If Apple launched its own search engine and made it the default on its products, everybody would welcome that and absolutely no-one at all would denounce it as an abuse of their dominant position on their own products and call for tougher antitrust enforcement against Apple.
Also:
> Apple already designs its own [list of digital services].
> The only thing missing is a full-blown Apple search engine.
I dunno, being able to deploy their own baseband processors feels like a higher priority for Apple.
Will it happen though? I get that COVID will be disrupting this but in my mind this has been the next big hardware (vs device/software/service) target since Apple Silicon.
In my mind it's almost Apple Car levels of vaporware.
They're certainly trying. They bought up the old Intel modem IP and team. Qualcomm is an existential threat to their business so it makes sense to try to get them out of there.
Their continuous rent-seeking, anti-competitive behavior.
Anything they touch basically seems aimed at the goal of extracting as much money as possible. All their designs basically force companies to use all their products instead of using competitor products (notice how designs with Qualcomm SoCs also have Qualcomm power, audio, and numerous other QC chips). They're a blight on the industry.
And yet they provide the best there is? Qualcomm does not bound you into audio, but do bound for certain parts where together they made better product package such as power amplifier or RF Front End with their Modem for better pricing. i.e It doesn't mean you cant buy their standalone Modem or their SoC without the others.
And it was really Anti- Competitive and monopoly as many has lead ( or not lead ) to believe, Mediatek would not be eating their market share ( in terms of Unit Shipment ) top and bottom.
The article completely overlooks the cost of providing the service, which for Google is balanced against a bit more than "a revenue stream about the size of the Apple Watch".
It also has what it needs to launch its own Facebook… oh that didn’t turn out well for google, almost like there’s more to a market than just making the tech.
Local / online hybrid LLM’s could replace googling and Apple is in a great position to provide them. The scary question is how they will monetize this.
I'm curious if you can back this up from a technical perspective. I work with mobile (but all Android these days) and it is a mess. Best training ecosystem is Pytorch (and maybe Jax). For inference, you need to endure all manner of pain to optimize/quantize your model (PTQ has perf loss so Training aware quant is key). There are lil operator optimizations (fusion helps perf) but then certain ops have issues on the accelerator you care about. This android experience is mostly related to ARM issues. Q seems to be investing heavily in tooling but now that they are laying people off, I think their toolchain is a bit risky (and may not work if not on Q hardware). Is the story for optimization much different for Apple silicon?
Simply launching another search engine, even if it's better, won't be enough. There needs to be a transformational change to get people to switch and I don't think llm integration is enough.
One area they could gain an advantage over Google is by downgrading sites with advertising on them. Google won't do that because it would hurt their bottom line even though they know much of their results are garbage.
>There needs to be a transformational change to get people to switch
All they need to do is change the default search engine in the next version of iOS. As long as the results don't suck that would probably be enough of a leg up.
is this just google propaganda trying to sway public opinion towards them not having a monopoly because “well apple could do it”. Why would apple invest anything into search when they get paid billions not to?
Honest take: I’m an Apple customer (no, I’m not interested in hearing how terrible that makes me). I want to see them focusing on innovating in hardware that works for their products. I want to see them improving and stabilizing Swift and SwuftUI. I want to see them actually investing in fixing long-standing issues with macOS. The last thing I want as an Apple customer is for them to launch a giant new undertaking like search. Right now they have been fairly good, with some significant exceptions, on user privacy issues. A search engine, especially one relying on ad revenue, would muddy those incentives. And it would be a massive investment in an area they just don’t have any reason to touch aside from trying to stick it to Google.
I share your concern, but I'm a bit ambivalent on this. I though the same of Apple TV: I don't want them wasting energy on things like bad movies/TV shows when macOS's got so many missing features and bugs.
It turns out, it didn't diverge their attention, at least from the outside, and they created good content.
Google needs a good push, that's for sure. I doubt Apple has the core talent now to compete when Siri is what is and has been for more than a decade, but who knows, they've got the money. Maybe they could buy Kagi.
AppleTV felt to me like it wasn’t so far outside Apple’s reach, especially taking into consideration iTunes as a major product/precursor.
I’m more concerned about the incentives that being in search would create, to be honest, because if it is at all based on ad revenue for support, it injects incentives exactly opposed to the ones they have now as far as user privacy, etc. are concerned.
The incentives issue is concerning. The move to services has already gave the Settings App a grocery store feeling (“click here to get more storage/family pack”)
I will place a bet against Apple successfully launching their own general purpose Google competitor successfully before 2028.
