I guess we’re not the customers. Use duckduckgo or something else.
I don't know how technically hard this would be to implement. Or what business model will resist adding it. I just know that most of my web searches are NOT me trying to buy stuff. They're usually looking up information.
Like, how does the load sensing algorithm work on my LG washer? No chance I'll ever find out by searching the web. Only 200 links to different places I can buy a new one. Or maybe some videos explaining that the feature exists, and "oh and would you like to buy one?"
Maybe I want to know if my GadgetMaster 3000 will work with my WidgetWanker Mark II. Nope, but here's 100 places you can buy those, or maybe the THINGDOODLE brand of knockoffs.
Ok, I'm ranting. Point is: give me a way to restrict my search to only those pages that aren't selling anything. Maybe the heuristic is that the page hasn't changed in 6 months, or uses outdated tags like <marquee>, or some fancy semantic analysis of the page copy. I don't know. I just wish search results weren't so utterly dominated by e-commerce sites or affiliate spam.
The primary purpose of ads isn't to get you to buy something. It's to plant a brand in your brain so that next time you do want to buy something, you'll think of the brand.
Ads are very good psychological manipulation.
Of course those are also really ads in a sense, but somehow there’s got to be a way to provide a real search engine that doesn’t show only e-commerce sites. I’ll pay for it!
Here's a thing that I am absolutely convinced happens. Ad network tracks purchases. Ad network track ad impressions. Ad network correlates the two and finds that 90% of people who saw the ad also bought the product. But they conveniently "forgot" to check if the ad impression was before or after the purchase.
I'm confident this is happening because the only times I see ads for anything I'd be interested in buying is after I've already bought the damn thing.
But do they at all take into account the fact that probably about 50 out of 100 ad impressions lead to an "I'll NEVER purchase anything from the purveyor of this ad-spam shit!" reaction?
> A built-in load sensor automatically detects the laundry load and a microprocessor optimizes washing conditions such as ideal WATER LEVEL and washing time. Turbo drum washing (option) When "Punch + 3" washing wings turn, the washing tub turns in the opposite direction.
I accidentally searched for this:
load sensing algorithm work on my LG washer?
Adding "how does the" to the beginning of the search gives this as the second hit:
> Water level sensing in LG washers is done differently from how you may be used to seeing it done in other brands. Whirlpool, GE, Electrolux and others use an air tube connecting an air dome on the tub to a pressure sensor with a physical diaphragm or transducer that "feels" the water level increase as an increase in pressure inside the air tube.
Which seems more promising, but I didn't click through.
I've watched it at work and as best as I can tell, it takes inertial mass readings by wiggling the drum using a specified current, and measuring how far the drum oscillates. Or maybe it predetermines the amplitude and measures the power required?
I'm just guessing. What I want to find out: is either of my guesses close? Is there something else I'm not seeing? Why does it do this 4 times in a row? (In between "wiggles" it will rotate the drum a few times to—my guess—redistribute the load and get an average from the four readings.
Maybe its rose-tinted glasses looking at the web of yesterday, but I have this sneaking suspicion some nerd out there has an entire website devoted to arcane appliance details like this. I would very much like to find and bookmark that site!
Hopefully DDG and others don’t turn into that tho
As long as you click on an Ad, google makes money, regardless if you buy something or not.
The only one I know of that at least plans to have a simple two-way relationship with search users as their customers is Kagi. Currently it's in free beta with a waitlist (took a week for me). When they go live they intend it to be a paid, subscription service with no advertising. So far I've found it good, and I'll gladly pay when the time comes.
They can grow their market by offering 10 free searches a day, and I think that would be enough to entice a lot of folks to switch.
Offer a paid API, and I'd bet they'd have a real growth story.
I hope they give me a easy way, to set my own priorities (but I would choose that option).
You can choose to boost particular sites, or to down-rank sites with trackers.
When did you apply? I filled out their survey at the end of december and haven't even received a confirmation email. I tried again a few weeks back with a gmail address, and still nothing. So I guess at least it's already at Google-level customer service.
There doesn't even appear to be any contact email aside from the discord which is also invite-only. Kind of soured me on the whole experience.
That makes this situation even weirder. Why not just re-send them, considering that they should have the collected email address sitting right there? If it's a typeform issue, then some sort of message on their site would have been nice. I mean it's one thing if this was a free product, but the fact that they are planning to provide a paid search engine but can't manage the task of sending email invites and don't have any contact address* just ruins any enthusiasm I had of being a customer.
