Hacker News Reader

Comment by hombre_fatal | Hacker News Reader

Every time I try to use it, almost every search query I write while trying to get work done I end up retrying with "!g". Eventually I find myself using "!g" preemptively because I don't want to miss out on results like long tail StackOverflow hits.

Though "elm list" is still my classic example.

View on HN

Hum... My first 3 results are tutorials about how to use lists in Elm. The 4th is a link to the Elm language home page, followed by the official documentation.

Google, instead, shows me a list-extras package first, followed by one tutorial on lists, one on types, an unrelated blog, a SO answer for "are Elm lists faster than JS arrays", followed by a bunch of random stuff, including the Facebook profile for A Elm List.

When you type "elm list" into Google and it understands that you mean the language instead of the tree, it's because they have an extensive profile on you. Of course DuckDuckGo won't do that since that's their unique selling point.

I can see this being a legitimate gripe of an end user that just compares search engines, but I was under the impression that this crowd understands the proposition of DDG and doesn't complain when it's working as designed.

I hear this presumption a lot from people trying to defend DDG, but you can trivially disprove this by googling it anonymously from a random IP address. Try it with "elm list".

Frankly, when I compare my logged-in, static-IP Google results to anonymous queries, I can see that the degree of "personalization" that people claim is vastly overstated; not much seems to change.

It just seems like wishful thinking when the obvious reality is that Google is just more advanced in its ability to semanticize input.

DDG "rust cargo" gives me images and results for cargo ships.

I followed your advice, turned on VPN, went private browsing and searched for "elm list".

On DDG the first result was elm-lang homepage but the second was a tutorial on lists in Elm. Down the page I also saw a wikipedia entry for a list of elm trees. Seems ok overall to me.

Then I tried the same thing with Google and first I was stopped by captcha and had to spend a minute clicking hydrants and traffic lights. Then I got a whole page of results about various programming discussions on lists in Elm, but nothing in the entire page about the trees. That does not seem right, I think regular person would be much more likely to search for a tree, not for an obscure programming language.

Verdict: I found DDG results better as in more relevant for general population while still surfacing the programming ones. Also no hydrants on DDG.

Would someone searching for the trees search "elm list" or "list of elms"?

How is there anywhere enough information in that query to know its about programming? I guess the difference is I don't expect DDG to know anything about the query (imagine talking to a random librarian vs your coworker). This example helps - I think its down to a matter of expectations and tradeoffs. Maybe fundamentally DDG is a "search engine" and GOOG is a "personal AI research assistant".

As I point out in my sibling comment, you can google anonymously to prove that it's not personalized. In fact, you can do this with every google vs ddg query to prove it's not personalization that makes google superior. No need to guess.

Also, ddg's first link is programming related, it's just a worthless result (Elm's homepage).

You ask how someone/something could know that "elm list" is programming related, but I don't see why you think it's unfeasible for someone/something to come up with confidence levels for the various possibilities.

Let's say you were sufficiently omniscient to know about the Elm language, elm trees, the elm email client, Professor Elm from Pokemon, and the tiny town of Elm, Switzerland. What do you think is a better result for the query "elm list"? If you can form an opinion, why couldn't the world's most lucrative machine learning project form a similar one?

Reply on news.ycombinator.com