Assuming that people don't clear their cookies is wrong nowadays. Even when your website is a little snowflake, the rest of the web isn't.

I honestly would recommend a Web Extension that adds your search as a search provider, or something like an Open Search Description.

Also: the landing page of GitHub annoys the shit out of me. Every damn time. I don't wanna register with three clicks if I already have an account.

Make a landing page easier for regular users to login, and prioritize login over signups...otherwise those regular users won't be happy. Any company should focus on keeping the regular users happy (UX wise) and not focus on one time trial registrations that will be a burden on infrastructure.

Fucked up incentive structures like this are why we cannot have nice things.

I strongly suspect that this has to do with attribution and shallow "data driven" choices - there is a heavy bias towards things that are easier to measure. Especially in startups with limited resources.

Measuring retained users regardless of anonymous/known state is very hard and lossy. So the nearest easy to measure analog that is often picked is logged-in users and we start the fucked up cycle you mention.

Similar things happen around all metrics that have to do with retention. We know that a better experience through higher performance and better UX is likely to create happier users that return... but we cannot prove it (easily) so we focus on optimizing other things.

The number of users you lose by doing that is smaller than the number you gain as long as you don't have enough users (and spends enough on publicity). This is also the one thing that limits the number of users a company will ever have on most cases.

What I do mind is the price...if you read the FAQ, they want to charge as low as $10/month, $30/month for families once it exits beta. ("As low as" means "possibly higher")

That's very high. The family price equates to 16 Office 365 Family seats, which includes 16TB of cloud storage.

In my opinion, the price for this kind of product should be more in the realm of $50/year. I'm thinking a price closer to what password managers charge.

This price also will need to be adjusted for global customers if that is the eventual intention. Spotify in India costs less than $1.50 a month.

It is also an independent company, bootstrapped by the founder, without a ton of VC money to subsidize the cost. And at $10/mo per user it would be barely breaking even, if at all. So there are a lot of factors of play, and I completely understand why $10/mo may sound excessive given all the free alternatives out there. It is just one of the many uphill battles we are fighting ;)

Right now you're charging nearly as much as something like Netflix, so your product has to be nearly a Netflix' worth's more value than DuckDuckGo (rather than Google as it's closer to your value proposition) for me to rationally justify the price.

One thing that's great about search engines in terms of entrepreneurship fundamentals is that is is such a massive market. You only need a fraction of a percent to be a viable business, I'm sure.

I can also understand how underpricing a product can be a big mistake. Somebody who wants this kind of product has already made the conscious decision to pay. Pricing a product too low can be a signal that it's not a good product.

Perhaps one might say that this is an example of how complexity and cost of production is a bit detached from the customer's willingness to pay. The most complex software product, the operating system, is essentially given away for free. People are mostly not willing to pay for it, even businesses don't want to anymore. On the other end of the spectrum, there are SaaS products with very low levels of complexity pulling in a lot more than $0.

If it's truly barely breaking even at $10/month, that fascinates me. It must mean that search engines have to do a lot of computing.

In any event, I wish you the best, because the features do look cool. I like the idea of being able to set site preferences (e.g., when I search for a movie I might like to see Wikipedia over IMDB).

Spot on. Not that long ago we used to pay for OS, even beta versions of OS. Nowadays, most complex software products imaginable, like OS, browser, search engine, social networking... - are all free, and over the years we as a society have been slowly drawn into an expectation of these to be free. Of course someone or something is paying for it, as all these also "magically" generate hundreds of billions of dollars of revenue. And when someone shows up with an idea to actually sell such product and tries to create a business where user is also the customer, and incentives between the user and the business are aligned, they are usually looked at with a raised eyebrow.

I am personally of the opinion that you always get what you pay for and that one should not shy away from paying and supporting products they love. I hope we are building one such product, that some people (and admittedly you need to be a little different to use and pay for Kagi) will fall in love with.

Is it that you're doing a large amount of compute per search which yields better results but higher cost per search?

