The quality of Google search results has been awful for many years now, but if you still want to use it, the only usable way is via a frontend like Searx[1]. Using any of Google's frontends for any of their services is an exercise in frustration from dodging ads and fighting their hostile UI.
Generally great post, but did he say Gmail was a Hotmail clone? Really? Did Cory get too old to remember things or is his anti corporate paranoia that bad?
I still remember the gmail announcement: it was on April 1st and near everyone just thought it was an April fools joke that they promised a GB of space when Hotmail gave 5 MB. And their conversation UI, no delete button, labels: they innovated on every single aspect of how you could do email. Oh did I forget to mention it was the first true web application that was mass market end user focused? And did a killer job at it too?
At one point Google was a true innovators company. Failing to acknowledge that shows Cory’s own biases and continues to relegate him to one loud corner of the internet. We get it Cory, you loooove GNU. Doesn’t mean you can’t objective about other things.
craigslist is one of the few sites that has stayed the same and is still serving its users. The owner got kinda rich with it, but refuses to do anything more to the site. It's amazing.
The most annoying new “feature” is that every for sale search on my local CL also returns “more from nearby areas” results from New York, which is hundreds of miles away.
Depends what you use it for. Searching estate sales for shippable small stuff, fine. Shopping for cars or romance not so much.
The problem is the meaning of distance is relative by culture. 250 miles is nearby in American west. Some people drive that far for groceries, a date, work even. Without leaving the county.
Meanwhile from say [checks map] Springfield Massachusetts, same 250 miles spans eight states. Half of which hate each other.
Now try buying a car on Craigslist in Hawaii, where half the listings are on the wrong islands. Part of this is CL's fault and part is the spammers mislabeling location (which is also CL's fault, indirectly).
In the parts of UK or EU where people don't have cars, search radius of 250 miles is worse than useless. Some people don't travel that far their entire lives, unless there's a war on. It makes the whole site feel like buggy American crap.
californian: it is not. that's a 3-4 hour drive and can get you from Los Angeles to Las Vegas. If I'm driving that far to meet someone, it better be some crazy rare item that can't be shipped, not for groceries.
Even as someone raised in the rual parts of California, the nearest walmart was "only" 40 miles away or so in the 90's.
Yep... I actually use(d) those limits because sometimes I want to find older news about a topic, sometimes an entirely different event than whatever is popular at the moment with many of the same keywords.
But no, now that the market has been oligopolized, it's more profitable to show you what you don't want, and you'll have to be satisfied with whatever incidental successes you manage while wading through the output of the Search^h^h^h^h^h Sell-You-Things Engine.
Imagine trying to find pertinent information about MetaCompany renamed themselves, or searching any term after any sort of tool or company took a common word and got popular
What's up with software becoming less powerful (useful) over time? This is a wider trend that has presumably been accelerated by the shift to mobile apps (it's crazy to me that you would need different features based on the platform you use, but this belief seems to be wide spread).
"you don't need that" is being taken to the extreme. Designers with simple needs, not imagining that somebody would need more?
I have to switch to the desktop view to use date selectors on mobile. It feels like almost inevitably at some point someone will go "of our entire user base almost nobody uses it, so we should remove it!".
Recommendation algorithms are another field. Presumably they have tons of parameters. And how many do they expose? 0. Leading the algorithm to be mis-tuned for how I feel at a given moment ("give me new stuff" vs "show me something familiar").
YouTube's "related videos" have recently completely gone down the drain too. They've always mixed not actually relevant videos in there, but now they're ONLY showing them, leaving me without a way to actually find related videos.
Am I just an old man lamenting the world changing or does this type of stuff seriously feel inhibiting to anyone else?
This trend is one of the major reasons I prefer free software - some kind of respect for users' needs and the theoretical option to change these things.
At least somebody at YouTube figured out that only being able to sort channels' videos by "Latest" and "Popular" is silly so couple of days ago they introduced or reintroduced sorting of channels' videos by "Oldest".
Also few months ago or more, they were experimenting with being able to sort videos' comments by "Timestamped comments", which is actually quite smart because people timestamp videos in comments a lot and viewers are indeed interested in clicking timestamps when they see them in comments.