Search is incredibly complex and expensive to build and run. Apple lacks the server side chops to build super low margin services (iirc they’re around double the cost of gdrive). Say, it’s 2-3 years for them to build out capacity, and another 1-2 years to get the software working well. That still would make it a mad dash.
That being said, I can see Apple building a special purpose alternative to Google, indexing a small subset of the web, which is curated - think what gets put on Apple News, or similar.
Right, but hiring, building a team, getting them performant — takes time. Apple can do it as a skunkworks style operation, probably better than most other companies out there, but such an undertaking would be a multibillion dollar hole that I expect would be hard to explain away, even if it classified as “iCloud R&D”
I'd be extremely skeptical, as Apple has shown over time that this kind of large, data intensive software is not in their wheelhouse. I think Apple makes some great products, but if I had to make choose the shittiest products Apple makes, it would have to be Siri, Apple Maps and iCloud. These are all areas where I think the Google equivalent (Google Assistant, Google Maps and GDrive/Docs) are much better, and the these are all much more closely related to search.
It really depends on what and how you use it. I know that's not saying a lot, but Siri is genuinely good at some things, and incredibly terrible at others.
I know the same can be said about Assistant, Alexa, etc.
That may be true only in the U.S. and a few other countries. I recall Apple said several years ago (maybe around 2017?) that it would improve maps in India. It’s 2023 and it still can’t find common addresses in major cities. Considering that Apple has repeatedly stated that India is a key market, is it the competency of the maps team that ought to be questioned?
> ... it would have to be Siri, Apple Maps and iCloud
Probably. And I think a large reason for that is because they don't like to solicit feedback for how their products work and for those types of products you NEED feedback.
It's just not in their DNA. Steve Jobs famously laid out the philosophy of "the user doesn't know what they want until we tell them" and it's a core part of their formula. At any rate, they have enough money and time to get it wrong for a long time before getting it right.
Apple Maps is actually really good these days. It doesn't have feature parity with Google Maps, but I don't necessarily consider that a bad thing. The core things: directions, POI and overall design are just as good, if not better than Google.
What features are missing at this point? I think their addition of offline maps this year was the last thing that they were missing. At least as far as I can tell.
I wonder if anyone is starting to think that OpenAI And it’s ilk has the ability to commoditize web search.
Right now the others lose because Google’s really good at it. I have to admit every time I’ve tried a different search engine I’ve been massively disappointed. Heck the same for any map besides Google (except Waze which well…)
Total spit ball here - but I wonder if the thought is more trying to get to where “the puck’s going to be” where the search Apple is working on is personalized and integrated across all your life/apps/platforms as opposed to something you do typing into a field on a website.
But also with a focus on improving productivity and user value then as a way to target ads.
I see this comment a lot recently and it perplexes me every time...
I don't see how LLMs are search tools. Am I using them wrong? I've asked them for information with sources and they've hallucinated entire answers including accurate-looking but entirely non-existant URLs as sources.
They've been great for getting some ideas, and summarizing things, but if I venture into an area that I'm not a domain expert in an LLM seems like a terrible way to get info.
But everyone seems to think they're the next search replacement... So I have to wonder if I've just got bad luck with them, or somehow haven't groked how to get them to give valid truthful info.
They're a direct replacement for most of the ad-filled search queries you might have used Google for before like "Trip ideas for Paris" or "Meal plan ideas". The best part is that you can ask follow up questions with the same interface, which you can't after clicking onto some random web page match from the search results.
Using them for something like "SQL error: <some copy/pasted error>" is much more dicey, but Google is unlikely to get much ad revenue from a query like that anyways.
Sure, the use case of "find this page" isn't going away, but I'd be quite scared if I were Google, because the use case where they make the most money is going to be handled by LLMs shortly instead of search.
Totally agree. Somewhat ironically, recently I was playing with plugins with ChatGPT 4 and was testing out some of the "trip ideas" questions. I actually got great results - interesting suggestions that were personalized to what I told it about myself that I wanted - when I used "plain" ChatGPT 4. Then when I turned on the TripAdvisor plugin, the recommendations went totally to shit. It basically just spat back the first 3 items from "Top outdoor things to do in X" from TripAdvisor. No personalization, and I had just mentioned outdoor things as one of the qualities I was looking for, but I listed a bunch of others. Also no "off the beaten path" recommmendations.
