Hacker News new | ask | show | jobs
by keiferski 2687 days ago
The "must include" functionality makes zero sense to me and I'm dumbfounded as to why it was ever added. If I didn't want my results to include a particular word, I wouldn't have included it in the search. Boggles my mind how Google could mess up something so simple.
5 comments

I'm dumbfounded as to how you don't see the usefulness of this feature. Sometimes there exists more than one word for a situation, and you might miss a forum thread, for example, that is relevant to you, but you're saying to Google "only use these exact words" and not accounting for differences in dialect etc
Honestly, you're both kind of right.

It's totally correct that often we want search results that don't include the exact word which was specified in the search request. We often want close synonyms, stems, elimination of redundant words etc. and flexibility on this helps to retrieve much better search results.

On the other hand, Google has a particularly frustrating habit of dropping words which are absolutely key to a particular search. You know, those cases where you are searching for details about a very specific set of explicit keywords, and 90% of the results exclude one of them, making them totally useless.

It's like an "uncanny valley" in some ways. When it says "You probably don't need the word 'and' in your search, and also 'video games' is the same as 'computer games' so I'll just include both" then the process makes sense, feels helpful, and indeed you probably don't notice. When you search for two explicit technologies and one of them is ignored, it feels like you're fighting a system that's trying to second-guess what you mean.

Ultimately the search engine is tuned for common queries, and the majority of users aren't likely to have the required technical skill to express exactly what they want to find without some help. This results in mismatch between the expectations of different users.

Verbatim search helps with this, in that it allows advanced users to better control their results. I'd be super happy if I could just toggle this "on" all the time.

Because often the key word is the one being dropped by Google. If I search for "Einstein quotes about social democracy" and it drops "Einstein" it defeats the entire purpose of the search.

A single button that says "try searching without word for more results" would be far less frustrating.

I search for "suicide" (for my work) very often. It _fucking sucks_ that Google seems to think "death", "died", "murdered" etc are all valid synonyms. It means that for every single search I do I need to do extra work.

I miss the days of +forced +terms.

Surely using quotes around the term does the trick?
Because people are dumb and type in entire questions or sentences, so google tries to be "smart" and remove words that hold no real meaning.
Actually I think it's useful to google whole questions - for forum thread titles. The problem of course is that Google does not search for the whole thing and wrapping it in quotes makes it useless as well.
Asterisks within quotes still seem to work. I use those when I have a good idea of what string I’m looking for but I’m not sure about some words or want to find posts where someone (for whatever weird reason) apparently typed in or pasted an error in chunks.
One of the original defining features of Google was that all results included all terms. Now even when you search for +”term” you get results without “term”.

It seems just stupid to return these completely useless results but I guess it makes them money somehow.

Search engines - Google amongst them - are moving from using the search term as (possibly stemmed) literal words to search for to attempting to interpret the 'meaning' of the term. This interpretation can be somewhat... lacking, with the observed result of nonsensical answers.
They have definitely messed up in many ways, but having a must-include operator was not one of the ways.

It allows (allowed?) you to tell the search that the other words can be used for ranking and are nice-to-have words, while the must-include words are treated differently and must be present and match exactly.

If all search terms were automatically must-include with no override, search results would be garbage because as a user you aren’t going to predict the perfect inflected forms of words used in documents every time.

Being dumbfounded should not lead you to think other people are wrong. It should lead you to wonder what you are missing.