Thanks for sharing. I have been working on vector search engines (IR) for a while (Aquila Network). And I believe works such as this as well as the "differential neural computers" (couple of them from Deepmind) will be the next breakthrough in IR. I can't see the direct path yet. We're eagerly waiting to see somewhat a usable architecture yet. I believe current vector indexes will eventually get modified into hierarchical random access memories that stores compressed information in higher dimensions (static & replicated part of the distributed system). On top of this, an application specific information decoder (that's the dynamic part of the system, UX) will use the underlying information accordingly.

Thanks for this article. Adding to other "Google is dying" discussions that I have collected over time as part of my personal research: https://docs.google.com/document/d/1cSMY5wXSKhJdMxeJEvTUJ21e...

It's not dying that much since it's obviously your preferred publishing platform.

Yes that might be true, I'm just collecting information.

Agree, Mullvad provides really good VPN service. I faced almost zero downtimes / speed throttles. It establishes quick connection with server (maybe because it uses wireguard). Anyway, I'm a regular user and I think paying 5E worth it.

> It establishes quick connection with server (maybe because it uses wireguard)

I'm actually kind of curious about what Wireguard does here. I think Wireguard says it's connected almost immediately even when it isn't, presumably holding traffic back locally while it waits for the connection to be active. I was wondering because I spent some time confused by a non-Mullvad Wireguard connection that wasn't working (turns out the server wasn't available at all) that nonetheless appeared as "connected".

network-manager UI, if I recall, just shows that the WG connection is turned on, not that there are packets coming back. (It's not a client-server, so this "makes sense")

Seconded. Very occasionally I'll have to swap servers in a location but that's super infrequent and it's not exactly a primary tier location that I'm using. One of those things I can generally just setup and leave working.

In my opinion, Google's way of doing information organization worked really good in the past, now it's time to move on. The same system that separated signal from noise is now helping to generate too much noise while trying to fight it at the same time.

Search should be the native feature of the Web itself (even if it's just basic), just like any other discovery protocol.

I've had similar thoughts, though there seem to be a number of challenges.

How would you suggest:

- this be done?

- site / source rankings be done (the real value of current search portals)?

- avoiding (or penalising) false indexing --- either including index terms not in the indexed document(s), or omitting significant terms (excluding stopwords)?

- updating indices with new content submissions or modifications? Particularly for simple / technologically primitive sites.

- access points to search for users of the system?

No problem, I really appreciate it. Definitely going to work on that.

Glad to see that many people share the same feeling that pushed us to do this in the first place - until this point, we were skeptical about whether it existed or not.

We're planning to double down on the features that supports a feedback loop from consumers to creators.

We wanted to build this product in a way to provide more value to the content creators, while benefiting content consumers. We hope with this, creators will get direct exposure to their follower's taste and expectations.

Thanks for your time, much valuable information. Have a nice day!

Sorry for the late reply. We were not expecting this post to take off on HN. Got hit by unexpected traffic. The problem has been fixed with an alternative solution.

Thanks.. Right now accuracy is validated on a test dataset that we keep on our side. If you could use the feedback url on website to share some details, that dataset can be augmented further.

Thanks. We're kinda solving this problem for ourselves and people share the same emotions.