Hacker News new | ask | show | jobs
by putonyourshoes 3141 days ago
Hey, I work on Google search. For any of the search queries you mentioned, would you be able to share examples of webpages you would have rather seen in the results?
9 comments

If you want a specific example, here's one that came up in the news recently. Try searching for "rasberry ketones", a nutritional supplement. It's filled with hyper tuned SEO pages with very little content. Many suggest they are informational, but are really article/adverts tuned for search engines. Even those that appear on ostensibly legit websites have ridiculously click-baity titles like "What They Don’t Want You to Know About Raspberry Ketones", which in my mind raises red flags that BS is ahead.

If you read a newspaper, you clearly know when someone is trying to sell you something, but on Google search, you don't. I don't mind being marketed to, but I would like to know when I'm marketed to, and the vast majority of google results are organic search marketing.

Interestingly, I just searched for it and the results are:

DDG: 1. WebMD 2. Walmart 3. examine.com 4. Wikipedia 5. Amazon 6. some site apparently dedicated to r.k. 7. livescience.com 8. some webshop 9. HuffPo with the clickbait title you mentioned 10. Some kind of webmd clone

Google: 1. WebMD 2. healthline.com 3. examine.com 4. Dr. Axe (no idea what this is but doesn't seem to be purely spam) 5. livescience.com 6. Wikipedia 7. HuffPo clickbait 8. Some kind of webshop 9-10. Amazon

I wouldn't say it all spam, though there are questionable sites in top 10, but there are also a number of legit ones.

I note though "rasberry ketones" is not a correct spelling, it should be "raspberry". Maybe that's the reason you've got bad results?

Interesting, thanks. Did you ever manage to find better-quality material on raspberry ketones that you think should have been in the search results instead?
Yes, I did, although it took quite some time.
Would you mind sharing a link or two?
Search just about any health related query, the vast majority are pure snake-oil salesmen. Product reviews basically either point to CNet or some relatively trusted source, or if they don't, they point to complete garbage. Try looking for any public records, most bring you to "people search" websites that trick users into thinking they do more than they do. Searching for anything historical brings up wikipedia, and if your lucky, some google books results, which is good, except that Google stopped scanning books. I could go on. It seems like Google is fine at picking the major trusted sources that everyone knows about, but is really bad at differentiating smaller quality sources from spammers. I mostly use Google as my Wikipedia/StackOverflow search machine. I usually find better material searching the references of a wikipedia article.
> except that Google stopped scanning books

That's really a shame (and an opportunity for someone else).

Search for example on "deep learning". Results:

+ nice: open books (deeplearningbook.org)

+ not nice: online seminars (udemy, udacity, edx)

+ not nice: wikipedia copycats (investopedia)

+ not nice: SEO optimized business analytics sites recognizable by their top 10 this, top 20 that (kdnuggets)

+ not nice: commercial software (mathworks)

+ not nice: big companies hitchhiking on buzzwords (sas)

What would be nice are websites and people that are often linked from HN. For example http://www.inference.vc rather than sas. Or https://distill.pub/ rather than investopedia. Or https://www.tensorflow.org rather than mathworks. If good sites are showing up in the top results more often, there will also be more incentive to keep up the good work as well.

Better delineation between commercial results vs. noncommercial results; Where by "commercial", I mean a web page whose primary purpose is to sell me something. (Yes, HN is there to sell me on ycombinator, but as long as there's no credit card involved within 5 pages of following the description, it's probably on the "informative" side)

It's an easy distinction to make for a human, I'm sure Google's omniscient AI can easily be trained to make that distinction. And it would also work in Google's favor, at least with respect to users who do not have ad blockers - if (as a general rule or thanks to explicit user choice) "commercial pages" are demoted, the google ads become more valuable.

what results were before Google got rid of the discussion filter -- namely blogs, forums, and communities

these are all hidden from most search results and consumers are now forced to settle for SEO optimized trash results, rather than discussions

If I am searching from a country google endpoint please show more results from that countries' domain.
Signal to noise on the defunct Google Discussions search was much, much higher.
why doesn't google search implement downvoting / upvoting search results? I never understood this...

what if human judgement was better than the AI powering the results indexing, for particular keywords that don't seem to be machine learned well enough

From waht I've read, they do implement voting of a kind, by measuring clicks (and considering that a 'vote') and time on the result page (if you immediately hit 'back' and click on another link, you obviously weren't satisfied).
If you think human judgement is going to deliver truly good content, then you either haven't been introduced to the concepts of "clickbait" and "the reddit frontpage".
Hacker News works just fine ?
For incredibly specific topics. The number of topics which don't make it to the front page, or which are flagged off the front page, is pretty large.
> Hey, I work on Google search. For any of the search queries you mentioned, would you be able to share examples of webpages you would have rather seen in the results?

No, because you don't show them to us.