[mlthoughts2018]

I sincerely believe that with a little shoe polish, a little more sincere effort into information retrieval mathematics, and better marketing, DuckDuckGo would be a superior search product.

Currently I use DuckDuckGo for probably 85% of searches. There are specific search cases that don’t work well, and specific types of automatic categorization and presentation (e.g. sports scores, rich location data) that Google is well ahead on.

But in many mundane searches, the result quality is indistinguishable and the lack of personalized tracking would break any ties strongly in DuckDuckGo’s favor, so strongly that it’s obviously worth it to split searches between two different engines on a case by case basis.

And in fact, I like some things about DuckDuckGo better, particularly I like the visual experience of its Open Street Maps results better. It’s not as information rich as a Google Maps result, but Google Maps is visually too cluttered and often suffers performance issues that a lower-tech maps service doesn’t. (DuckDuckGo also lets you select from a few choices for the maps backend, including Google Maps).

Don’t get me wrong: I’m sure Google will remain far ahead as the search product leader.

I only mean that in terms of the implementation and actual user experience, DuckDuckGo doesn’t have far to go before it would be entirely a substitute product that completely replaces an average search experience on Google.

(I’m speaking as someone who had worked on all types of personalization features for an Alexa top 400 site’s product search engine — so I have a lot of work experience informing my opinion that the actual distance, in terms of the investment to reproduce feature parity, between DuckDuckGo and Google is not that high.)

Google’s original efforts to create internet search were amazing. But now the underlying search tech is totally a commodity, including most of the fancier machine learning and information retrieval features. It’s why they have to integrate advertising so tightly to it. Search features alone don’t differentiate it as a product anymore.

I do hope a service like DuckDuckGo invests in that last 10% of the squirrly little extra features it needs to provide to seriously compete for overall market share.

[amelius]

I wish DDG would include a button in the search results that says "Search in Google instead".

That way, I would actually use DDG as my main search engine because it would be easy to resort to Google in specific cases.

[gbear605]

Check out Bangs on DDG [1]. Just insert "!g" into your search text and it automatically redirects your search to Google. It works for thousands of other websites too. (Wikipedia's !w is one I use all the time)

[1]: https://duckduckgo.com/bang

[amelius]

Yeah, I know about bangs, but on mobile they are cumbersome to use: I have to tap my screen 7 times to change from DDG to Google:

    1. tap DDG search bar
    2. tap space
    3. tap shift key
    4. tap "!"
    5. tap shift key
    6. tap "g"
    7. tap enter key

If DDG included a simple button, it would be 1 tap. Of course, I can't blame them for not linking to Google, but this way I will stick with Google.

[mlthoughts2018]

Wouldn't it be amazing to create a rent-seeking feeder website similar to e.g. Expedia, Priceline, Hipmunk, Orbitz, etc., that sits on top of data provided by other providers (in those example cases, data from airlines and so on, but in the search engine case, data about the characteristics of search results).

Then a user can go to my site at www.mycashcow.com or something and type in a query, which behind the scenes farms out queries to other search engines, gets characteristics about their results and predicts which of the search engines you would most prefer for this particular query, and then navigates you to that search engine's results page.

This way you always use one portal to search, and to the extent that you approve of its underlying prediction model, it automatically routes you to your preferred search engine for a given task, without needing to open a different tab or even click a button.

Expedia-like sites have been able to do this because by acting as aggregators, they solved a general search traffic problem for a fractured set of service providers (hotels, airlines, etc), which in some cases have to make some of the data publicly available. So the providers could not easily avoid a race-to-the-bottom commodity effect or avoid paying the aggregator sites for preferred placement. Seamless is doing this now with local food delivery.

We would get destroyed by Google if we tried this though, because then it would mean Google traffic would depend on Google actually functioning as a quality search engine, rather than happening to be a monopoly search engine (good quality, sure, but still monopoly) and reaping benefits in the form of advertising integration. Advertising would only pay for Google ads if the traffic driven to Google from www.mycashcow.com was in line with what they wanted, which would incentivize Google to either pay me to rank them higher, or do things like ceasing to track users because that would make them competitive with what users like about DuckDuckGo.

Since Google could sue me so hard regarding whether I am allowed to farm out a "pre-query" to their search engine, analyze "their" search results data, and recommend if a user would prefer it to the other options, I don't think this business could get off the ground unless it was created with a pre-arranged collusion with Google from the start.

[amelius]

Why would it be illegal to combine data from different sources? As a user, I'm doing this all the time. So why would it be different if a machine does it for me?

Of course, if somebody is making money off of it (mycashcow.com in your example), then it might be a different story.