oldfart@yahoo.moc (Old Fart) wrote in
news:467675e9.3737984@news.west.earthlink.net:

>> Or, as is more likely, do you merely wish to prevent an
>>association between a particualr IP traceable to you and particular
>>net activities?
>
> Exactly, and the anonymity of a dynamic IP is really all I was after,
> but it sounds as if I may be forced into what for me would be the
> overkill of a proxy.

"Proxy" is, more or less the answer. But before focussing on the answer
let's explore the question a little.

I'm going to concentrate on surfing rather than email, newsgroups, irc,
etc. but some of the principles generalize.

Somewhat arbitrarily, let me propose several levels/motivations for
keeping things private: you fear inconvenience/annoyance, embarrassment,
litigation, prosecution, or murder ;-) Obviously, if you fear criminal
prosecution or being killed for your online actiovities then you must
adopt more secure methods than if you are just trying to avoid junk
email.

Potentially you can disclose your identity in a number of ways, which I
group for convenience as being determined by either the *content* of your
activities, or the *access pattern.* Content can be disclosed by
cookies, by java/javascript and such, and obviously by anything you write
in a message or even your writing style. Although it bridges to the
access pattern category, other disclosures, such as user-agent string or
even computer fingerprints (packet timestamps, etc.) could establish your
identity. Fortunately, aspects such as hardware fingerprints usually
only arise with the most serious adversaries. The content question also
depends on whether you make "ostensibly unrelated" visits to sites or
present a quasi-permanent persona to them (e.g., repeated posts under a
nym).

Determining who you are by access pattern could follow a number of broad
avenues: by info available at one destination site, by info available at
one destination site but with backtracking, by info available at/near
your origin, or by a global view (source through destination).

The last category is the province of major criminal/intelligence
investigations and will subsume the other categories; the best view of
your origin (and also all your destination activities if not
proxied/encrypted) is your ISP. The destination site looking backwards
is the view of end users or end destinations backtracking you.

As you can infer from the above a frequently-changing IP (from the same
ISP) provides only weak protection against only the weakest category of
adversary (a destination site or end user who is unwilling/incapable of
backtracking). If you need more than the weakest protection you must do
more.

As a minimum you should adopt strategies to limit your exposure to
endsites and - especially! - your ISP. That means some sort of proxy
with the link from your machine to the proxy encrypted. A single-hop
proxy (cotse, etc.) will give mild-to-moderate protection against
snooping by your ISP and backtracking without a subpoena. A single-hop
proxy gives reasonable privacy protection combined with little loss in
speed (and sometimes perks such as extensive email options to minimize
spam, etc.) Multi-hop encrypted proxies (a la Tor) are needed for more
serious protection but usually impose a performance penalty. Combine Tor
with helper apps (privoxy, vidalia, janusvm) for even stronger protection
minimizing some forms of "content" leakage. If you're a real nut, add
running your own Tor node into the mix (perhaps even an exit node,
although this can bring its own problems). This gives plausible
deniability that any of your surfing activities are attributable to you.

And harden your browser (at least something like Firefox with the
Noscript extension and cookies at least cut back to session-only).

Regards,