OP or mod should edit the title and add [2019] and btw you can sort Google search results by date. This was probably one of their experiments to see how would users react and start behaving without being able to sort search results by date. But tbh, most of the Google users (casual users) don't even use sort by date.
Though it's still the case that it's missing in the Android Google app (which is the interface called by the main search pill on most Android homescreens).
Despite the web app looking almost identical, save for the presence of the date drop down.
I just tested it on my Android phone; when you search for something in Android Google app, you need to swipe right and then tap "Tools" and you will see "Any time" option/filter. Unfortunately "Custom range" date filter is missing but others are present.
I've noticed this come and go over the past few months and just assumed it was some A/B test. In most cases jumping to another browser was enough to get it back.
I take it this is looking more widespread and/or permanent?
Yes, polluting the information ecosystem with unverifiable AI garbage results would be the dumbest thing humanly possible, I expect Bing to lead the way here and Google to scramble to ape them.
On the upside, the self petrification of giant redwood companies creates space for new businesses to create actual search engines.
Over the last few years it's became an outright necessity to use search operators to find what you want on Google. Back in the day the only skill you needed was how to phrase questions to find the best results, but nowadays I spend a lot of my time telling those around me about site:, before:, etc.
This was part of one of their A/B tests....you can still sort Google search results by date. Speaking of YouTube, YouTube still is not good enough features wise. I'm a software maximalist and I would like to be able to search comments on videos and more importantly being able to search your YT playlists. Also there is no search feature to search channels' videos on mobile unlike YT on desktop (web version). But the most tragic thing about YouTube is its main search engine; it just sucks, it gives you very few relevant results and most of the other results are recommended garbage or garbage that is not even remotely close to your search query.
Speaking of recommended videos and their corresponding algorithms. One huge thing for YouTube would be feature that would allow you to tweak your recommendation settings. For example filters like: Recommend me only videos in English or recommend me only Gaming and Music videos etc. There is a lot of room for improvement but Google is public profit driven company so they have to balance their expenses and chase revenue which in return means serving users "en masse" and not focusing exclusively on the quality of the product.
In my experience sites were pushing updates to their pages to make a 5 year old article appear as if it had been updated just this week, so as to prevent filtering out such content for being irrelevant.
So to some extent the value of the date function was already being destroyed by corruption, but I did still get use out of it so I think this is premature.
It feels like Google have got to the point where they've actually given up on an algorithm and have many hand curated lists of results which they serve up for any vaguely related search so now it's impossible to add nuance to your searches as they all get lumped into the same bucket.
I can imagine that 'sort-by-date' searches are both significantly more expensive for google than regular searches, and used rather infrequently relative to their main search. As such it's no surprise they would end up on the cutting board at some point.
I'm not convinced this feature was ever particularly useful. A lot of web pages don't have meaningful dates associated with them. (What's the date of the HN home page, for example?) Worse, a lot of pages which do have dates have incorrect dates -- either ones which are always current, or ones which represent a date that isn't relevant to the content, like the date a dynamic web page was first created.
> A lot of web pages don't have meaningful dates associated with them
All sites indexed by Google have at least one very important date associated with them: the date they were indexed, and by extension what they looked like then.
And you can bet Google isn't throwing away this information for their own use. Temporal signals give causal signals, and casuality is extremely important for AI.
I used the feature semi frequently when trying to find data from a certain time period and being able to filter to only results that had a page unaltered from that time was very useful to filter out false positives. Sure I lost some hits like what was the front page of hacker news or Reddit at the time, but I’m usually looking for blog posts when I filter by datetime
Google bot is really good at crawling new pages within days, sometimes much faster.
Therefore, it’s useful because you can sort by approximate date of the content being crawled. It doesn’t need to be the exact date of the content being published. Just approximate is better than nothing.
That is a bad idea, because some articles are updated constantly. I have one of them. Google bot saw it around 12 years ago for the first time, but I have updated and rewritten it a thousand times.
This isn't journalism, where articles never change. It's providing current information about a topic and the URL is a shopping window – it changes.
[1]: https://github.com/searxng/searxng