[HN Gopher] Show HN: IPDetective - An API for IP bot detection
       ___________________________________________________________________
        
       Show HN: IPDetective - An API for IP bot detection
        
       IPDetective collects data from about 60+ different sources such as
       official cloud provider endpoints and public VPN/Proxy/Tor/Bot net
       lists. Then aggregates this data into a fast and easy to use API
       that can be integrated into applications or scripts easily.
       IPDetective started as a hobby project for my other hobby projects
       :) and I decided to wrap a simple website around and offer it as a
       service.  Let me know what your thoughts, if you find value in this
       service or if you have any feature requests.
        
       Author : AndrewCopeland
       Score  : 29 points
       Date   : 2022-09-12 19:57 UTC (1 days ago)
        
 (HTM) web link (ipdetective.io)
 (TXT) w3m dump (ipdetective.io)
        
       | bootsmann wrote:
       | How do you plan to keep the list up to date and especially
       | reliably accurate? (providers change, bot nets get cleaned, ASN
       | especially in 3rd world countries are very unreliable, crawlers
       | might scrap weird things etc.)
       | 
       | Disclaimer: I work for a company that operates in a similar space
       | so don't go into too much secret sauce if you don't want to :)
        
         | AndrewCopeland wrote:
         | So I have a scraper that gathers the information from about 60
         | different sources using a wide array of techniques which is run
         | everyday.
         | 
         | The largest thing I try to look at is to make sure my sources
         | are not older than 1 year. Sometimes these sources break and I
         | update them or they go offline all together. I am always
         | looking for new hosting providers, vpn and proxies.
         | 
         | Also I can do further port analytics on these ips as well. A
         | lot of the VPN services have specific ports open used by VPN
         | services. I would say that automating every source from the
         | start was a good way to start so I can stay up to date.
        
       | stevenicr wrote:
       | I checked the posted privacy policy - and I have no idea what
       | your system logs and for how long.
       | 
       | There are a few sites I could use this for, but some of them
       | would also end up sending private customer ips for lookups, and
       | for that reason I just manually check the suspicious ones and
       | manually don't look up the ones I know have passed the 'already a
       | member and human not hacker' gate.
       | 
       | I keep look for something I can just add lists to my server and
       | check against that, but not going to spend thousands on it.
        
         | marvinblum wrote:
         | Same here. I would like to integrate a simple blacklist into
         | our system. Does anybody know of a good IP lists that contains
         | known bots, data centers, and so on?
         | 
         | Also from a performance perspective, I can't do hundreds of
         | millions of requests a month over the wire. That just a waste.
        
       | bragr wrote:
       | >Let me know what your thoughts
       | 
       | Do you have any data showing your service is better/more
       | accurate/better false positive rate than competing offerings?
       | 
       | >IPDetective collects data from about 60+ different sources such
       | as official cloud provider endpoints and public VPN/Proxy/Tor/Bot
       | net lists.
       | 
       | Are you on the up and up with all those public sources? I'm not
       | sure which ones you're sourcing, but many do not allow commercial
       | use, resale, or at the very least have some attribution clause to
       | keep security companies from mooching off crowdsourced data.
       | 
       | >No, currently IPDetective does not support ipv6 addresses.
       | However this feature is on the road map.
       | 
       | That's a major shortcoming in 2022.
       | 
       | Finally, as itake pointed out bellow, I'm not giving you my email
       | just to run a simple test query and see what the results look
       | like.
        
         | AndrewCopeland wrote:
         | Thanks for your input.
         | 
         | >Do you have any data showing your service is better/more
         | accurate/better false positive rate than competing offerings
         | 
         | I have signed up to some of the competitors and my service
         | averaged about 20ms per request for a single IP address. From
         | the competition it was typically around 200ms to 300ms.
         | Regarding better false positives that would be rather hard to
         | do.
         | 
         | > Are you on the up and up with all those public sources?
         | 
         | I gathered from sources that did not have any licensing related
         | to non-commercial use. Would you recommend I reach out to all
         | of the sources regardless even if they say you can use it for
         | commercial use? Or do not say anything about commercial use at
         | all?
         | 
         | > No, currently IPDetective does not support ipv6 addresses.
         | However this feature is on the road map.
         | 
         | Yeah I know, just focusing on IPv4 right now. I am still
         | collecting ipv6 addresses however I have not implemented it in
         | the service yet.
         | 
         | Thanks a bunch for writing the comments above.
        
