Hacker News new | past | comments | ask | show | jobs | submit login

The only problem is that 90% of the time the image is removed from the site ! The image is often still hosted, but has been removed from the page content. This is very annoying! So the next step for Google is to actually make sure the site still has the image before including it in the search results.

btw Text search has the same problem, but is less annoying because we are so used to web sites not having the info and will just continue searching. It's extra annoying with images because we then actually knows it has the image we are looking for, but then it has not. It's like when the information I look for comes up in the site description, but when I visit the site it's not there.

I guess this is a hard problem to solve, but would improve the search experience a lot. One idea is to have the browser fetch the site content and do a full text search on it before it's displayed in the search result. Could be done with a browser extension.




>The image is often still hosted, but has been removed from the page content.

The same applies to web searches,

>So the next step for Google is to actually make sure the site still has the image before including it in the search results.

And they should do the same for videos and DDoS the whole Internet just for 0.004% of chances making minor manual action a bit inconvenient.


You are arguing to save a minor amount of bandwidth VS saving thousands of man-hours.


>save a minor amount of bandwidth

I dont think that's minor:

https://www.cnet.com/news/google-goes-down-for-5-minutes-int...

now imagine google has to check if every link it will show still exists. That's like billions of trillions of request per hour world-wide. And another thing is a privacy issue, I know, google doesnt give a shit about it, but some people do. Think that you search for 'smelly poo' and 10 different servers get HEAD request to articles that are top ranked in this keyword, potentially 9 of the servers won't get a human-made visit, 9 sysadmins or developers will find a way to abuse that metadata in the future.


I doubt Google account for 40% of Internet traffic. How would you even measure that !? What most likely happened was they saw a 40% drop in traffic on their servers while Google was down.



Same source. Basically the same article too. Probably a marketing piece by GoSquared.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: