[knorker]

You're focusing on the quality of results, though. That I think is actually the easiest problem, to the point where I didn't even list it.

Remember when Google+ launched, and Microsoft as a joke created a clone in a week or so? That was completely missing the point about what's hard.

Or how "make a twitter clone" is basically the "hello world" of web apps. If it's just you and your friends on it, actually yes you can make a twitter clone that you can basically not tell the difference, and you can do that in a weekend (another weekend to make the app).

To make a twitter clone for 10 people you can run it on your laptop. For 1000 people you buy a VM in some cloud. For a billion apple devices you need world wide pops, fibre deals, plots of land, construction companies, resource planning, legal teams, government contracts, etc...

Again, Apple could possibly do this. But this is not their core skill. And do you know what happens when a company throws billions on not their core skill? Google+ happens.

[lotsofpulp]

Are you suggesting DuckDuckGo would not be able to handle Apple switching the default to DuckDuckGo for technical reasons?

I do not know enough about what goes into delivering people search results worldwide. I just know that my family’s experience switching to DuckDuckGo has been seamless, but I also do not know how representative our search behavior is.

[knorker]

> Are you suggesting DuckDuckGo would not be able to handle Apple switching the default to DuckDuckGo for technical reasons?

Right, I think they would not. Not without A LOT of work, on all sides of the business.

If you have a 1qps service, you can scale 100x[1] fairly easily. If you have 1kqps[2] then scaling that to 100kqps is a different beast alltogether. Every single design in load balancing decision, TCP termination, request routing, peering aggreements, geolocation, backend load balancing, and failover, can be assumed to be wrong.

Or if not wrong, then at least untested and many parts will not survive first contact with the enemy.

And that's just to get the SEARCH working at all.

Remember, in order to replace Google's $15B you probably want some sort of revenue, too. And revenue of $15B+cost of service.

So… ads?

Google has buildings and buildings full of people doing nothing but ads. They have presence in pretty much every major city in the world. No, I'm not just talking about the engineering offices that are well known.

To ask Apple to create a search engine is actually to ask them to create "A Google" (except Cloud).

According to the latest earnings report Google spends about $160B and Search ads takes in about $120B. These numbers are not comparable, since they are different level line items. But it should be kept in mind that a very naive reading of this means that yes, Apple has $195B in the bank, but if they tried to "just create a Google" then they'd be broke in just over a year.

Especially since it would be MUCH more expensive and risky to build this in one big shot, than to organically grow it at great profit over many years.

Maybe better to get an earnings report from Google pre-cloud, when it was essentially an ads company by income and investment. Of course it won't be comparable unless Apple decides to also do an ad network. Which they probably would because if you have the tech and the customers, then it's free money.

But you said technical reason. So let's scratch ads, and never mind the money. Yeah, they could be able to do that. It's not clear to me how much of their search index they actually own, though. They say they have a crawler[3], but it sure also reads like "we're just a frontend for Bing". Truth is probably somewhere in between.

So what do you think MSFT would say if Apple started hammering Bing (albeit indirectly) for search results, sans ads? Or even with ads?

So duckduckgo is good because they don't actually have their own index. Bing took years to not be ridiculous (it's good now). We saw Cuil completely fail, even full of ex-googlers.

All this to say: Writing their own search engine is hard (see Bing, and how much MSFT plowed into that to make it work), and DDG can't just be used as a backend. And switching to DDG is just throwing $15B in the lake and giving it to someone else for free.

[1] that's the order of magnitude difference between google and ddg according to https://www.wired.co.uk/article/duckduckgo-google-alternativ..., of course that doesn't take into account that this would only move apple traffic, but I like the round number.

[2] 1.5B per month is about 578 per second, and with seasonality that it at least 1kqps at peak.

[3] https://help.duckduckgo.com/duckduckgo-help-pages/results/so...

[PaywallBuster]

DDG is just a frontend to bing's results

Why would it be difficult to scale?

[lotsofpulp]

I assumed it would not be difficult to scale. I was asking knorker why they thought it might not scale, as that is what I thought their comment implied.

[piyh]

Why couldn't they use AWS? They already do for iCloud.

[knorker]

Reportedly not just AWS: https://www.datacenterdynamics.com/en/news/report-apple-is-g...

Admittedly they don't have to vertically integrate, and they wouldn't pay list prices if they use any other cloud.

But honestly, if the idea is to get off of Google search, what exactly is the gain by relying on a third party albeit at a lower level?

You have to ask yourself: Is running a search engine the best thing that Apple could be doing with its time? Is the fact that they don't run their own search engine a danger to their core business?

In the end it comes down to projected cost and income, and obviously I'm not in a position to calculate either one for Apple, not being in the room with their ruthless negotiators.

But yeah, the starting point of dropping google is losing out on these $15B. So already that's what you have to work with. And then the cost of public cloud egress traffic, which is famously ridiculously expensive.

Your comment seems a bit like "why don't they just…", which seems a bit naive when dealing with business at this scale.