           | kosolam wrote:
           | Yeap... measure the amount of requests for ipv6 support you
           | get from your users, not hn users :-)
        
             | bragr wrote:
             | https://www.google.com/intl/en/ipv6/statistics.html
        
       | itake wrote:
       | 1/ needs authenticationless way to search. all other ip look up
       | tools don't require an account for me to test this. I'm not going
       | to register.
       | 
       | 2/ I've tried blocking vpn or proxy users (by ip hosting
       | provider), but found too many false positives. Using a VPN is
       | common for IT professionals or consumers trying to 'protect'
       | their privacy and this impacted the growth of my app.
       | 
       | How is your tool better at detecting a bot vs a human using a
       | vpn?
        
         | AndrewCopeland wrote:
         | >1/ needs authenticationless way to search. all other ip look
         | up tools don't require an account for me to test this. I'm not
         | going to register.
         | 
         | Okay I will look into a solution.
         | 
         | 2/ I've tried blocking vpn or proxy users (by ip hosting
         | provider), but found too many false positives. Using a VPN is
         | common for IT professionals or consumers trying to 'protect'
         | their privacy and this impacted the growth of my app.
         | 
         | That good to know, I do not find that I have a lot of false
         | postivies but I would imagine it all depends on audience.
         | 
         | >How is your tool better at detecting a bot vs a human using a
         | vpn?
         | 
         | It does not know the difference as or right now. I was thinking
         | about adding a user-agent validation as well which could add
         | another layer.
        
           | michaelmior wrote:
           | > I was thinking about adding a user-agent validation as well
           | which could add another layer.
           | 
           | Presumably this service exists because bots try to avoid
           | detection. I don't think UA validation would really help much
           | and there are plenty of libraries already that do this.
        
       | joshmn wrote:
       | Does this handle residential or LTE proxies? Obviously harder to
       | identify, but it's not unfeasible.
        
         | AndrewCopeland wrote:
         | Currently we collect the data across several free public proxy
         | lists. Definitely more difficult to do wit the proxies you
         | mentioned. I think traffic analytics and port scanning would
         | need to be implemented to get these residential and LTE
         | proxies.
        
       | mousetree wrote:
       | Good work on the pricing and free tier. Much cheaper than your
       | competitors (we're busy testing out ipinfo.io). But would
       | recommend you add a bit more data in your API response if you
       | already have it or can built it out easily. For example, any
       | geocoding, ASN name, whether its a proxy or a VPN etc.
        
       | AndrewCopeland wrote:
       | Please feel free to leave comments if you have any questions or
       | any feedback.
        
       | mxuribe wrote:
       | I didn't see my question in the FAQ, so asking here: do you have
       | a process to contest accidental inclusion of an IP address?
        
         | AndrewCopeland wrote:
         | I do not, but it seems like a great feature request. I will
         | look into adding something like this.
        
           | mxuribe wrote:
           | Thanks so much @AndrewCopeland!
           | 
           | I should have added context (or at least an anecdote) to help
           | you during any product roadmap meetings...I used to oversee
           | web ops for global real estate company (one of the biggest in
           | the U.S. and the woirld)...and our consumer-facing websites
           | would show tons and tons and tons of listings of residential
           | homes for sale. Of course we really were just showing
           | listings from our data store as well as other real estate
           | companies who agreed to share listings data. As in many data
           | sharing and data synch arenas, there are data issues. The
           | most common scenario: "hey, you're showing a home that sold X
           | time ago...stop showing it!" And, in all cases it was
           | "someone else's data". But to the customer, or even other
           | realtors, we were the bad guys. Even realtors who should have
           | known betteer that there are always data issues that we can
           | not fully control; at least nopt at the source...would
           | complain to us. Now, one might assume that once the source
           | data updates properly, the "correct data" should flow
           | through, right? Well, not in real estate! Clearly this is a
           | different arena. But i've learned that even if its only to
           | remove a local cahce of data, its a good idea to give users
           | and/or customers a mechanism to at least properly communicate
           | to you about stale data...and if appropriate, maybe evenallow
           | self-service for a user/customer to get rid of the "bad
           | data". Obviously this merits putting in place protocols to
           | avoid abuse...but i hope you get the idea. Good luck!
        
       ___________________________________________________________________
       (page generated 2022-09-13 23:01 UTC)