I mention this because Google essentially operates like the TripAdvisor plugin, and I always just get back "Top X" lists that more often than not seem like SEO spam. I use ChatGPT now pretty much exclusively for those types of questions where I'm not simply trying to find a page on the web.
It depends on what kind of information you need. If i need an authoritative source then LLMs cannot really replace search currently. If they improve however it could become the replacement. For example if you say find me authoritative links the AI could run the search, check the results and give your best ones. It does not do that currently but it potentially could in the future. Now this does not really replace the need for google as a search engine. It might mean that a lot of google traffic would suddenly come from OpenAI. That would probably impact googles bottom line. If the traffic gets big enough they will block it and ask for payment. Or they might even ask OpenAI to show their Ads to the user.
If you just want a quick answer then LLM can actually suffice (if you trust the answer and dont need to check). Like for example how many eggs do i need for my cake?
Apple Maps is now superior for me in every way. Two things at play: Apple Maps has gotten much better in the past couple of years, while Google Maps appears to be in unsupported bit rot mode.
Google maps is still better at searching for retail stores, Apple Maps is better at everything else, 10x so for driving.
I live in a major US city and I refuse to use Apple maps for navigation. In the past year, I've tried it every so often and it has taken me to the wrong place, on strange, longer routes, and tried to take me through roads that don't even exist. The traffic information is also terribly inaccurate.
So no, maybe in some cities it works, but its still a pile a garbage compared to Google maps.
Pretty subjective at the end of the day. Personally I find them roughly equivalent quality in mapping/traffic but Google has turned into a pile of sponsored content. Apple frequently has the exact locations of businesses within a building wrong or confusing but I'll take some walking around a building over a recommendation for Subway when I search for "camera store". Anyway I submit corrections to Apple Maps and they accept those corrections within a couple of days 90% of the time and when I need that thing in the future it is correct.
I mostly use navigation for cycling in London and driving in parts of South America. Google Maps does what you're describing to me more often than I care. Apple Maps has been fairly solid.
In countryside England, both can give bad directions some times but I have still found Apple's to be superior.
Apple Maps isn’t great, but in Boston, MA, Google maps has been an awful experience. I think the most egregious issue Ive experienced is that Google tries to adjust your location to the route it has planned, rather than displaying your location on a map. It’s also unable to deal with the tunnels, and re-routes because it thinks I’m on surface streets if going too slow.
Finally, it’s instructions just flat out aren’t good for dealing with the complexity of Boston’s intersections.
I’ve had more luck with Apple Maps but mostly I just navigate by pre planning my routes using a map
Interesting…I have had the opposite experience. Apple maps has recently been much better at providing more complete results for searches (gyms, restaurants, retail, etc.), while still lagging behind Google for driving directions and traffic. Google maps appears to be excluding relevant results from searches (which appear in Apple maps), and instead offering up promoted locations.
Apple maps is absolutely horrendous in Croatia. Missing streets, wrong street numbers, wrong directions, and let's not even get into how many POIs and retail are missing.
Bad satellite images, missing street view and 3d views are just cherry on top.
Kagi, as everyone seems to forget, is a frontend for the Google Search API with some nice filtering/custom prioritization features. The results are good because they’re Google results.
Apple Maps and iCloud are excellent. Siri still does suck.
Your conception of the first two are probably due to their subpar launch. But they have improved to the point of being better and more tightly integrated than their competitors nowadays.
If they try this, and I'm not convinced they will, it won't be a general purpose search engine. It'll be something that only runs on their devices and in their ecosystem.
Apple does have a crawler of its own, which feeds into Siri's knowledge. But, there's no public front end to use it as a search engine, though there was up to about 2021 or 2022, provided as a bit of fun by some enterprising third party.
This is probably something Tizen-equivalent for Apple. Even if they don't seriously consider launching this to public in a foreseable future, this will be a great card for negotiation against Google, not just for default browser auction but lots of other stuffs. This is especially important in the upcoming DMA context so that they would incentivize Google to keep the Duopoly situation rather than putting the app store/browser space into serious competitions even after Apple opening up the third party app store/browser.
Wait, I thought they were going to replace Tesla with a self-driving Johnny Ive Apple-mobile?
Sarcasm aside, headlines like this about what Apple could do are the equivalent of flying car cover stories for Popular Science. The articles are probably like pre-written obituaries, just kept in the can for a slow news day.
<popular company> <could, should, must, will> <something popular company doesn't do>.
> Google already knows how to make great search engine. It is just 'choosing' not to, so it can make more money from advertisers.