* I realize the founder has his email listed in the FAQ page, but emailing the founder for an issue like this seems quite gauche. And besides I did try reaching out on social media but didn't get any response.
On the 70 searches point too, if you search for "business news" and then click the news360 tab to get news results, I believe this is actually counted as 2 searches so would get charged at 4 cents if in the overages category.
When I searched up I thought maybe $5 a month, or possibly $10 at a push, but $30+ will be too much for mass market adoption in my opinion. I also think there is an inevitable situation that someone searches a lot and suddenly finds themselves getting a $100 bill after they signed up for $10 a month!
Unfortunately there isn't a way around the price too from an architecture perspective because it's a high cost engine to run - the costs are driven by the fact that Kagi uses the paid Google and Bing api's, then adds a bit of it's own index and instant answers, but the Google and Bing API's both charge quite a lot per request (c$0.015 per request).
Search quality is usually pretty good (until it isn't), and it's getting better as it's manually tweaked to meet common queries, but the maps search is pretty much unusable. The search doesn't seem to "understand" queries in the same way Google does, particularly ones that are semantically hard (despite the results coming from google, Kagi reshuffles them). The founder, Vlad, seems pretty great though and is very active in the discord community. I really wish them luck, and particularly that they manage to crack the cost/pricing challenge they have.
However I have gone back to Google with adblock as my daily driver as I just didn't get the same quality of results, and found myself going back to Google enough times to make it frustrating. I appreciate it is a beta though, and that it will only get better over time.
Our $10 plan is likely going to give the user 1,000 searches. 80% of users search less than that monthly, even average HN users (who are majority of our beta users). We will also have pay as you go option ($1 per 80 searches) and $30/mo for people who want to support Kagi.
I would argue that our monetization strategy is valid as it is the only one that at least has a chance of making the business sustainable. Whether enough people would pay (which is what I think you were getting at) is another matter.
Yes our Maps is completely broken now. But at least we are building our own Maps (together with our own Search and a Browser, not an easy feat). The vision is ambitious, I'd love a world where at least a portion of it is realized.
Do you intend to make the search limit a daily thing, or will it only apply over the whole month?
For me at least, I have spikes in my search behavior, like when I'm working on some difficult project, while on other days there's not much going on. While having a search budget would maybe get some getting used to, I could see myself working with it if I had a way to plan ahead.
It would help if there was UI support for this, similar to what mobile phones have in the "Mobile Data" settings (data used/remaining, usage graphs and so on). I'm also on the beta right now, but haven't seen anything like that when looking just now.
Monthly for sure, even passing unused one to next month.
We are building a UI for consumption data right now :)
I suspect 80% of your users also are not daily-driving though, are also using other search engines at the same time, and have not set Kagi as a default on all their computers and their mobile devices (which I did!). In order to do that I had to change my mobile browser from Safari to Firefox, which is part of the reason I suspect most people have not done this.
1000 sounds like a lot more than 33 a day - particularly when if it is a search with a lens applied that drops to c16 per day, or a search that then clicks through to images.
I also think that the people that are willing to pay for search are likely to be the 'power users' which use search more. If your target market is 'people who are willing to pay for search, but don't use search that much' I suspect you will be limiting the market that find the subscription valuable.
> I would argue that our monetization strategy is valid as it is the only one that at least has a chance of making the business sustainable. Whether enough people would pay (which is what I think you were getting at) is another matter.
I think the problem is a monetization strategy is only valid if people are willing to pay for it. Anyone can create a monetization strategy that is designed to cover costs, but it's not going to lead to a sustainable business unless it generates more value for customers than the fee provides.
My point being, there is a lot of potential for "show the same search results again, because the user re-searched 5 minutes later" and "show the same search results, but filter locally on an additional word" too.
If Google/Bing give "1000 results per search", then there is also value in storing that single search too. I guess:
- I'm not sure how the API works, so I can't add much value here
- But, even though it adds local cost for storage and DB, caching searches per user could reduce costs in some scenarios
Because, while I love the idea of paid search... I think I do 100s of searches per day, so if you pay for each page of search results, and winnowing down results with additional terms....
> But, even though it adds local cost for storage and DB, caching searches per user could reduce costs in some scenarios
There may be some opportunity to do this, and caching results would be a good idea, but most forms of caching are explicitly banned from the Google and Bing API T&Cs. Some caching is allowed in some circumstances, but this will limit the usefulness of caching to reduce costs.
