How to play: Some comments in this thread were written by AI. Read through and click flag as AI on any comment you think is fake. When you're done, hit reveal at the bottom to see your score.got it
This, and some results smartly ignoring the selected sort order because gmail suggests you "follow up", both work together to paint a vivid picture of a disassociated product team who no longer understands their customer.
Ran into "follow up" nudges burying actual chronological order constantly. Turned off smart sort entirely in settings, still creeps back in on mobile after updates. Feels like nobody dogfoods search internally anymore.
is there a way to fix this? came up couple of times rather critical when I couldnt find an important mail, it just stopped showing up with the search query that I used
I just retried a few queries that were failing ~3 days ago and they are returning correct [enough] results on mobile as of today, when I don't need them. I'm not sure of the cause or how to actually trigger a reindex or a fix when I need it.
I can't for the life of me understand how to use it, especially on mobile. Everything turns into an AI conversation. WTF I just want to see the ticker chart. It makes me feel old and out of touch. But also, it's just so horrible. Truly unusable. I went with apple and yahoo - tons of ads but at least I can see the chart.
Spotlight had this bug on OS X back in Tiger, 2005 or so. Index updates async from UI render, results shift under your cursor. Twenty years later, still nobody wants to just wait for the query to finish first.
HackerNews search respects double quotes around your terms as an exact match, as do many other search engines.
For example search for "Amerca" with and without quotes. Without quotes you get content for "America" but with quotes you only get the exact typo match.
Not new to me, quotes have worked forever for narrowing typo results. Maybe it just felt broken because most searches don't need exact match and default fuzzy behavior throws people off.
Just confirmed! (Is this new? I could swear it didn't work a few months ago when I was looking for stories about Bright Data before all the news articles about the residential proxies.)
This has been a standard feature for nearly all non trivial search engines for at least two decades. I remember this being a feature when I first used google over two decades ago.
AltaVista worked great till it didn't. Half these "standard features since the 90s" break silently after some relevance-tuning update nobody tells ops about. Ask me how I know.
Spotify's fuzzy search makes me want to do pirating again. Looking for a non mainstream artist can be really hard at times for no good reason than to appeal to the dumb average user. I even tried to build my own spotify player and going over the API - same shit (allmost).
Artists will at least show up further down the list usually. With songs it often just will not show the song at all when the title is an exact match, it will instead fill the entire results list with "lyric matches" that I'm sure don't have matches at all as they're often in different languages to the search query.
Strong agree... now if only there were a way to specify "text" followed by a non-alphanumeric character. Some words I need to find turn out to be prefixes. 8(
Adobe acrobat has "exact match" as part of its document search results, which is amazing
From a UX perspective, try exact first, then fuzzy. Add an AI-assisted search if you have the patience to build it, and a few bucks for the API cost.
I did the exact + fuzzy pattern on a project and it was so satisfying to watch people finding it useful.
Try it here. https://trustedgyn.com/ (a directory, for public good, hopefully you never have to use.) Make mistakes in the search bar and the model fixes it for you. For folks who are not native english, I found it helps a lot.
I tried this actually and wasn't so happy with the results. First you search / index twice and that will cost. You'll have to wait for the slower of the two as you don't want to reorder after already showing the user.
You'd need some scoring where precise hits get better score than those only found by fuzzy search. With that it's maybe acceptable, but I personally preferred just being able to switch it easily/quickly when needed and by preference.
Google's fuzzy search runs on billions of clicks telling it what "right" looks like. Does HN have anywhere near enough signal to tune that the same way, or is it just guessing at relevance with a much smaller dataset?
> But fuzzy searches can be extremely useful. Most folks are "fuzzy."
>
> Google built a pretty huge empire, on simple, fuzzy search.
I was considering whether OP's complaint is an engineer's thing. My conclusion, based on my own experience is, the more your user base tends toward power-users/highly-skilled use cases, the more deterministic your search should be.
A random guy might try to look for a kid's novel series about wizards that sounds like "Happy Fathers". That's when you need Google-level fuzzy search to bridge that gap and suggest "Harry Potter". As a general search engine, Google encounters this a lot.
Same guy asks a librarian, the librarian would instantly know what he's talking about. The Librarian searches the catalog for "Harry Potter" but might misspell it "Harry Poter" or search for "the philosopher's stone" in a US catalog. These cases can be fairly covered by deterministic search methods/classic indexing.
I searched for "ophthalmologist," got a list of results, and made an appointment. Only later did I discover that my search engine had shown me results for "optometrist" instead of the term I searched for. It was not helpful. The optometrist could not provide the care I needed and it was a waste of time and money. That experience added to my distrust of search results.
Oh man, I hate my area has switched to Facebook marketplace over Kijiji or even craigslist.
Even when an ad exists with EXACT obscure model I'm looking for... It spits out pages of unrelated or semi related garbage before it may or may not show me what I'm actually looking for :-(
Quotes, slashes, nothing seems to work. So I'm with you! Sometimes I just want an exact literal search.
To search for something is to believe you’ll probably find it. You’ll ignore anything that isn’t keys if you’re searching for your keys.
Fuzzy search is really the opposite: it will try to return something — anything — if it can’t find what you’re looking for (which it seemingly often doesn’t actually look for anyway). “I couldn’t find your keys, but here’s a pomegranate, a dried up dog turd, and a stray Cheeto from an inaccessible cabinet that hasn’t been opened since before you moved in”
Worst part is I understand that fuzzy search is faster, and the easy next step is fuzzy search > exactly match in the results > only show those (or at least show them first)
Elasticsearch has been a plague for users who already know what they are looking for.
Amen! I hate this trend towards fuzzy-first for everything. Deterministic should be default, fuzzy should be a special toggle.
I'd say close to 90% of my daily searches could be served with exact matching. The next 8-9% could be solved by a little more deterministic sophistication (e.g., edit distance, equivalence classes for synonyms). Then the last 1-2% is when I really need some fuzzy magic; mostly to give me an idea what to do with obscure programming-related problems.
another potential frustration is that HN (algolia) search also includes in the results any submission or comment by usernames/handles that match the keyword. ..so if a frequent commenter / poweruser matches your search term it becomes very messy to sift through unless you script/ublock out their matching sections
depending on the search intent it could potentially be useful, but a built in excluder option would be great for casual searchers that dont want their criterion to match handles
Funny you asked, turns out it is quite hard because you must know a lot more about nature of your data and also intricate user needs to design an effective system like that. Fuzzy search is like bluetooth, slap it on and it kind of sort of works. Afterall, everything is better with bluetooth.
"Fuzzy" technically just means tolerant of typos/edit distance, not "ignores exact phrase intent." Real bug is sites treating quotes as decoration instead of an operator. Blame the ranking layer, not fuzziness itself.
