diff options
Diffstat (limited to 'ars-technica/published/notes-moz-richard-barnes.txt')
-rw-r--r-- | ars-technica/published/notes-moz-richard-barnes.txt | 90 |
1 files changed, 90 insertions, 0 deletions
diff --git a/ars-technica/published/notes-moz-richard-barnes.txt b/ars-technica/published/notes-moz-richard-barnes.txt new file mode 100644 index 0000000..889e8ac --- /dev/null +++ b/ars-technica/published/notes-moz-richard-barnes.txt @@ -0,0 +1,90 @@ + +> Which is still open so I would qualify that if I mention it. +> +> Okay, here are a few specific questions I have about this effort, though +> again, I'd much prefer to have something more like a discussion because +> my take on all this is somewhat open ended right now. +> +> 1) First and foremost what is the big win for users visiting a flat HTML +> site (that is, no login, no data exchange)? Which is to say, how does +> HTTPS help users outside of situations where they already have it (e.g. +> their bank, Facebook, et al)? + + +HTTPS give you auth and *integrity*. They know they're connecting to you and they know that the content is what you intended to provide. + +- subresource integrity spec: https://w3c.github.io/webappsec-subresource-integrity/#use-casesexamples let's the benefit propogate. If my 1st connection is https, and I download a link to some other site, SRI specifies a constraint of the image, SRI protects against dependencies (jquery, etc). + +Systemic benefit -- more things are HTTPS, great cannon can be prevented. widely used site + +little sites are more likely to get caught up in tracking or advertising. + + +> 2) The best answer I have come up with to the above question is that +> HTTPS stops unsophisticated MitM attacks. Do you have any numbers or +> research of any kind on how common such attack are? + +No one knows. Mozilla is trying to get such stats, but so far, says Barnes, "we don't have it. + +> 3) HTTPS consists of several layers, will Firefox be grading these +> layers on a per-site basis and letting the user know the overall level +> of security? That is, I might have implemented HTTPS, but done so in +> such a way that my server is vulnerable to Heartbleed, BOOST, POODLE, +> etc or supports a weak, possibly compromised cipher suite, will Firefox +> warn users about the potential vulnerability? If so how? If not, why +> not? + + +is there a date? + +It's already happening. New features are https, fido hardware auth, etc + +- Gradually phasing out access to browser features for non-secure websites + +For every features that goes away, the question becomes, "how much are you going to break the web for it's own good?" + +To be completely frank, I don't care about URLs I care about secure connections. So if you can get a secure connections via HSTS et al + +geo location api, get user media (mic camera), + + +HSTS and the upgrade-insecure-requests CSP + +Still treated as mixed content, HSTS you discover as you browse, + +"HSTS priming spec" + + +> 4) LetEncrypt is great, but it's still way beyond the capabilities of +> non-technical users. Yet part of what makes the web amazing is how +> simple it is to just create a few text files, put them in the folder, +> upload it to a server and you have a site (this is I believe one of the +> central parts of Mozilla's Maker efforts, that anyone can create things +> on the web). + +Fix the transport level. + +Big site concerns: + +- not too complex +- dependecies -- media sites can't go HTTPS without their ads being HTTPS as the ecosystem moves in that direction the big sites don't have to worry as much. + + +Little site concerns + +- complexity (config, etc) +- same level of automation as DNS - caddy server +- dependencies + + +> 5) Tim Berners Lee has called the move from http to https, "arguably a +> greater threat to the integrity for the web than anything else in its +> history." Given that URLs breaking, changing and disappearing is already +> a massive problem, and that this move will absolutely mean more broken +> sites, how is that a win for the web? Is a secure web that's only 10% of +> the web better than an insecure web? +> + +Tim has been a really useful contraian voice. His views have driven the browser and web community to address concerns he has raised. HTST priming is designed to address. + + |