> Because, while I love the idea of paid search... I think I do 100s of searches per day, so if you pay for each page of search results, and winnowing down results with additional terms....
Agree with this - and at your usage levels plus $1 per 80 searches plus the $10 fee for 30 / day, you will be talking about $40+ per month, or $480 per year - enough to buy a new Macbook every 2.5 years :)
But IMO having been on the discord group the founders have got themselves into a trap of thinking "Charging per query is the only way we can make the sums financially work because we use both the Bing and Google APIs, thus we will need to get users to accept to paying per query".
The proposed pay per use is an independent tier, not on top of any other.
Pay per use is on all tiers if you go above the quota, the quota just changes based on tier. It’s been made pretty clear on discord that none of the tiers will be ‘all you can eat’.
A bottom PAYG tier is different to what I last heard although it doesn’t actually change the monthly cost of $40 for OP though.
There are people willing to pay for it, the question is will there be enough. A doomed monetization strategy would be one where people are willing to pay for it, but you lose money on every transaction.
> 1000 sounds like a lot more than 33 a day - particularly when if it is a search with a lens applied that drops to c16 per day, or a search that then clicks through to images.
Not sure where you got this from. Search via lens does not cost as 2 searches. It is costing 1 search and we are making changes in the future where it will count as less than 1 search.
So far I've been quite pleasantly surprised. I didn't need to de-prioritize anything yet, and most of my queries are technical. I would pay for it, but like you say, I'm not sure how much.
5$ for unlimited searches is something I'd go for without even thinking. For 10$, I already question whether is it actually better than using searx where I get comparable results. I can get a VPS cheaper than than and run my own customizable searx instance for that cost.
I'm not doing anything fancy, but I don't want to think about the query count. You can stack-up hundreds of searches per day sometimes, just because you're looking up something interesting, then do next to nothing for days.
I'm not saying I wouldn't pay for more, but I don't see the gain.
I definitely see the problem with the API cost of the underlying calls though. That feels like a huge argument that having your own index is important to compete in the market.
A Google size index that is potentially able to replace google results would probably cost $100m-$1B to build. We have a fraction of that built for noncommercial results (try Noncommercial lens in Kagi to see how it works).
I guess kagi is currently running at a loss?
My best guess is at about $0.5 per 1,000.
> I guess kagi is currently running at a loss
We are in beta and not chargining any money, so yeah by definition we are running at loss.
Unfortunately, I’m sure they rake in magnitudes more cash from ads than they ever would from subscriptions.
I do believe we should pay (a reasonably small amount! :)) for products that companies provide. It’s the “free” model that got us into this mess.
Though this kind of goes to the related question: why Google don't allow you to setup your profile to nuke domains from consideration when you search, I do not know - it would completely divorce them from any responsibility for search result manipulation and make everyone happier.
It's so bad you'd almost think google has a stake in pinterest at this point.
Such a blacklist could be limited to one's personal account though but, I guess, since there is no monetisation benefit for google they don't care to offer it?
The Majority.
Duckduckgo hasn't been any better for me lately. It seems like search as a whole is failing to respond to the latest round of SEO tricks. It takes a lot more work to find what I'm actually looking for, regardless of which engine I use.
If you are not logged into some of these sites you can't actually get to stuff either. Annoying.
Not sure why they are allowing them to take over the rest of search though?
It is inevitable. It's coming. Algolia? Please?
Disclaimer: I work for Google but in no way related to search, business models, or product decisions. My words are my own.
Hey now, SEO spam is often packed to the brim with high revenue ads.
Whatever it is it's really annoying because usually these results link to another search on a shopping site and more often than not those results come up empty anyway.
Like you see in the cached textlines: "Buy reptilecandy here, low prices!!" or something, and the URL to the site seems legit, trustedstore.com/reptilecandy/ but if you click on it you just get to the shop top page with no reptile candy in sight.
Clicking on one of the sites linked in the search results gives:
https://www.zazzle.com/reptile+candy+favors
And on their shop page it seems they have a section called "Reptile Candy & Favors". But if you click on the category above, Candy & Favors, there is no sub-section list so the sub-section is probably entirely dynamically generated.
So there's something weird going on, I wonder how those dynamic categories end up in googles search excerpts. Can they pay google to inject non-existant pages and sections into the results? (the search result in question was not marked as an Ad though)