So it seems to me the most pressing issue is still creating a viable alternative to Google Search. I've tried to use Duck Duck Go and even Bing in the past but I've just not found the quality of the results to be nearly the same.
However - recently I've found G Search is getting steadily worse. They've made UX changes that make it harder to view cached pages, and it seems to ignore search operators frequently.
This is the biggest foothold to their dominance. YouTube, Gmail, Maps and everything else is in a supporting role to this core product.
Who is actively working on this and what is proving so challenging? What is stopping us from creating a viable Google clone, with the same level of high quality results?
I've been using DuckDuckGo for a long time now and I find Google to generally be worse than DDG.
A lot of crap in the results (SEO doing his bit I guess), more pages taken down in the results and terrible UI.
The code preview cards are great for copy pasting from stack overflow without even opening the thread.
Bangs let you search something in another engine if you can't find what you're looking for.
The only times when I need to use Google are:
- checking opening times of businesses
- checking maps, as Google maps is the only decent thing out there
- I never use Google images but DDG images safe filter doesn't catch everything (you get nsfw in images - maybe this has been fixed, but I experienced it as an issue in the years)
1. Ongoing investment -- for all that HN harps on Google being an "ads company", they still employ XX's of engineers to work on search continually. For every thousand engineers, you're looking at nearly a billion dollars of continual spend to continue employing them.
2. SEO -- for every engineer working to make Google better, there are probably ten engineers outside working to make the web worse. Sorting wheat from chaff is actually easy -- you just drop it and let the chaff blow away -- so I'm not actually sure what the appropriate analogy is here, but it gets harder and harder is the point.
I’m not sure if it’s a difference in how I structure queries (I often structure queries with “”, site: limiters etc), what type of things I’m searching for (often technical/troubleshooting documents but also videos), or what. But I have found the opposite. In fact I have use DDG almost exclusively for at least about 5 years now, if not longer. Rarely do I need to !g a search against google (though it does happen)
You get used to Duck over time. Many of us have lived in pre-internet times. It was fine. You don’t have to always get the best search results immediately.
You're asking a lot of an average user, though, if there are two options on the table and one is always going to be worse. We don't tell people with AM/FM radios "Just stay on the AM stations; you get used to the AC wire hum over time."
And to tip the scales on the Google hegemony, we need average users on board.
Yea I really wonder why they don't have competition. You'd think a search engine built with java in the mid 90s would be easy to replicate at least to it's 2005 era standard using modern tech.
The unique value Google has is the mass of users who provide them with data, by clicking (or not clicking) their search results. It's the same way with most of the big websites - they wouldn't get to where they are if they started out today. But having all this pre-existing traffic is a huge part of why they're more useful than their competitors.
A lot of super technical stuff I'm googling at work all day I feel like is fairly static content that I think old school google excelled at. Modern Google is great for finding a local restaurant but not so much at figuring out the nuance in some searches.
We've been working on one such alternative - Kagi Search [1]. Kagi Search is currently in beta and we are also creating our own Maps app and a browser. In the future we will have email too so hope to offer an alternative way to experience the web.
Google's special sauce isn't particularly special: they have a couple solid signals by which they can tell that a given user didn't like the search result they vended, and they tune appropriately. Couple that with a little bit of user categorization ("our data shows there's significant difference between what users geolocating from the US and from Britain want when they search for 'biscuits'") and a history of collecting and refining that data, and you get a good search engine the same way Amazon gets a good product marketplace: sifting huge amounts of user traffic.
Google takes advantage of the network effect of being the hugest player, which is a positive feedback loop. A smaller player has less users to tune signal on and will adapt more slowly over time to user needs, on average.
> Who is actively working on this and what is proving so challenging? What is stopping us from creating a viable Google clone, with the same level of high quality results?
Search is hard. Organising the amounts of data that Google does is hard. Just look at Bing. Even with MS' resources it's terrible.
As for maps, look at Apple maps. Again, nowhere close to Google. There's also OpenStreetMap, Waze and probably others, none are quite on the level of Google. Definitely none can give you the experience of search -> navigation so quickly. Searching for and then navigating to a location on Google is just so much easier than the alternatives.
As for mail, at least there's tons of viable options there. Protonmail, Outlook, my personal fave is Hey.com (I hate email so their minimalist/focused functionality appeals to me).
Google got to their position by being very, very good at what they do. It's just a hard problem. Most other tech companies have tried and fallen short.
However - recently I've found G Search is getting steadily worse. They've made UX changes that make it harder to view cached pages, and it seems to ignore search operators frequently.
This is the biggest foothold to their dominance. YouTube, Gmail, Maps and everything else is in a supporting role to this core product.
Who is actively working on this and what is proving so challenging? What is stopping us from creating a viable Google clone, with the same level of high quality results?