Comment by ansy | Hacker News Reader

Comment by ansy | Hacker News Reader

This is really an example of taking Google's strengths and making them out to be weaknesses. Google's search algorithm is tuned to what actually works for most people. Does anyone search for Java and click on coffee links? Or links for Java the island in Indonesia? Based on these results I would bet not. The thousands if not millions of people that typed Java before you didn't want coffee or islands. They wanted the Java virtual machine and programming language so all those long clicks to Java programming sites and JVM downloads have corrected the results page to reflect that.

But if you typed "cup of java" or "java island" there would be no mistake. Those people wanted coffee and Indonesia. So that's what links you get.

I have used DuckDuckGo as my primary search engine. But it is not competition for Google. Google is much, much more advanced.

If I typed in a proper name and that proper name was tied to an address in my area such as by being in the local white pages or having an address on a website, I would see the local result for that name because Google looked up the geolocation of my IP. That sounds a little creepy, but I got what I wanted. Google gave that link extra weight for being near my location therefore it wasn't buried on page 50. That's incredibly smart. How would DDG solve the same problem? It doesn't even try.

Google's search algorithm is tuned to what actually works for most people.

Yes. And that is where I think it goes wrong. The fact that a lot of people misspell should not be used to correct a person's right search to a wrong one. That's not the right thing to do. It should suggest a different search, but not change my search altogether.

As an example, if I place an order for a Chicken Burrito but am served a Beef Burrito because that's the most popular dish, it would not be okay. Same applies to search.

Google is much, much more advanced.

Yes! I said they over did it.

I think Google should definitely be over-careful when deciding to vacate the searcher's terms due to suspected miss-spelling but I don't agree that it should vacate the feature altogether. I'm sure Google has evidence to support a net increase in searcher satisfaction with the behavior. Perhaps there's a middle ground when the algorithm is less certain there's been a miss-spelling and includes some results satisfying the original query.

I think the burrito analogy is flawed... searching for a word Google doesn't know about is more akin to ordering something that wasn't on the burrito menu at all, or say... if the cashier misheard your order and thought it could have been either chicken or beef so they made an educated guess and served you a Beef Burrito at the same time asking if that was what you meant, giving you an option to receive your correct order if it wasn't.

Obviously in that situation they would just ask you to clarify your order before making a burrito, but if they had the ability to serve burritos as fast as Google serves results, then for the majority of cases (assuming the beef burrito is most popular) the outcome is satisfactory and it isn't worth the time or effort to ask and wait for an answer.

But Google did show the correct results when I clicked on the link suggesting I was searching for the actual string. So it was on their menu all the time. I also communicated my request correctly to the other end; the other end deliberately changed my order!

no, it really is a terrible analogy :)

the menu contains almost every word in almost every language that's online (plus many misspellings and made up words and lists of meaningless character permutations and...), so if you are going to automatically correct a query, some set of words that are on the menu are going to have to be excluded.

On the other hand, if we accept that this case was suboptimal, the solution seems to be one of:

1) never autocorrect, only offer a "did you mean." I would bet this would result in more people having to click than the current situation (in spite of the annoyances reported in this thread).

2) be more conservative with corrections and offer a "did you mean" more often. 500,000 results for "uncollege" and the fact that correcting it to "college" repeats a word in the query string is a good argument for this particular case, at least. On the other hand, testing for "[some random word] and uncollege" never seems to correct for uncollege unless the other word is education related, which is interesting. While the autocorrect isn't smart enough to notice that it's correcting to a repeated word, it is doing some kind of clustering and deciding you probably meant college in this case.

I'm pretty sure that Google being Google, they did exactly this kind of testing.

Put 500k people in bucket 1, and 500k people in bucket 2, and see what happens.

It's pretty obvious to me that the vast majority of times I see that correction message, that I've actually misspelt a word, and that it's pretty rare that I did actually want to search for "dictonary".

Tyranny of the majority. The ultimate expression is to remove the search box altogether and just show the most popular porn sites - because that's the most frequently searched-for result.

Or just show the Wikipedia article on Philosophy :)

I think the analogy is flawed, but for a different reason - if you're given the wrong dish in a restaraunt, you can't correct it in less than two seconds.

I would rather have a search that corrected typos and aimed for the best result which I could rapidly correct than a search which slavishly returned "No results for 'hgihway', did you mean 'highway'?" (as a simple example).

DuckDuckGo corrects typos: https://duckduckgo.com/?q=hgihway

If you actually want to search for "hgihway", search for "+hgihway": https://duckduckgo.com/?q=%2Bhgihway

Of course, Google has results for "hgihway": https://encrypted.google.com/search?hl=en&sa=X&ei=mV...

We find that our users really like disambiguation. I find it most useful on searches for people (https://duckduckgo.com/?q=Doug+smith) or places (https://duckduckgo.com/?q=Deerfield).

Re: local search, we're bringing in datasets from services like Yelp. For an example see: https://duckduckgo.com/?q=Wood+tavern. We're always looking to improve, so if you've got interesting local data, please reach out.

It’s questionable that in the case of Java the disambiguation helps much.

Here is what you have to do to get results for Java coffee if you are using Google:

Type “Java”,

parse results and see that you didn’t get the desired results,

type space, “c”, “o”, “f”

And DuckDuckGo:

Type “Java”,

press enter,

parse results and see that you didn’t get the desired results,

expand right category in the disambiguation list,

click on “Java coffee” in the expanded list

It’s a wash, it really is.

So try a different example.

Scala also has multiple meanings. As of 3 months ago, it was almost impossible to find relevant pages about "Scala" or Lift through Google without knowing including more relevant information ("Scala Language" or "Lift Scala").

I'm able to easily type in a second word when I search. In the case of Java, I'm able to get the language, island, or coffee on the first search in google - for this 'princely' cost of searching correctly in the first place, I don't have to do the (apparently wonderful) ddg process:

1) visually skip the sizable disambiguation list

2) parse the main results, find they don't match (DDG is all about java language just like google on 'java')

3) go back and parse the disambiguation list

4) click the category to expand

5) parse the newly expanded list

6) click the relevant entry

7) wait for a new search

8) parse the main results again

I mean, seriously, a second word in your search is hardly cognitively onerous. Are we really so spoiled as to demand perfect results on short, ambigiuous topics?

Sorry, but Google's results are not defensible in these two cases. A search engine that spends a fraction of real estate to suggest specificity on rather generic term is performing better. It would be difficult to test but I suspect on a search for "java" that coffee or geography would fare better in relevance than the second half of Google's results.

Likewise, it doesn't even make any sense to search for "college or college" without quotes. It makes even less sense when the internet includes an exact match on "college or uncollege". Oddly, a search for "college and uncollege" (without quotes) finds the article in the #1 position. Again, is it too much to ask that Google include a few exact match results even when it strongly suspects a miss-spelling?

To those who downvoted: is your position that Google's results are better in these two cases? Or that it doesn't matter? Or that the cost to match/exceed DDG's results outweighs the benefits? Or that since Google works differently behind the scenes it should not be expected to deliver these kinds of results?

Downvote? Really?

Reply on news.ycombinator.com