Hacker News Reader

Comment by manimino | Hacker News Reader

One thing I find amazing is that "single box" search engines are still the standard.

The biggest pain point of using Google is when it incorrectly guesses what you are trying to search. But the user is only allowed to type in one little box, so it's hard for the user to even express what they want.

If the user had multiple inputs, such as the ability to drill down, define general topics, etc., it would be way easier for them to express what they actually want.

A Google competitor that did this well could win.

View on HN

People don’t enjoy more complex search inputs. This can be readily seen at libraries where users strongly desire Google-style catalog search and don’t like being shown separate fields for author, title, keyword, etc. Worldcat hid that traditional search view years ago and, like Google, keeps it around for those who prefer it.

Increasing max query length seems more promising since it keeps the simple interface. Search expectations are changing as we Google more verbose errors and eventually expect to shape our search queries as we do our GPT prompts with extra moods, language styles, countries of origin, etc.

> Worldcat hid that traditional search view years ago and, like Google, keeps it around for those who prefer it.

In fact, it's so true that in order to find Advanced Search on Google and it doesn't occur to you to go into the settings menu in the top right (the gear - because why would anyone bury Advanced Search there anyway?), you have to actually google Advanced Search.

For Google to have buried it like this, Advanced Search must've been confusing the vast majority of people who tried to use it.

> For Google to have buried it like this, Advanced Search must've been confusing the vast majority of people who tried to use it.

Apple, the industry leading expert of UX, didn't have copy/paste in iOS for years. Compelling evidence that copy/paste is too complicated for people, right? Then one day they added copy/paste, I guess the nature of people changed over night and Apple merely responded to this new tolerance for complexity. They're the experts, after all.

Or maybe, [tech company] not implementing [feature] isn't good evidence for that feature being undesirable to the public.

> Apple, the industry leading expert of UX, didn't have copy/paste in iOS for years. Compelling evidence that copy/paste is too complicated for people, right? Then one day they added copy/paste, I guess the nature of people changed over night and Apple merely responded to this new tolerance for complexity. They're the experts, after all.

> Or maybe, [tech company] not implementing [feature] isn't good evidence for that feature being undesirable to the public.

Not sure Google burying advanced search is a good analogy here since Google once used to make advanced search easily accessible from the main page.

In fact, they still do depending on your user agent. Case in point: Wayback Machine's view from today. https://web.archive.org/web/20220909000247/https://www.googl...

The fact that its appearance is variable would seem to suggest it's a very deliberate decision to determine whether to surface it, unlike Apple just trying to force its own worldview.

My point is that these companies are not ruthlessly scientific user experience optimization machines. Google burying advanced search should not be taken as evidence that users didn't like advanced search.

Apple added copy paste two years after the iPhone was introduced

https://www.engadget.com/2009-03-17-iphone-finally-gets-copy...

> For Google to have buried it like this, Advanced Search must've been confusing the vast majority of people who tried to use it.

Or, pushing people towards the simplified "let us figure out what you want" showed slightly better ad conversion rates.

How about instead of doing really dumb complex forms we put our thinking heads on and try to optimize the process.

If you type 'Tolkin' the UI could suggest 'Author' and if you click on it adds it as a constraint. There are all kinds of things you could do by having a simple switch to complex search and have interesting options with a good UI.

I think something like that can work, but at Google scale, the suggestions would have to not hinder users who accidentally click them or think they did something just by appearing. Google has largely succeeded by trying to guess when the searcher means the author Tolkien without any UI prompt/change.

Again, I don't want to replace the initial model, but have a 'Complex Search' page that does this.

The issue is that sometimes you know you want to the 'Rust' programming language, not the 'Rust' video game. And in a complex search you could express these things and you could get much better results.

Sorry, I missed that you were proposing this for a separate search page and not the main search.

Isn't this the classic HN misconception, that complex, customisable interfaces ought to be popular?

Yes. Classic approach here is that exposing more behavior is better and that there should be rich formal languages for interacting with systems. Good for HN, more questionable for the rest of users.

They are though. The most profitable paid search engines have incredibly sophisticated operators and interfaces. Westlaw has a super sophisticated topic sorting system that goes back about a century. Lexis has something similar it generates by software.

Free search isn’t the only way. When your users all charge hundreds or thousands of dollars an hour and really need good information, they will pay a lot.

What’s ultimately limited is general user free search. Normal person time just isn’t very valuable and they mostly just want stimulation. Google is mostly a stimulation engine pretending to be a knowledge engine. It started off as more of a knowledge engine and over time optimized itself for providing a buzz.

Firstly, those are apparently geared towards legal research; I was replying to a comment about "a Google competitor". Secondly, B2B software (which I'm guessing those programs might be) has much less incentive for good, simplified UI/UX, even when it's obviously feasible and useful. The focus is almost always on customisations or new configurations that will draw in more leads.

