Ask HN: Any obvious reason for duckduckgo vs. Google speed difference on mobile?

[fdsak]

Over mobile network (4G) I am unable to open duckduckgo from time to time and google is as usual very fast. I checked duckduckgo is not down and on my desktop duckduckgo works fine. I was thinking if there is any priority assigned to google requests by my network carrier and intentionally blocking duckduckgo. Using firefox over android, Pakistan

[notjtrig]

I would guess more of Google's landing page is cached, and it's a smaller page.

I'm getting Google in 0.5-3 secounds, DuckDuckGo in 3-4 secounds, StartPage in 2 secounds. USA 4G 3:30AM

On another note, both Google and DuckDuckGo seem to link directly to search results but actually redirect your broswer to capture the link you clicked. StartPage links directly to search results.

I've asked DuckDuckGo about this and they can't provide a sastifactory answer. Maybe someone could clue me in why that redirect is nessicary, but until then I consider them untrustworthy.

[aaronhoffman]

Isn’t it to mask the search terms that you used? If you go directly to the page the current url including the query would become the referrer

[londons_explore]

No https cross domain referr

[tagawa]

Hi. Sorry you haven't had a satisfactory answer from us - I'll try to do that here.

The click redirects are to stop the subsequent website from seeing a user's search terms, but we only use it for browsers that don't support meta referrer. Even then, we obviously don't collect or share anything via the redirect server, in line with our privacy policy ( https://duckduckgo.com/privacy ). There's a bit more about the redirects here: https://duck.co/help/results/rduckduckgocom

Also, it's possible to disable them completely here: http://duckduckgo.com/settings#privacy

Disclaimer: DuckDuckGo staff

[smt88]

I also noticed this and stopped using DDG because of it. My assumption was just that scaling is hard and DDG wasn't as good at it as Google is.

[puzzle]

Google has more POPs, search clusters everywhere, etc. I was going to say that they also use QUIC, but the poster mentions using Firefox, which I don't think supports it yet.

[DarkWiiPlayer]

I just tested this myself and, while google loads its DOM content almost twice as fast as duckduckgo for me (1.2s vs 0.7s), that's still way within a reasonable time for a webpage. For comparison, youtube took me 3.5 seconds to load the DOM content.

That said, it may still be some difference like browser caching, or maybe android even does that. My tests were on a Windows PC using chrome.

[fdsak]

I was thinking if there is intentional slow-down by carrier, as it takes sometimes a minute to open a web page ( over mobile) while google works fine

[bradknowles]

You could try doing a traceroute from your mobile phone to both Google and DDG.

The number of hops and latency you see in traceroute may help you figure out what kind of CDN and global server load balancing that DDG is doing versus what Google is doing, and thus why their respective web pages are the speeds they are.

[bradknowles]

If you go to gmetrix.com and generate page speed reports for both Google and DDG, you will see some interesting differences.

Of biggest importance to the OP is that DDG doesn't appear to be using any kind of CDN.

For DDG, their web page is the same size (384 KB), and the score pretty well on most subjects, except they don't defer parsing of Javascript (noted on the first tab of the Page Speed report), and on the second tab of the page speed report, you will note they don't use a CDN and they don't have "Expires" headers.

But for someone on a mobile device in a country that might have slow connectivity to the main web site for DDG, I think the lack of the CDN is probably the biggest factor.

[bradknowles]

I did try to see if I could generate some quick graphs with pingdom.com or runscope.com to highlight the importance of using a CDN, when sampled from various places around the world.

Sadly, while both pingdom and runscope do allow me to test sites and get HTTPS latency data on a per test basis, they don't make it easy to collect and graph that data -- on a per test basis. Sure, they can both give you uptime reports and tell you average response time across all tests, but I want more detail than that.

Still working on this.

[rurban]

Google serves content from a data center near you, others just from 1-2 central servers. You cannot beat the shorter connection.

That's btw Google's trick, from the very beginning. It was not pagerank or any better algorithms, it was purely massive investments getting into all the data centers worldwide.

[breakingcups]

[citation needed]

[siproprio]

Duckduckgo is slow for me as well. I stopped using it because of this.

If you're looking for an alternative to Google that is faster than duckduckgo, try Bing. It is still slow, but faster than duckduckgo.

[ggg9990]

Could be that Google is paying your ISP for traffic prioritization.

[fdsak]

Is this legal ? And how do we know if there is such thing ?

[krageon]

It's not legal in a lot of countries, and I also doubt it is the case. It's much more likely they are just positioned better all around the globe (with better peering, etc). Arguably this means the same thing in practice, but this is the thing that happens all the time (as opposed to direct deals with network operators).

[ggg9990]

It’s legal in the US.

[segmondy]

Google has more money and resources than duckduckgo. I suffer through duckduckgo because I rather anyone than Google. At this point, I might start giving bing a try.

[mkbkn]

In my case,

DDG: 4-5 seconds.

Google search: 1-2 seconds.

Carrier - Airtel (4G) in India.