several web pages. I want a list of relevant pages to look at myself.

It seems the entire AI industry and even Google itself does not understand this.

I searched my name and "caver".

It did guess my LinkedIn profile - but copied bits wrongly (although it did helpfully(!) list what STEM in "STEM Ambassador" stood for)

It listed all the books I have written - no I have not (there is an author that I share part of my name with)

It listed caving clubs of which I am a member - except the one that I am actually a member of

It invented me a political life - I was an independent Chairman of a Parish Council, but none of what it invented was true

When I tired to find the sources of the information, it told me that there were no sources

I just tried it for a typical query I would make in google “best Indian restaurants in the city of London”.

Good points were lack of ads cluttering the page (this will change when they need to make money) and I like that it shows the sources it used.

Bad points were the sources it used were solely the websites of Indian restaurants, all of which claim to the best restaurant wherever they are located, so it picked five random restaurants.

I found google’s results better as it included links to rankings on websites like tripadvisor and timeout.

When AI was trumpeted. It's useless.

I agree Google is now poor to useless.

Perplexity is small enough not to be targeted yet. But at the end it's an unethical engine that presents harvested text from other web pages. If I ask which OLED TV is the best, it'll extract the information from comparison sites and present it. Eventually those sites will block it or they'll die.

You can't just make a web page that shows information from all other internet web pages. Perplexity never compared TVs, so its results are practically stolen, even if there's a reference. They're providing a service based on other people's work without paying loyalties.

Traditional search engine keep the quoted texts at minimum for this reason. I expect the lawsuits to start sometime next year.

I asked it for tonight's winning lottery numbers but sadly it didn't provide those, just the results of the last draw.

Incredibly amusing to see Ed Zitron quoted in an article advertising an AI search engine, given his strong opinions about AI.

Just tried it (again): it did not have the right information, either pages not indexed, or query mi-interpreted.

You can't even tell why, since the haphazard parrot will always produce an assertive answer that is trained to look right.

I'd much rather prefer "sorry, no results" or "I interpret your query as ..."

Aside from that, totally agree about Google's enshitification.

I switched to Brave.