Haystack: A small utility helping to protect your searches | Hacker News Reader

Haystack: A small utility helping to protect your searches

github.com

20 pointsnorova17 comments

I prefer GoogleSharing.

http://www.googlesharing.net/

"With Google's introduction of SSL support for search requests (encrypted.google.com):

The result is that Google knows what is being searched for, but doesn't know where the requests are coming from. The GoogleSharing proxy can tell where requests are coming from, but can't tell what the content of the requests is. And the user can avail themselves of Google services without having to trust either Google or GoogleSharing."

Not to knock your efforts, but this looks like a naive and wasteful approach to privacy. It seems like it would be pretty easy to separate the needle from the haystack when dealing fake searches (are any links clicked from fake searches? are cookies used? javascript? useragents? what about google's realtime search? are you searching for nonsense terms?).

Also I'm not confident this thing isn't going to randomly pick embarrassing search terms.

This is why I shared it. In the unlikely event that someone find it the least bit useful, they might want to expand on what already exists (which admittedly isn't much) and improve it. All of your points make complete sense, and as I've said in the github notes, the project was just something I cooked up quickly a few years ago. Small ideas can grow into big things; figured it was worth the bandwidth to publish.

(You should be aware that Haystack's also the name of a failed and cringeworthy attempt at creating privacy software.)

I think it's pretty irresponsible to publish privacy tools that you know don't actually give adequate privacy. It's easy for me to see the holes in this technique, but not everyone is technical. If someone's reading this and considering deploying this tool, please just use tor + torbutton instead..

As well as a Django search tool: http://haystacksearch.org/

God forbid I put myself out there in an attempt to share something with the world (which is admittedly naive) and maybe let someone else improve upon an idea. I'm so terribly irresponsible.

You did nothing wrong and should carry on learning, coding and sharing. Thanks for sharing your project with HN.

Anyone with the chops to pull and compile C# code from Github is technical enough to understand the implications of trusting their privacy to this tool. Please don't be so dramatic.

Haystack? You mean the MIT project? http://en.wikipedia.org/wiki/Haystack_(MIT_project), the search tool http://haystacksearch.org/, the horribly-implemented privacy tool http://en.wikipedia.org/wiki/Haystack_(software) or...

... the Facebook object store? http://www.facebook.com/note.php?note_id=76191543919 or ...

Why not just use scroogle (http://scroogle.org/)?

Even if it randomly searches for stuff, the searches are still linked to you, and I would be concerned over what terms it was searching for if my search history were ever subpoenaed. I could images several cases where this could get you into trouble.

Why not just use DuckDuckGo.com as your search engine? They don't save any of your personal information, and they return less content-farm spam than Google does. (My second choice would be scroogle.org, which somebody else already mentioned -- an anonymous Google proxy, and doesn't require a browser add-on like googlesharing.net does.)

Just a little note. I just pushed this to Github recently after discovering the code on my machine. I wrote this little utility a few years ago and didn't do much with it. Thought I would share the code and see if anyone else finds it useful! Thanks for taking a look.

Would be nice to have a Linux utility like this that was browser-agnosic and os-agnostic (ie. configurable to mask as any browser/os).

Another useful feature would be to have the utility follow some of the "next page" links on some of the searches, to make them look even more like regular human searches.

The "next page"-type behavior and more human-like searches was something I originally intended on adding but never got 'round to it. After re-discovering the code on my machine though, I'm finding myself a bit re-motivated.

The current version isn't browser-specific. You could add user-agent headers pretty easily. The existing C# code might compile with limited effort on Mono.

Excellent, God forbid Google ever use your search queries to improve its ranking algorithms and give more relevant results.

This comment sponsored by Experts-Exchange and ehow

Reply on news.ycombinator.com