4

The question What information is sent to Wolfram Research when loading or working with Mathematica? describes the information sent to Wolfram's servers in routine Mathematica usage (Documentation Center/Wolfram|Alpha/Paclet updates etc).

Apparently you can assume that the following is sent to Wolfram's Servers

  • All your search terms used in the Documentation Center
  • IP Number
  • $Version
  • $SystemID
  • $LicenseID
  • $MachineID

But what about when sending a generic web query using URLExecute, Import etc. to any server? What information is sent and can you modify it to attain "Tor-like" anonymity?

Some Background:

Settings for a web-query can be wrapped in a HTTPRequest which includes a "UserAgent" setting that is by default the "Wolfram HTTPClient xxxx" (where xxxx is a version number). This "UserAgent" seems to carry information about the operating system and of course indicates that it emanates from Mathematica. There isn't a way of spoofing the IP number in the HTTPRequest wrapper as pointed out by Carl Lange in the comments but this can potentially be done by

  1. Setting the "UserAgent" key in HTTPRequest to apply that "UserAgent"'s proxy settings.
  2. Adding proxy settings in the Proxy Dialog found in Mathematica's Preferences.

enter image description here

As per the documentation by default on Windows the proxy settings are taken from browsers while on a Mac they are taken from the Network Preferences panel.

Another alternative might be to just use browser proxies by restricting to WebExecute

Saunter
  • 49
  • 2
  • It's not clear to me what you are trying to do (and how changing the user agent would be related to Tor). Can you edit the question and explain the problem fully (but concisely)? – Szabolcs Dec 02 '19 at 15:10
  • This is a Tor problem and not a Mathematica problem. Look at this page if you want general configuration advice, or ask at tor.SE if you are unable to get that to work. – J. M.'s missing motivation Dec 02 '19 at 21:15
  • 1
    @J.M.willbebacksoon The question is independent of Tor and seems very much a Mathematica issue related to its privacy provisions. The use of Tor as the "UserAgent" was simply an attempt to see if Mathematica can piggy-back off its proxy handling to ensure online anonymity. I will edit the question to clarify. – Saunter Dec 02 '19 at 23:11
  • In general, I don't think it is wise to use commercial software with something like Tor. If necessary, you can license the Wolfram Engine with a fake address and run it on a virtual machine. It should probably not be against TOS.(but What kind of situation is that?) – Xminer Dec 03 '19 at 00:26
  • 3
    Eventually, Mathematica is just doing the equivalent of a curl command to the remote server and everything that is sent except your IP address should be configurable in the request (by changing either the HTTP headers or the body). Your IP address is un-spoofable at the Mathematica level (though you can attempt it using Tor or similar, as you point out, or a simple tunnel). You can try this yourself by spinning up a local web server and firing a few URLExecutes at it. – Carl Lange Dec 03 '19 at 00:46
  • (As you've probably already seen, Mathematica has native support for proxy servers in its network preferences, and this is another way to modify the IP address that the resulting server will see.) – Carl Lange Dec 03 '19 at 00:50
  • @CarlLange Ok, thanks for clarifying - yes, not being very au fait in these networking issues I was hoping that someone might have some experience in actual proxy settings that adequately spoof but also, your suggestion about testing with a local web-server is "obvious" - thanks. – Saunter Dec 03 '19 at 01:47
  • @Szabolcs sorry, not sure the level of detail required and poorly worded but perhaps not unlike this question on automating web-browsing but updated with new Import functionality and an emphasis on understanding identifying information sent. – Saunter Dec 03 '19 at 02:21
  • Mods - why is this now off topic in the light of edits having pointed out that 1) this is clearly a Mathematica question 2) it can be answered without further information 3) it is similar in flavor to several other questions 4) there seems a modicum of community interest in recording a resolution? Seems very odd how mods here are so keen to close en mass but then just as quick to disappear without engaging/justifying? – Saunter Dec 06 '19 at 01:14
  • @Saunter The Stack Exchange UX has changed (in the last few hours), but the decision to close as off-topic was made three days ago (presumably before your edits). Further - this is a community-run website. Most people who vote on whether questions should be closed (or re-opened) are regular users of the site, not moderators, and have no major extra decision-making power. Your question has 4 reopen votes, and it only needs one more! – Carl Lange Dec 06 '19 at 09:26
  • @CarlLange Fair enough. Still, might have hoped the initial closers would have "unclosed" given response(s) but perhaps they (mostly) have. – Saunter Dec 07 '19 at 11:04

0 Answers0