4 ms·

Faroo: Peer-To-Peer Web Search

56 points by flippyhead 11y ago

wslh 11y ago
Sorry but where can I try FAROO without downloading an app or connecting to an API?
- charlieegan3 11y ago
  http://www.faroo.com/ http://www.faroo.com/
  - daveloyall 11y ago
    Query: sed split file by pattern Result: No results were found.
    - 8_hours_ago 11y ago
      python string also has 0 results. It doesn't seem like this is quite ready yet. The website also added a new page to my browser history for every character that I typed in the search box, which is kind of a pain (also, it's slightly amusing that it added so much to my browser history when the search engine presumably gives weight to websites that occur frequently in its users' browser history).
ZenoArrow 11y ago
So would it be fair to say that effectively Faroo works by: 1. Scanning your browser cache for the sites you visit. 2. Setting a ranking to the sites in your browser cache, based on how frequently you visit the site. 3. Merging this site + ranking information with other Faroo users in a search index. 4. Distributing the search index in a distributed way, perhaps all nodes only having a fraction of the total index to prevent storage issues. If that's a simplified version of what is happening then I could see it working. If I've misunderstood something let me know.
- dogma1138 11y ago
  That's pretty accurate and quite scary, mostly because it will start pulling off bio's from PornHub now instead of Wikipedia and that can never end well.
btown 11y ago
Sadly, by its very nature, it misses a lot of the "long tail" of rarely-visited sites that Google et. al. can crawl.
magila 11y ago
It appears they are relying on a combination of not documenting their ranking algorithm and updating it periodically to prevent spammers from flooding the system with bogus "attention" signals. I'll leave determining the likelihood of this approach being successful as an exercise for the reader.
- ljk 11y ago
  So... security through obscurity?
lrvick 11y ago
At first I thought, awesome, finally an open distributed search engine. ...Oh wait, you want me to install your closed source code on my computers, to give you all my personal data and CPU cycles for you to own and use as you like. (Which you "promise" won't be exploited or abused). Make no mistake, this is a voluntary botnet for a single party to use to build a company on. This voluntary botnet will provide search engine services from a central point so they can keep the majority of profits without the overhead of paying for resources like every other for-profit company. This is not Peer-2-Peer it is Peer-2-Company. Also they are late to the party on this approach, Microsoft already tried this with Bing: https://googleblog.blogspot.com/2011/02/microsofts-bing-uses-google-search.html https://googleblog.blogspot.com/2011/02/microsofts-bing-uses... This model is just begging for people to troll them by releasing tools to poison their database with nonsense data. They should build a -real- open p2p distributed system (the blockchain is not perfect but a good example). Short of that if they want to be a for-profit centralized search engine, fine. They should go buy some servers and databases they can directly control. and try to be a closed source competitor to duckduckgo.
- drakecoffeecake 11y ago
  Yep, yep, yep. You seem to be exactly right.
- fungos 11y ago
  There is one open source: http://yacy.net/ http://yacy.net/ The only downside is that it requires Java.
- joshu 11y ago
  > They should build a -real- open p2p distributed system what's stopping you?
  - CJKinni 11y ago
    I'm not the individual you're responding to, but probably a mix of time, money, and interest. You don't need to make a better version of something for your criticism to be valid.
mdip 11y ago
I found this a little problematic: FAROO indexes only pages which are located in the Internet, but no Intranet pages or HTTPS protected pages If it simply can't see HTTPS pages, it'll leave a large chunk of the internet invisible to the search engine. I understand the reason for this, but it's a technical limitation they'll have to find a way past to make it useful as more and more sites encrypt by default.
- dogma1138 11y ago
  It indexes pages you visit, it does some MITM/Browser snooping but at least it's not intrusive enough to do SSL stripping.
bhouston 11y ago
We tried using various for an all we developed and its results were fairly poor, so we had to switch to bing.
curiousjorge 11y ago
Gene Kan (https://en.wikipedia.org/wiki/Gene_Kan https://en.wikipedia.org/wiki/Gene_Kan) also had the same idea, until he shot himself while working on a distrubted peer to peer real time search engine in front of his computer I think. I heard about him when I watched a documentary about Napster. No documentary about Gene Kan exists. He is virtually unknown but his work was an important contribution.
jzelinskie 11y ago
Something similar was posted to HN a while ago[0]. However, it was made specifically for indexing scientific papers without keeping all the content. Totally open source and runs in the browser. I feel like a more generic implementation could be done well, but hasn't been so far. [0]: http://juretriglav.si/an-open-distributed-search-engine-for-science/ http://juretriglav.si/an-open-distributed-search-engine-for-...

