Apple rejected Bing for wrong answers about Annie Lennox | Hacker News Reader

Apple rejected Bing for wrong answers about Annie Lennox

theregister.com

71 pointsLaSombra42 comments

> Giannandrea wasn't impressed by that, nor Bing's lack of local language coverage in markets Apple cared about, nor the poor progress Microsoft had made improving the search engine since the two tech giants previously discussed search preferences during 2015 and 2016.

Sounds like they had much more valid reasons than Bing being wrong about Annie Lennox's first band to me.

“Apple rejects Bing for being bad” isn’t clickbaity enough. But yeah, one would definitely think Bing would have support for every language/market under the sun after 14 years, or that Microsoft would have more to show Apple since 2015.

Google has it far too easy

For all its size, Microsoft is generally horrendous at making their products available globally. For example, Xbox Live Gold / Xbox Game Pass only became available in all EU countries in February 2023 (previously, you could only access it by buying pre-paid cards from another country).

Their (now defunct) music streaming service was never accessible in many of the poorer EU countries - not to mention globally (to add insult to this, they didn't even pull the advertising from it in local versions of Windows - you'd get big ads for it, but following them would only lead to a mess of redirects).

Even Sony, a much much smaller company, has more global coverage and faster than MS - the PS4 had full support in Romania for example since launch in 2013, and PSPlus has also been available since around the same time - nearly a decade before the MS services.

xcloud gaming is still not available in India, Idk what is holding them back

But I think Annie Lennox can be a great M&M test, too.

Yep. It's evidence that "AI" isn't automatically the saviour of Bing's search results, because in this instance it gets a pretty basic Google-able query (that an LLM can handle and be cajoled into handling correctly with enough RHLF) wrong. Google's "AI" results aren't much better at not being regularly wrong tbf.

All that will matter is that they're the only results, so no one can prove otherwise.

This has nothing to do with AI, the anecdote is from 2018.

  Giannandrea's testimony, as recorded in the filing, also states: "Microsoft was willing to sell Bing, which you wouldn't do if it was a strategic asset."

I think safe to say, it wasn't only about Annie Lennox. Even Microsoft knew they had a turd on their hands.

I don’t get this: “Privacy-centric search engine DuckDuckGo, the filing alleges, sometimes preferred to funnel fresh sources of funding to investors rather than improve its own service.”

They are using cash flow to pay dividends to investors?

That phrase also stumped me. It sounds like DuckDuckGo are taking funding and giving it to investors, which doesn't make any sense.

I also don't see how it is a defence of Googles quality. DDG would rather give money to investors than improve upon Bings search results, sounds like Bing is excellent and as good as it gets.

Or DDG is just sucking the well dry and isn’t bothering to build value for the long term.

Disclaimer: I applied for a job at DDG several years ago and they had the nerve to turn me down :)

> Privacy-centric search engine DuckDuckGo, the filing alleges, sometimes preferred to funnel fresh sources of funding to investors rather than improve its own service.

What does this mean? Paying back loans / reducing debt? Or more like dividends?

The way that's written, "funnel fresh sources of funding to investors", appears to me to be alleging something that sounds like a pyramid scheme or a Ponzi scheme.

The filing (1) claims they are a Bing Search and Ads reseller that invests almost nothing into their actual search engine, and instead repays investors with every new funding round.

(1) From page 137: https://regmedia.co.uk/2024/02/26/us_v_google.pdf

I'll never understand why Microsoft, with all their resources, never made Bing a better search engine.

They don't have to be conservative. It's not their only cash cow. They can experiment.

Instead, they opted to nearly copy Google's UI and leave it at that.

I think they would have done a lot better if they invested in a separate company and gave them freedom + lots of resources, like they did with OpenAI.

That's what Microsoft used to do with everything, they had so much money they could just make incredibly bland versions of everyone else's hit products.

I assume it's because Wall Street was mithering them about why they hadn't invented said hit product and they built their own versions just to shut Wall Street up while the Windows/Office money flowed in.

Bing is a remnant from that time that they never killed.

I think Bing is more important than that.

Bing is an insurance policy against Google having too much control over the internet. I don't think Microsoft care about market share, they care about having a foothold that they could invest in if they needed to without needing to go from a standing start.

Did they stop? Teams is still very much a poor copy of Zoom and Slack. Like they can't even make good copies.

