- In my last post I asked for a Pangramic Palindrome- a sentence that was the same backwards and forwards and had all of the letters in the alphabet. I had spend some time on Google and other search engines trying to find such, with no success. Within an hour of posting Gareth Rees pointed to Peter Norvig's Pangramic Palindrome! I also found out (NOT to my surprise) that someone else had asked about such things. (NOTE- my spell checker wanted me to replace Pangramic with Pancreatic.)
- In this post I asked if the following was true: Given 2k-1 integers there exists a subset of k of them that sum to a multiple of k. I had looked at Google ALOT for this one and also asked some people, but didn't find anything. The comments pointed to a paper on the subject which answered the question (yes) and also gave additional theorems. If I had asked just a few more people I would have gotten it without asking my readers; however, like the HALTING problem, its hard to know when to stop asking and when to start posting.
- In this post if a certain sum that my co-author Clyde Kruskal proved was new. (I was sure it wasn't but couldn't find a reference.) I got a WONDERFUL combinatorial proof in the comments. I'm much happier with the combinatorial proof. (I also got a reference- Euler beat Clyde to it. Oh well.)
- In both this post and this post I asked readers if they wanted to review books for SIGACT NEWS. I got far more responses then I usually do when I just print it in my column.
Will search engines ever be so good that they are better than asking experts or asking your readers? (In the future we will all have blogs and hence we will all have readers--- though with FACEBOOK the future may be now. In the future we will all have 15 readers.) I doubt it. For several of the questions above I didn't know quite what to look for. For example I didn't know to look for Palindromic Panagram.
On a related note- has anyone tried BING? How does it compare to Google?
Part of the problem with your search for a palindrome was your unusual choice of word (and spelling), namely "pangramic".
ReplyDeleteThe usual adjectival form of "pangram" is "pangrammatic" (this is the only form given in the OED, for example) and had you searched under that form, you'd have found the Google answers page on your question.
And even if you prefer the simpler form of the adjective, the expected way to spell it in English would be "pangrammic"—and had you used this spelling, Google would have asked you, "Did you mean: pangrammatic".
So I think the answer in this particular case is that search engines could do better at handling alternative and unusual spellings.
Try Comparing Google and Bing on this query: p np
ReplyDeleteThere are experts in my family on our history going back centuries, yet occasionally I find things on the web that people either never knew or had forgotten.
ReplyDeleteI use Google in order not to bother an expert. An expert blogging and asking for help is another matter, but I'm glad to see experts opening up.
And yet another option is to use a system like Aardvark, which, based on the terms in your query, searches for an expert who might be able to answer your question.
ReplyDeleteJust a link regarding situation (or the description of) number 2.
ReplyDeleteSearch engines may not ever be that good since as commercial products their goal is to make 90+% of the money with the least effort possible. If you're asking direct, college level questions, you're outside the 90+%.
ReplyDeleteWhat a coincidence, I just posted on my blog on "How to improve search". In it I casually mentioned that one approach is to use what I call crowd searching. One allows subscribers to amend one's search terms based on feedback, thereby gaining expert search for free, and guidance to more direct sources: http://josefbetancourt.wordpress.com/2010/01/21/how-to-improve-web-search/
ReplyDeleteI haven't looked at Ardvark.