56 points by flippyhead 11y ago

wslh 11y ago

Sorry but where can I try FAROO without downloading an app or connecting to an API?

charlieegan3 11y ago
http://www.faroo.com/ http://www.faroo.com/
- daveloyall 11y ago
  Query: sed split file by pattern Result: No results were found.
  - 8_hours_ago 11y ago
    python string also has 0 results. It doesn't seem like this is quite ready yet. The website also added a new page to my browser history for every character that I typed in the search box, which is kind of a pain (also, it's slightly amusing that it added so much to my browser history when the search engine presumably gives weight to websites that occur frequently in its users' browser history).

ZenoArrow 11y ago

So would it be fair to say that effectively Faroo works by: 1. Scanning your browser cache for the sites you visit. 2. Setting a ranking to the sites in your browser cache, based on how frequently you visit the site. 3. Merging this site + ranking information with other Faroo users in a search index. 4. Distributing the search index in a distributed way, perhaps all nodes only having a fraction of the total index to prevent storage issues. If that's a simplified version of what is happening then I could see it working. If I've misunderstood something let me know.

dogma1138 11y ago
That's pretty accurate and quite scary, mostly because it will start pulling off bio's from PornHub now instead of Wikipedia and that can never end well.

btown 11y ago

Sadly, by its very nature, it misses a lot of the "long tail" of rarely-visited sites that Google et. al. can crawl.

magila 11y ago

It appears they are relying on a combination of not documenting their ranking algorithm and updating it periodically to prevent spammers from flooding the system with bogus "attention" signals. I'll leave determining the likelihood of this approach being successful as an exercise for the reader.

ljk 11y ago
So... security through obscurity?

lrvick 11y ago

At first I thought, awesome, finally an open distributed search engine. ...Oh wait, you want me to install your closed source code on my computers, to give you all my personal data and CPU cycles for you to own and use as you like. (Which you "promise" won't be exploited or abused). Make no mistake, this is a voluntary botnet for a single party to use to build a company on. This voluntary botnet will provide search engine services from a central point so they can keep the majority of profits without the overhead of paying for resources like every other for-profit company. This is not Peer-2-Peer it is Peer-2-Company. Also they are late to the party on this approach, Microsoft already tried this with Bing: https://googleblog.blogspot.com/2011/02/microsofts-bing-uses-google-search.html https://googleblog.blogspot.com/2011/02/microsofts-bing-uses... This model is just begging for people to troll them by releasing tools to poison their database with nonsense data. They should build a -real- open p2p distributed system (the blockchain is not perfect but a good example). Short of that if they want to be a for-profit centralized search engine, fine. They should go buy some servers and databases they can directly control. and try to be a closed source competitor to duckduckgo.

drakecoffeecake 11y ago
Yep, yep, yep. You seem to be exactly right.
fungos 11y ago
There is one open source: http://yacy.net/ http://yacy.net/ The only downside is that it requires Java.
joshu 11y ago
> They should build a -real- open p2p distributed system what's stopping you?
- CJKinni 11y ago
  I'm not the individual you're responding to, but probably a mix of time, money, and interest. You don't need to make a better version of something for your criticism to be valid.

mdip 11y ago

I found this a little problematic: FAROO indexes only pages which are located in the Internet, but no Intranet pages or HTTPS protected pages If it simply can't see HTTPS pages, it'll leave a large chunk of the internet invisible to the search engine. I understand the reason for this, but it's a technical limitation they'll have to find a way past to make it useful as more and more sites encrypt by default.

dogma1138 11y ago
It indexes pages you visit, it does some MITM/Browser snooping but at least it's not intrusive enough to do SSL stripping.

bhouston 11y ago

We tried using various for an all we developed and its results were fairly poor, so we had to switch to bing.

curiousjorge 11y ago

Gene Kan (https://en.wikipedia.org/wiki/Gene_Kan https://en.wikipedia.org/wiki/Gene_Kan) also had the same idea, until he shot himself while working on a distrubted peer to peer real time search engine in front of his computer I think. I heard about him when I watched a documentary about Napster. No documentary about Gene Kan exists. He is virtually unknown but his work was an important contribution.

jzelinskie 11y ago

Something similar was posted to HN a while ago[0]. However, it was made specifically for indexing scientific papers without keeping all the content. Totally open source and runs in the browser. I feel like a more generic implementation could be done well, but hasn't been so far. [0]: http://juretriglav.si/an-open-distributed-search-engine-for-science/ http://juretriglav.si/an-open-distributed-search-engine-for-...