I would guess they are trying to do both. I'm assuming the search team is 1000s of people with many of them trying to improve it.
I think part of the reason search has gotten bad (once you get past the ads) is the SEO spammers. I use DuckDuckGo and it has pretty much the same problem of shitty websites with copied content clogging up the search engine. The more zeroed in I am on some of my searches the more duplicate websites with fucked urls comes into play.
I would like to see less of Apple's revenue come from advertising, not more. So if it is a paid search engine with no ads, then awesome. But I doubt that will happen.
More than search, I wish Apple would compete with Google Workspace. Correct me if I am wrong, but from what I can tell, even with the new support for custom domains and Apple Business Essentials, iCloud still isn't designed for teams. The number of people you can share a domain with and the number of email addresses you can have per domain seem very limited. Or are there different limits for managed Apple IDs?
>More than search, I wish Apple would compete with Google Workspace.
One of the things I like about the Apple ecosystem is that it doesn't cater to enterprise customers very much. Their focus is on what the individual wants rather than what the CTO of some huge corporation needs as is the case with Microsoft.
> assuming it can sell advertising and search slots at the same price as Google
I'd imagine if Apple did launch a search engine (which I don't really believe is fully within their proficiency wheelhouse, though maybe buying something like Kagi could work?) I'd think they could sell their ads for more than Google does for the same reasons developers make more on iOS compared to Android.
All that said I'd really not like to see Apple go down the advertising route. I'm already unhappy with the amount they do and would like to see that shrink, not grow.
I’m always kinda bewildered by this idea. I see the plausibility, but over the last 10 or so years, apple has sold itself to people as the tech company that cares about privacy, so much so that you have to pay a premium and forego the ability to repair their products to enter their walled garden thats supposed to be privacy oriented. Would there be any fear on their part that customers would be upset that the company is now in a position to sell users data or services based on user data? Would this rollout of an ad supported search engine from apple have no backlash and be friction free?
101 comments
[ 2.5 ms ] story [ 191 ms ] threadAlso:
> Apple already designs its own [list of digital services].
> The only thing missing is a full-blown Apple search engine.
I dunno, being able to deploy their own baseband processors feels like a higher priority for Apple.
Will it happen though? I get that COVID will be disrupting this but in my mind this has been the next big hardware (vs device/software/service) target since Apple Silicon.
In my mind it's almost Apple Car levels of vaporware.
Which part of Qualcomm is an existential threat?
Anything they touch basically seems aimed at the goal of extracting as much money as possible. All their designs basically force companies to use all their products instead of using competitor products (notice how designs with Qualcomm SoCs also have Qualcomm power, audio, and numerous other QC chips). They're a blight on the industry.
And it was really Anti- Competitive and monopoly as many has lead ( or not lead ) to believe, Mediatek would not be eating their market share ( in terms of Unit Shipment ) top and bottom.
If they really need a new op, they add it.
>There needs to be a transformational change to get people to switch
All they need to do is change the default search engine in the next version of iOS. As long as the results don't suck that would probably be enough of a leg up.
I otherwise agree but defaults are a very powerful tool.
It turns out, it didn't diverge their attention, at least from the outside, and they created good content.
Google needs a good push, that's for sure. I doubt Apple has the core talent now to compete when Siri is what is and has been for more than a decade, but who knows, they've got the money. Maybe they could buy Kagi.
I’m more concerned about the incentives that being in search would create, to be honest, because if it is at all based on ad revenue for support, it injects incentives exactly opposed to the ones they have now as far as user privacy, etc. are concerned.
Search is incredibly complex and expensive to build and run. Apple lacks the server side chops to build super low margin services (iirc they’re around double the cost of gdrive). Say, it’s 2-3 years for them to build out capacity, and another 1-2 years to get the software working well. That still would make it a mad dash.
That being said, I can see Apple building a special purpose alternative to Google, indexing a small subset of the web, which is curated - think what gets put on Apple News, or similar.
https://www.protocol.com/apple-hires-cloud-open-source-engin...
Has much materialized from this?
I know the same can be said about Assistant, Alexa, etc.
There's multiple roads where it will tell you constantly to take a U turn while indicating a straight line visually.
It routinely lags weeks behind road closures.
It's lane suggestions are laughable at best.
I've driven many many miles around the US and Apple maps is easily a distant third between Waze and Google
Probably. And I think a large reason for that is because they don't like to solicit feedback for how their products work and for those types of products you NEED feedback.