[0] https://news.ycombinator.com/item?id=2781133

[1] https://news.ycombinator.com/item?id=13843743

On premise search. Businesses have lots of resources on their internal websites, but no easy way to search them in one place. Addition of role based search, when I'm logged and search only shows me resources I have access to.

$10 a month is less than one lost hour of work.

I hope they price it around there somewhere so that many will use it but even at $20 it is a no-brainer for me.

The results are in another league compared to what I get elsewhere and by now it also feels like an insult every time I realize Google has altered my query: do you guys really think I don't know what I am doing?

Google probably works better for illiterate people these days, but in the process they have alienated me. I have lost too many hours searching for things I knew used to be there only to be ignored and corrected by their AI. It is rather patronizing at this point.

This is a pretty feeble argument for something, especially for a product where the user perception is that it should be free, and there exist multiple free options already.

While "one hour of work" might not seem a lot, a different way of looking at that is that for most people there are 160 hours of work in a month, and so this thing needs to be worth 1/160th of their life.

It's not so hard to make that choice for a couple of things that are of obvious value, but when they start adding up in your life for basic things, people will question the value of it to them, and I'd argue for most people, search just isn't worth that much to them when there are so many good free options.

You should still use the website's search function first. Github issues/SO search often turns up results that google will ignore.

Offtopic: That must be one of the best deals on online storage Excluding Google's enterprise unlimited storage option which always felt like they could pull the rug under you any day.

It's necessary to prove an app brings more joy than a good $5 drink does--most products fail to cross that threshold.

I think I wrote something like "if a product requires a subscription I don't consider it unless it replaces another subscription - or is so good that it is obviously worth it".

Well, for me this is the second category: For me Kagi is a fantastic deal at $10 (although I hesitate to say it since it probably means they'll end up charging more ;-)

For me there is absolutely no difference in paying $1/month and $10/month for a daily driver search engine and I’m letting everyone know this to establish dominance over OP.

In other words, shave off a year of your career at least and retire instead.

$5 tea or coffee is just flavored water that I can make at home for 50 cents and minimal effort.

Obviously, to each their own, but $5 tea was a bad example.

And a $10/month search engine is a bad deal when ad blockers are free.

Even if Google was completely free and didn't track me and didn't insult me with their ad matching, paying $10 a month for results that are in another league is fine with me.

Still, $5 for a cup of tea is a really terrible example because it's an incredible waste – tea is very easy to make. Technically, when you buy some tea somewhere else you're wasting time rather than saving it, since an electric kettle is much faster.

A mildly better example: at least when you pay $5 for a Starbucks nitro cold brew, you're getting something that isn't realistically available at home (unless you want to buy/clean/store/operate a nitrogen coffee kegerator).

$10/month for a search engine...I mean, you can just block ads and trackers for free. Microsoft literally pays you money to use their search engine. It's not like streaming services where paying for ad removal saves you time.

This is one of the first "asshole design" moves that I recognize, that unfortunately most sites use. If the "Sign up" button (which will be used once by each user) is highlighted more than the "Sign in" button (which will be used hundreds of times), I know that your company doesn't really care about its users.

I'm surprised how common it is.

?

If you're not signed in, `Sign in` is available on literally every page in the upper right and one click takes you to the sign in box. If one click is too many just bookmark github.com/login.

If you don't want to sign in the search box is also in the upper right on every page to take you wherever you want to go.

This will make me be detected as a bot and I see captchas everywhere afterwards. I was using that for a while.

A better alternative would be content identity through URLs.

If your marketing page is at github.com/index.html, then make something like github.com/dashboard.html for logged in users, and redirect them to /login.html when they are required to be logged in.

This way users can easily bookmark the page that they actually want to see.

The sole purpose of these is to get you to globally log into google and stay logged in.

So I don't think it really makes much of a difference.

Curious how would you handle authenticating for a paid product on the web today?

Edit: I do authenticate on several paid products, but I was always introduced to them in some hassle free manner, and only got to the point of paying and dealing with adding another app to deal with maintaining after I’ve grown to like the service

Ultimately it’s laziness, but dealing with authentication is annoying and I can’t be assed to do it for something I’m not already valuing

After I made the password manager a part of my routine that I trust absolutely that I never skip, all of that anxiety went away. I still have logins from the before times that I wish I could round up, but for the last ten or so years I can guarantee I know every site that I’ve signed up for.