Slack maybe, but I doubt they copied anything from Zoom. The Zoom UX is pretty inconsistent and horrific for some very basic things.

E.g. Ever tried to get a chat history from a call that's finished with Zoom?

My guess is that since they had most Windows boxes in the bag, they didn't really need to throw resources at that department to innovate. Search is really, really hard, and expensive. They had a built-in market, and it's cheaper to grow the market with one big deal than improve your product incrementally.

I would give a simple answer: they don't care. They care about marketing and sales and Bing came "preinstalled" in zillions of computers around the world. That is enough for them. The crux of Microsoft is being harm at their ecosystem (PCs, Windows, Office, Azure, Dev Tools, etc). They are the business drivers. But even if their software and infrastructure is bad, it is used in almost all of the top companies in the world and the switch cost is currently very high.

People always claim Bing has bad results, but I honestly don't think Google's are any better. Their extra metadata is regularly incorrect and their actual results have been polluted with SEO bullshit and ads for over a decade now. We simply lack any decent search engine and have for a long time.

Maybe the complaint was against the UI of Bing, which isn't great, it's very busy and cluttered. The results are as good or better than Google at this point. It might vary depending on use cases and how proficient you are at using either search engine, but in general Bing is a very strong competitor to Google.

As for the UI, Microsoft seems to have given up an any type of UI research after the Office ribbon interface lift. I don't know how you'd improve the UI of a search engine, but Bing could stand a bit of improvement, and so could Google.

Both are polluted with rubbish, but Bing's inability to find it's own features is a bit special

(For example if you search "houses for sale near $placename" it surfaces a property search UX which is actually quite neat. If you type "houses for sale in $placename" which is not exactly an obscure variant query, the same feature is completely hidden (although there is a big button saying "houses for sale" that will reorder your search results without surfacing the house search ui). Battling spam is an ongoing battle, the "best" results are subjective, but not being able to find its own features feels like half-arsed implementation.)

Maybe i'm nuts but I think attempts to be senantically cute with queries almost always backfires. I'd much rather treat it like a full text search.

Generally I do full text search if I want to find words and phrases in text and natural language queries if I want cute widgets to solve natural language questions.

But "house sale $placename" works fwiw!

My guess is worse bureaucracy and lower average engineering skill. These things matter.

As someone unfamiliar with how these sorts of proceedings work, how are the statements in the linked filing from competitors/ other companies in the space sourced? Is it discovery done on competitors as part of the defense case, existing public statements, sourced from previous public filings, something different entirely?

The headline is misleading and destructive to discourse by inaccurately simplifying Apple's rejection of Bing to a single, trivial issue, neglecting the broader, critical reasons such as Bing's inadequate local language support and lack of significant improvements. I'm SO tired of headline writers (AND the people who only read headlines).

JG did always care about knowledge graph factuality to a fault.

i would still call it a virtue; he's a very honest guy

And bing still gives the wrong result

I imagine that for MS pushing Bing to Apple could be beneficial even for free:). This would force Apple to use their own search engine instead of working with Google, thus hurting both Google and Apple. So it's natural that the deal didn't happen, I am sure people who make decisions in Apple are very smart and experienced ones.

>The thrust of the filing is that Google invested a lot to make its search service excellent, while rivals never succeeded in attempts to match it – so Google is not a ruthless monopolist, but a restless innovator. Even the presence of Google search in Android and Chrome, the filing argues, did not preclude competition.

Or others gave up on innovating because Google is a ruthless monopolist and would kill it either way. Happened to Edge.

I have even less respect for Apple after reading that they care who some popstar's first band was.

Please, do not look at the specifics. In this case many things were put to the test, one of those is the ability of returning answers from the search engine. That is a very important selling point for 'advanced' search engines and it is something that almost consistently under-performs.

Information is information and Bing got it wrong.

Presumably it's able to get well-indexed information right (e.g. "capital of France" => Paris), so to test its breadth and depth you have to get more obscure.

The particular query is irrelevant, just that they're testing like a reverse Gell-Mann Amnesia effect — asking it stuff that you _do_ know the answer to as a proxy for how good it is generally.

Try reading the article

Probably just looking for confirmation. The evaluation approach is totally valid.

Reply on news.ycombinator.com