Right now the others lose because Google’s really good at it. I have to admit every time I’ve tried a different search engine I’ve been massively disappointed. Heck the same for any map besides Google (except Waze which well…)
Total spit ball here - but I wonder if the thought is more trying to get to where “the puck’s going to be” where the search Apple is working on is personalized and integrated across all your life/apps/platforms as opposed to something you do typing into a field on a website.
But also with a focus on improving productivity and user value then as a way to target ads.
I don't see how LLMs are search tools. Am I using them wrong? I've asked them for information with sources and they've hallucinated entire answers including accurate-looking but entirely non-existant URLs as sources.
They've been great for getting some ideas, and summarizing things, but if I venture into an area that I'm not a domain expert in an LLM seems like a terrible way to get info.
But everyone seems to think they're the next search replacement... So I have to wonder if I've just got bad luck with them, or somehow haven't groked how to get them to give valid truthful info.
Much much better than anything I get with an equivalent google search.
Using them for something like "SQL error: <some copy/pasted error>" is much more dicey, but Google is unlikely to get much ad revenue from a query like that anyways.
Sure, the use case of "find this page" isn't going away, but I'd be quite scared if I were Google, because the use case where they make the most money is going to be handled by LLMs shortly instead of search.
I mention this because Google essentially operates like the TripAdvisor plugin, and I always just get back "Top X" lists that more often than not seem like SEO spam. I use ChatGPT now pretty much exclusively for those types of questions where I'm not simply trying to find a page on the web.
If you just want a quick answer then LLM can actually suffice (if you trust the answer and dont need to check). Like for example how many eggs do i need for my cake?
Google maps is still better at searching for retail stores, Apple Maps is better at everything else, 10x so for driving.
So no, maybe in some cities it works, but its still a pile a garbage compared to Google maps.
In countryside England, both can give bad directions some times but I have still found Apple's to be superior.
Finally, it’s instructions just flat out aren’t good for dealing with the complexity of Boston’s intersections.
I’ve had more luck with Apple Maps but mostly I just navigate by pre planning my routes using a map
Bad satellite images, missing street view and 3d views are just cherry on top.
Do we have recent numbers for this? I find Kagi’s results hands down better on most queries than Google.
Your conception of the first two are probably due to their subpar launch. But they have improved to the point of being better and more tightly integrated than their competitors nowadays.
Only using 1 companies services seems like asking for trouble.
Sarcasm aside, headlines like this about what Apple could do are the equivalent of flying car cover stories for Popular Science. The articles are probably like pre-written obituaries, just kept in the can for a slow news day.
<popular company> <could, should, must, will> <something popular company doesn't do>.
Google is going to just sit idly?
Google already knows how to make great search engine. It is just 'choosing' not to, so it can make more money from advertisers.
If it didn't have to pay $20B/year to Apple, they can re-deploy all that capital however they see fit to maintain its dominance.
Probably, that is what most companies the size of Google do.
I would guess they are trying to do both. I'm assuming the search team is 1000s of people with many of them trying to improve it.
I think part of the reason search has gotten bad (once you get past the ads) is the SEO spammers. I use DuckDuckGo and it has pretty much the same problem of shitty websites with copied content clogging up the search engine. The more zeroed in I am on some of my searches the more duplicate websites with fucked urls comes into play.
More than search, I wish Apple would compete with Google Workspace. Correct me if I am wrong, but from what I can tell, even with the new support for custom domains and Apple Business Essentials, iCloud still isn't designed for teams. The number of people you can share a domain with and the number of email addresses you can have per domain seem very limited. Or are there different limits for managed Apple IDs?
One of the things I like about the Apple ecosystem is that it doesn't cater to enterprise customers very much. Their focus is on what the individual wants rather than what the CTO of some huge corporation needs as is the case with Microsoft.
They made $30B in profit last quarter. $166B in the last 4.
That level of profit, coupled with their existing manufacturing and software capabilities means they can do basically anything.
What they do is very carefully chosen to maintain their current steady path. Speculating on what they could do in these terms is odd.
The feature roll out is slow but the UI is great.
A better map if not better mapping.
I'd imagine if Apple did launch a search engine (which I don't really believe is fully within their proficiency wheelhouse, though maybe buying something like Kagi could work?) I'd think they could sell their ads for more than Google does for the same reasons developers make more on iOS compared to Android.
All that said I'd really not like to see Apple go down the advertising route. I'm already unhappy with the amount they do and would like to see that shrink, not grow.
They could probably have sufficient ad inventory at launch to become a big player in the ad space.