I feel bad for people making new services now who have to deal with this, but the internet isn’t new and they have to overcome our experience with all the bad behavior of shysters who operated like these apps and ended up lying about the service

As of now, I log in only when needed, and usually to a single account at a time, at most two.

I am willing to try new models for web search if the provider does not log any information. I'll prefer it to be stateless (other than account settings and payments).

This. I suppose it's become so prevalent due to A/B tests finding it more efficient to optimize for new users, but at some point in a company's growth it would make sense to put the focus on user retention. Then again, it wouldn't be that hard to make both signing up and logging in easy and attractive.

It is just like being back in 2007 or something when Google worked this way and it is wonderful.

I've used DDG for the past few years and I still hope they too will succeed, but with DDG the quality is the same or (slightly) lower than Google and the bang operator comes in handy many times a week.

With Kagi I only think I have used the bang operator once (yes, they have it). It is that good: when I look at the results I know immediately that there is no way Google or DDG can do it better except by pure luck and also my results are already there on first or second try so why bother shopping around?

Bonuses:

- bangs work like in DDG (I think, I don't use them anymore)

- doublequotes actually work unlike in Google and DDG

- even without doublequotes it doesn't stray very far from my query

- I can add lenses (?) or something (I have not bothered yet, the defaults are good enough)

Yep, I'm a walking billboard now, but I'm not paid in anyway and don't know any of the people behind kagi, it just happens to be the biggest improvement I have seen since Google replaced everything else sometime around twenty years ago.

Edit: yes, it is a paid product but it is free during beta and so good that I would consider Kagi gift cards for friends in tech if it was paid, just like I'd consider gifting WhatsApp subscriptions before Facebook bought and subverted it (yes, I was considering worse words).

Edit 2: I'll do this for your product too if I love it. I did for WhatsApp and Google+ and in real life for the insurance company I have used the last few years.

Lets be clear that this goes both ways, Kagi is a really nice Christmas present: so far this is the biggest software improvement I've seen in a number of years: on the same scale as when generics was introduced to Java (and this was a big deal for me) or in fact even on the same scale as when I was introduced to Google. I'll have to keep using it for a few months to see exactly how big it is but feels like a really bif deal so far :-)

Wishing you a merry Christmas whenever you celebrate it (we are celebrating tonight).

This drives me nuts. If you're going to have a competitor to Google (ddg) at least give me proper functionality to refine my searches

There is a lot more going on than 10 blue links these days, including continual scroll on mobile. Perf versus features is always a matter of trade offs.

Okay. I'll bite. What does this mean?

I hold my iPhone 13 the exact same way I held a Nokia brick in 2003. What did Apple tell us?

They were HIT WITH 18 LAWSUITS about it OMG.

They settled all the lawsuits by ... offering a free cover or $15 refund.[1]

Depending on your viewpoint this is either one of many blips along the way of manufacturing tens of billions of products over a dozen generations, or a way for "mactards [to] realize Steve Jobs are out to milk their clueless a$$es with his oh so clever marketing gimmicks" (Yahoo News comment).

Bonus points if you can find anywhere Apple said "you're holding it wrong" or blamed users. Steve Jobs wrote "Just avoid holding it (the iPhone] in that way." which is somewhat pragmatic, and rather what you'd expect someone to say if they weren't filtered by a PR firm. Probably what I would say. Apple haters can't face this.

Macworld.co.uk said:

> "We don't really think there was anything substantial to the problem, and the iPhone 4 sold tremendously well. Customers didn't seem to notice, let alone mind, and Apple repeatedly pointed out that other mobile phones suffered the same effect."

[1] https://www.cnet.com/news/settlement-reached-in-iphone-4-ant...

[2] https://www.macworld.co.uk/feature/antennagate-11-biggest-ap...

Apple aren't my favourite trillion dollar company and just because they have money doesn't mean everything they do is bad; or that the fundamental attribution error applies differently (similar applies to other companies, groups, organisations)

You’re looking at cold start download sizes, but that’s not what people actual see after everything is cached follow the first load.

