[jedwhite]

Thank you. While I don't want to distract from the topic here, I do really appreciate your feedback and you trying it out, and would love to talk more if you'd like to reach out! Speaking generally, the world needs more people working on search, and I think there is a lot of room for completely new approaches.

[aconsult1]

I'm pleasantly surprised by the high quality of your top results. That shows the search algorithms (whatever your choice was) are good enough and Google can be eventually disrupted. As a fellow software engineer I'm curious about your stack and size of your operations. Are you blogging about this somewhere? You definitely should!

[foxylad]

> That shows the search algorithms ... are good enough and Google can be eventually disrupted

I doubt this. I'm guessing that 90% of Google Search's development effort is concentrated on defeating SEO, and as soon as a new competitor succeeds well enough to attract SEO attention, their results will suddenly turn to custard.

The competitor then has to combat the extremely sophisticated SEO practises developed against Google over the years. This is likely to be an even bigger barrier to disruption than building market share against Google.

[jedwhite]

You're assuming here that Google's primary interest is consumers, but their paying customers are the advertisers. The reality is that the worse the organic results are, the more likely a user is to click on the ads instead, and that's how Google makes money.

At the least that's a perverse incentive, and at worst it's a corrupting influence.

Whether it is conscious or not, Google does better financially when there is more SEO spam in results.

That's why the better Google does financially, the worse the search results are getting.

[aconsult1]

I doubt this very much myself too! I little wishful thinking to be honest.

However you're arguing as if new competitors had to monetize the same way Google does and that's not the only game available. Imagine a competitor that doesn't monetize clicks but works on donations (like wikipedia does). If results are higher quality than Google's the users will follow. Or maybe a search engine that promotes a particular product like a CRM tool for example.

Disruption typically doesn't come in obvious ways or else someone else would have done it (including Google and its gazillion dollars).

[jedwhite]

Thank you. I appreciate that very much. We plan to share on HN how we've built it and how it works, and get as much feedback as we can. That really needs its own thread, and I'm conscious of not getting off-topic or promotional here because this is such an interesting thread already, and it wasn't my intention to distract from that because this is one of the big problems in the world right now. Plus, to be honest, we have a few things that are broken and that we're working to fix! So to answer your question, yes! We hear the feedback asking us to share more, and we're going to get a new release up and share with HN soon. And sincerely, thanks for the encouragement to do that. We think it will be interesting to the community here.

[jamiek88]

I’m impressed! I did a search for ‘Oregon coast gardening’ and bunch of indie sites I’d not see before popped up. Some blogs too!!!!!! Old school.

Why don’t you do a show hn so we can have a proper thread I’d love to learn more about the decision making of the AI.

[jedwhite]

Hey thank you. Seriously I appreciate the feedback and encouragement to share Andi with HN. I know there have been a few comments and questions, even though I'm trying to not get off-topic because this is such an interesting thread. You're absolutely right that this needs it's own thread where we can answer questions properly.

So we're definitely planning to share what we're building with the community here soon. Andi's very much an alpha and a few things are broken, but we're working on a new release that fixes them, and then we're going to share it and learn as much as we can from the feedback and wisdom here.

[girvo]

"And I" or "Andy" in terms of pronunciation? I wanted to also mention how impressed I was with the results, well done!

[jedwhite]

Thank you so much! And yes! To answer it's pronounced Andi like Andy :)

[pacoverdi]

I really like it too, will keep on testing.

Small bug I found:

   use the "go" command (e.g. go reddit)
   close the newly opened tab (e.g. reddit)
   click "Learn more about me"
   click the top left icon (Andi)
   => goes directly to reddit.com

EDIT: formatting

[jedwhite]

Thank you! And ugh! I thought we fixed that bug. On it :)

One of the secrets with Google is search is that the top searches aren't searches at all. Number one search on google? "facebook" - it's people navigating. So the idea behind that is to let people navigate directly when they want to go somewhere specific, like "go youtube cute cats".

[SubiculumCode]

Looks like HN killed it :)

[jedwhite]

Without wanting to go off topic here, may I ask what sort of error did you get? All looks fine from here, although it is an alpha. If you were happy to reach out I can try to help figure it out with you.

[SubiculumCode]

Secure Connection Failed: An error occurred during a connection to andisearch.com. PR_CONNECT_RESET_ERROR. Also Firefox is flagging it as "Not Secure" in the address bar.

Interestingly, once I turned off my VPN, it works again, Firefox doesn't display the message either. Note that other web pages work fine on the VPN.

update: It work's with MozillaVPN, but not with my University's VPN. Perhaps UC Davis has flagged it from within their firewall.

[jedwhite]

That's weird. The endpoints are on Amazon Cloudfront, so possibly uni VPN didn't like the CDN it is on.

[SubiculumCode]

Also, I set it as my default search engine for a trial since my first few searches went well. Is there a place to submit bug reports?

[jedwhite]

Oh awesome, thank you! Yes just say "/bug" any time and tick the little box to attach details (we don't see what searches are so that let's us know) or just hop on our Discord from Contact Us. We really appreciate you trying it out. It is an alpha and we have lots of work to do, especially performance. I've been trying not to go off topic from the conversation here but wanted to answer your direct question, but we can chat more directly.

[SubiculumCode]

I love the HN view option...mostly because I find that form so easy to scan.

I've conducted a number of professional searches about neuroscience, obscure R packages, topics on D&D and Pathfinder, and it has done exceptionally well. The one thing my obscure demographic would love is a replacement for Google Scholar, which I use all the time.

[jedwhite]

I love the HN view too - it is compact, efficient and information dense. I think it works well for search results. And I love HN so it seemed like a neat homage :)

Thank you for trying it out and the great feedback too. There hasn't been too much discussion here about Google Scholar, but better research and academic results is something I hear a lot of people talk about as a growing problem.