Yes, this site skews heavily not just in favor of engineers, but the more old-school, "give me the box of tools and leave me alone" style engineer. 99% of humans who use a computer just want a single button that implicitly means "give me content" and they're satisfied.

https://www.google.com/advanced_search

This?

Yes, and very few people use that because most users can't tell an app from a website, spell URL correctly or tell you what "" does in google.

When I need those features I use the operators directly in the search field anyway.

The single omniscient search field is an amazing interface, it's incredible at what it does, which is understanding what our idiodic human brain is attempting to express with sausage fingers a monday morning.

> spell URL correctly

That's impossiboe by design. There are different cuaracters that look identical, foreign alphabets and special characters. Then there are GUids and queries in url

Oof. That's really complicated and still doesn't do the trick.

Concrete example -- when searching Reddit, I can check a box to confine the search to the current subreddit I'm on. That does a wonderful job of restricting results, and... it's one checkbox.

You don't need to make search feel like filing your taxes to make it better.

> Oof. That's really complicated and still doesn't do the trick.

> Concrete example -- when searching Reddit, I can check a box to confine the search to the current subreddit I'm on. That does a wonderful job of restricting results, and... it's one checkbox.

> You don't need to make search feel like filing your taxes to make it better.

I'm surprised at the Reddit example since Reddit's search capabilities are in-fact so awful that people have actually quit the platform for it or have at the very least resorted to using Google to search for specific Reddit threads (doable via Google's advanced search or simply adding the keyword "site:reddit.com").

In fact, the capability you described is present on the Advanced Search form. Look for the row "site or domain" and you can constrain your search to just specific sites.

I think the thing people miss is that Google effectively creates the function you're describing not just by having a multi-field search box (advanced search) but by allowing that to be achieved through just one text box (e.g "site:", "inurl:", "filetype:", and other key words), and for more tailored searches, Google provides entirely distinct UIs such as News, Shopping, Scholar (my favorite), etc.

I suppose I'm just lost as to what the main complaint is. I can't bring myself to work at Google because of how heavily they rely on adtech for revenue, but I can't hate their search.

That just creates the search with boolean operators for you... which Google then ignores. Attempts to search for an “exact string” are merely a suggestion to Google. They’d rather return 10 pages of definitely not what I wanted (i.e., wrong!) results than only return one or two hits even if those one or two hits are exactly what I’m searching for!

Nah, you can still force it. It treats exact-strings as suggestions at first, but you can still power through.

e.g https://www.google.com/search?q="search+enqine"

click the link that says "Search instead for 'search enqine'"

So Google understood my search query perfectly fine, but still chose to ignore what I asked for, and then makes me jump through additional hoops to get my initial question answered?

> So Google understood my search query perfectly fine, but still chose to ignore what I asked for, and then makes me jump through additional hoops to get my initial question answered?

Google's catering to the common audience, so yes basically. The hoops are for people who know what they're doing, which is annoying but understandable.

is there any proof that this has improved search for most people? Everyone i kniw says the opposite

improving search is not the main goal, rather it's keeping an interface that can still plausibly be described as "search" while allowing for the most ad impressions and clicks

Yes. Many years ago Google search leaders concluded it made more sense to optimize for what most people do: make bad queries. Ignoring quotes produces results that people click on more than respecting quotes (averaged over billions of users). This is also why the number one signal for ranking is user clicks, not something like page rank.

What percentage of searches require something this complex? I’ve been using Google for 20 years now and I’d be hard-pressed to remember a single time when I pined for a more complex interface.

The vast majority of my searches are simple, so I just do my first search and then (in order of frequency) find what I was looking for, do a second search that is either more or less complex than the first, or forget why I was searching in the first place.

Since the results from my first search are often enough, why would I want the interface to do that search to be more complex when the complexity is rarely needed?

When the subject is a generic or overloaded name, it's very common. For example, I've often done searches related to Smalltalk or Lisp (the programming languages) combined with names of various popular but generic libraries and get lots of results for small talk[1] and lisps[2]. There are various techniques you can use to cajole Google into giving you something closer to what you want but you end up losing a lot of the results that are the most useful due to terms you need to add to your search that aren't on most pages.

[1] https://en.wikipedia.org/wiki/Small_talk [1] https://en.wikipedia.org/wiki/Lisp

pay for kagi.com; it has “lenses” that define a corpus for your search

I'm trying it out for free and so far it's turning up great results. Thanks for the recommendation.

There's also Brave Search which has "Goggles" that perform the same function. https://search.brave.com/goggles

99% of searches are good enough with just a single text phrase. 1% needs more refinement. The current approach is personalization but I think having more dialogue like UX which can better capture the contextual information might be more suitable. But the blocker here is that our so-called AI technology is not there yet and those 1% searches are usually not very valuable in terms of revenue.

Reply on news.ycombinator.com