Google pages load plenty fast on every device I’ve used recently, including some very old android tablets. Most page loads don’t transfer much data at all due to caching. Combined with the high speeds of both browsers and devices today, there wouldn’t be much real change in a highly stripped down Google page relative to what they’re serving up today.

Here is what I get:

https://imgur.com/a/Dl2B5Zo

2 seconds to load and 1.12MB transferred, all on "warm" start.

I'm also seeing content loaded happen in under 1 second.

Where did the love go....

This. Nothing else matters.

While it is true, for almost all of the users most of it is cached. For me it is using 2.5 MB resources with 77kB transferred.

I really like the idea of a premium search engine too, I'll give it a go for sure.

I'm a little uncomfortable with having to be logged in though... Ideally if I'm going to pay for a search engine I would like to do that as anonymously as possible.

Best of luck with the project, I'd love to see some competition in search again. Google has been getting worse for years and they have no real incentive to improve given their dominance and lack of disruption in the space.

I'm not sure that I want to sign up and give you data when your premise is to not mine the data. Is that your premise? How long do you retain knowledge of what had been searched?

The personalise features put me off. The level of personalisation I want is course location... I.e. online shops in my country, or if I search for something in Liverpool pick the right Liverpool in the world... Both of these are relative to where I am and an IP geo lookup is roughly good enough with an override in the worst case.

I'm not sure precisely what your differentiation is (again, can't find it without signing up)... But I want to be very sure before I sign up to anything that you treat user data as a radioactive resource. Thankfully given that GDPR exists and everything done on search is related to a user profile I'm assuming that you're petrified of storing and using data... But I don't know as the site didn't reassure on that front and I didn't see the privacy policy at the foot of the home page.

I want to like what you're doing so much... But you've designed it to have friction to that discovery.

I'd echo this. I actually even went to the site to see what was so good about your search, then realised I can't even test with a single query without signing up, and lost interest.

Perhaps it is not obvious that Kagi has not launched yet. We are in closed beta, and email is used to invite people who want to be beta testers, nothing else. Once we officially launch, everyone will have the ability to try it.

>I'm not sure precisely what your differentiation is (again, can't find it without signing up).

The FAQ has a lot of information you may find useful https://kagi.com/faq

My other feature request is that I still like google maps, and OSM is just not a good replacement. The important feature for me here is: google a place name, click the map square that shows up in the Answer for that place to open it in gmaps. Both kagi and brave search use OSM which is not clickable, this really breaks my UX and alone is enough to make me want to go back to google search. Using !m sometimes seems to work but I think ideal would be a "which maps provider would you like" feature which just lets a user choose.

Kagi Maps is still work in progress (beta like everything else) and we just didn't get to shipping this feature. We are not using OSM only though, and many users have complimented the map to be better than Google maps already (for features that we have shipped).

What about a version of that where I pay a few cents for each search, using Lightning Network, for example?

If you just made those parts drop-downs with the most popular answers and had an 'Other' option that makes a text-field option, you would probably get much more valuable data and more of it. I would not be surprised if you had some amount of dropoff in your beta signup funnel due to questions that make potential users feel like theyre being 'put on the spot'

Accepting crypto payments could be a solution but I'd totally get why'd you want to avoid that.

The best we can do really is have a business model that aligns incentives with the users and weirdly enough, this is already an innovation in this market. Then, create the best possible product that users will love. Gaining trust is a process that can not be skipped.

I don't think that's an accurate view of the business model of Google. For them consumers are mostly locked in and their business model is in tracking users and mining as much information as possible from every facet of the experience.

Best of luck, looks really ambitious!

That is the problem: resources show up in abundance, not in scarcity. Only small or poor creators usually can do the latter.

Conways law at work, I guess.

Or I can use duckduckgo without needing to sign up and get your promised security with 1-2 second searches for free. Maybe you'll have enough takers to get off the ground but I'll rely on other tools.

> forces you to login

Turns out, this might not be true at all.