Hacker News Reader

Comment by transcriptase | Hacker News Reader

I doubt many will see this, but Yandex (yes, the Russian Google), is basically 2006 Google in the sense that it shows you what you want versus what Google’s lawyers and a bunch of SF dweebs working on their promo packet have in their best interest for you to see or not see.

Perhaps there’s domestic Russian things that are censored but that’s far outside my use-case.

View on HN

Yandex is only good if you search in English, because Yandex doesn't care about the English market :D. They only care about Russian and Turkic makers. (Maybe Vietnam? I forgot.)

Yandex is following the same trajectory. Yandex reverse image search used to find websites with an image, a la old Google. Now it’s the same “similar image” crap.

Exactly. At least Google gives you an "Find image source" button that restricts the result to exact matches. If the Yandex artificial stupidity decides you actually want some different image you don't get a choice. IME the Yandex image index also seems much smaller than Google's.

OTOH Yandex still gives your direct image links and is better at showing you the largest versions of an image (although too often that's just a blurry upscale).

Isn't most of Google from 2006 open source, or at least known in publicly available papers? So someone in the US should be able to re-create Google, and go back to basic page rank.

Is there anything preventing new search engines? Except scale and servers. But what most of us want is just plain old ordinary search as it existed in 2006, so that is probably reproducible.

I know there are some other engines. Like DuckDuckGo, but just found out they are really just Bing.

Which I guess begs the question, if Google Sucks so bad, why doesn't Bing take over??? It isn't as bad, even if not great.

Edit: After reading more posts. Appears Kagi is doing this.

PageRank relies on pages from one domain linking to pages on another domain as the main signal of quality. But no one does that anymore, because of PageRank. Most cross-domain links are from "partnerships" made under the explicit premise of boosting PageRank score. The metric became the goal, invalidating the metric.

Another factor is that since 2006, social media has displaced blogs. Blogs were a rich source of "authentic" cross-domain links, and the movement of online discussions to consolidated and closed platforms has dried up that well. Some even go so far as to pin the downturn of blogging on the demise of Google Reader.

I wonder if you could solve the PageRank problem by backpropagating negative scores for known bad sites. Then if someone sells out all they end up doing is ruining their own reputation.

The problem with openness has been turned on its head two years ago. Nowadays it’s the only way to not have your thoughts eaten and transformed in inexplicable ways by a myriad of LLMs.

Social media still has a lot of links. They are probably all have "nofollow" on them. Assuming google honors that.

Ironic, isn't it?

> so that is probably reproducible

Is it? How many more pages get published each day? If you include each YouTube video and social media post, it's easily several orders of magnitude. Organizing this kind of firehose isn't easy, neither is prioritizing/ranking it.

2006 was child's play in comparison.

The difference is 2006 Google just showed you what you were searching for. 2024 Google runs your query through 200 legal, moral, and “benefit to Google” layers of filtering, reinterpreting, and reordering before showing you results.

If you don’t believe me, search “watch frozen 2” (or anything else) on google, then on yandex. Both knew what you wanted, but only one is actually going to return the relevant results.

Strongly agree. It's like an alternate internet where Google stayed cool.

I use DuckDuckGo as my primary search engine, and Yandex as secondary

Reply on news.ycombinator.com