Slashdot Mirror
Is Google's Promotion of HTTPS Misguided? (this.how)
Long-time software guru Dave Winer is criticizing Google's plans to deprecate HTTP (by, for example, penalizing sites that use HTTP instead of HTTPS in search results and flagging them as "insecure" in Chrome). Winer writes:
A lot of the web consists of archives. Files put in places that no one maintains. They just work. There's no one there to do the work that Google wants all sites to do. And some people have large numbers of domains and sub-domains hosted on all kinds of software Google never thought about. Places where the work required to convert wouldn't be justified by the possible benefit. The reason there's so much diversity is that the web is an open thing, it was never owned....
If Google succeeds, it will make a lot of the web's history inaccessible. People put stuff on the web precisely so it would be preserved over time. That's why it's important that no one has the power to change what the web is. It's like a massive book burning, at a much bigger scale than ever done before.
"Many of these sites don't collect user data or provide user interaction," adds Slashdot reader saccade.com, "so the 'risks' of not using HTTPS are irrelevant." And Winer summarizes his position in three points.
If Google succeeds, it will make a lot of the web's history inaccessible. People put stuff on the web precisely so it would be preserved over time. That's why it's important that no one has the power to change what the web is. It's like a massive book burning, at a much bigger scale than ever done before.
"Many of these sites don't collect user data or provide user interaction," adds Slashdot reader saccade.com, "so the 'risks' of not using HTTPS are irrelevant." And Winer summarizes his position in three points.
- The web is an open platform, not a corporate platform.
- It is defined by its stability. 25-plus years and it's still going strong.
- Google is a guest on the web, as we all are. Guests don't make the rules.
"The web is a social agreement not to break things," Winer writes. "It's served us for 25 years. I don't want to give it up because a bunch of nerds at Google think they know best."
Google is never going to make Chrome unable to access HTTP sites. If for no other reason than because the moment they did, they know everybody would switch to a different browser. They're not in the business of making information inaccessible. Their strategy of giving preference to HTTPS sites is perfectly reasonable though, all the more reasonable because of the fact that HTTP sites are generally old and unmaintained. I want old data to show up in my search results, but I rarely want it to show up first.
This space intentionally left blank
Downloading executable files, downloading risky file extensions (doc, pdf), and downloading any document where integrity matters means that http is a risk. If someone downloads some old games from an HTTP archive, malware could be added. If someone downloads some PDFs with an outdated reader, there could be malware. If someone downloads some forms they're going to fill out later, changing the location they're supposed to be emailed/faxed/whatever means someone could give out PII or financial information. If someone is reading old news stories, changing the content of those stories to suit an attackers narrative could be very valuable. Just because the author can't imagine the security implications, doesn't mean organized crime, bored hackers, or nation state actors aren't thinking about it.
You can walk into libraries all over the world, pull a book off the shelf, and read it. Nobody maintains it; it just sits there. Some things work that way.
Except that the rules for HTTPS have changed at least 3 or 4 times, and recently. First keys weren't long enough. Then SSL wasn't good enough. Then TLS 1.0 is broken.
Managing ssl.conf across a few dozen servers has taken a fair amount of man hours at my organization in the last couple years-- and we have configuration management tools.
And all of this is to protect the transmission of unrestricted, publicly accessible information.
Do we really need https to display wikipedia? To see today's headlines on CNN? To read slashdot? Does the wayback machine of publicly viewable web pages need to be encrypted during transmission?
A large percentage of the web doesn't need to be encrypted during transmission.
In order to save the village, we had to destroy it.
Your criticism of insecurity has little to do with security in an httpd. It can be easily expanded to demanding that all machines connected to the net 'have their papers in order.' China loves advocates like you.
To answer your questions: yes. It needs to be default. Users, civilians, need to know when a web page is sending info across a network that's unencrypted, e.g. as plain text. They don't know the implications.
It would be a wonderful world if key management was simple, and it can be. CASB apps make it simple.
Wait until you find wire-sniffing apps inside your (expletives deleted) routers, or someone that's programmed a router port mirror to a tor listener. Security isn't that tough, but it eludes thousands of organizations. Look at this weeks, largest-ever breach in Florida, where most all of the living population of the United States had their names, addresses, and a few other juicy fields snarfed because of stupidity. The basics should include TLS 1.3.
Yes it changes. Anything valuable still requires paying attention to it. Civilians are clueless, and it's up to the responsible ones to do the job. So we do it. LetsCrypt is an easy method to get a cert and use it. I'm still unsatisfied that WPA3 is worth it, but I like how it works at a glance. In the real world, much stuff is broken and vendors are stupid and in it for this quarter's model, and this quarter's report to Wall Street and little else. Raising the standard from plain text to encrypted is an important step.
---- Teach Peace. It's Cheaper Than War.
LetsCrypt is an easy method to get a cert and use it.
Unless you're trying to obtain a certificate for the administration interface of an internal device on your home LAN, such as a router, printer, or NAS. Then you have to not only use Let's Encrypt but also buy a domain. If you try to use Let's Encrypt with a free subdomain owned by a dynamic DNS provider, you're likely to hit the weekly rate limit for the registered domain under which your subdomain was issued. Or have the major dynamic DNS providers completed the Public Suffix List add process for all their subdomains yet?
I shouldn't have to get a cert to pop up a website, period. The fact that people like you think we should is foolish, stupid and a road to hell.
Good-bye
Why do I need to use HTTPS on a website I create that is totally public, offers not login/forums, and takes no payments. Maybe a site dedicated to building Control Line airplanes?
Two reasons: So that the ISP can't modify the page in transit to include advertisements or other unwanted elements, which Comcast has been caught doing. Also so that the ISP can't use the URL paths that their subscribers visit to build interest profiles on their subscribers. With HTTPS, the man in the middle sees only the hostname (e.g. "tech.slashdot.org", not the path ("/comments.pl?sid=12295934&cid=56872990").
You can walk into libraries all over the world, pull a book off the shelf, and read it. Nobody maintains it; it just sits there. Some things work that way.
Just think of the lost opportunities!!
Why, with just 2 months and $200,000 we could start modernizing these "books" so that they use a proper 1px razor-thin font, a 20% contrast ratio, and nice 30% transparent pages. Another 4 months and $400k and we can upgrade them to require batteries and use AI to replace all those long paragraphs with summaries. And lastly, in just 1 year and a million dollars, we can add encryption, fingerprint readers, dynamic advertising, and pay-per-chapter so that only people with an active subscription or make use of the freemium model can read them!
Books-as-a-Service with nice modern UX, targeted advertising based on book genre, and microtransactions. Let's get started! Now, who will fund us?
"What do you despise? By this are you truly known." --Princess Irulan, Manual of Muad'Dib
/)
It is not misguided at all. Google wants a monopoly. They don't want any other company to have the ability to monitor what users are doing. Forcing https achieves this goal.
1) It reduces the number of trackers, which since they still track most sites through their analytics, raises the value of their data.
2) It gets people used to Google dictating how their websites look and function.
Your ad here. Ask me how!
I was on the side that agreed with your statement.. But then I thought about it for a while... non HTTPS traffic (plain HTTP) can be modified in-stream. I think it was Comcast that was caught injecting ads into HTTP traffic a few years ago. You cannot do that with HTTPS. Do you want your ISP injecting or modifing the webpages you are trying to read? Besides, nothing prevents anyone from having two or three browsers.. If chrome isn't cutting it for you, there's always alternatives.
So.. maybe a position reevaluation is in order?
Let me turn that around for you. You use somebody's public Wi-Fi, and it asks you to click on something that installs a new root cert. If it is easy, the average person will do it without hesitation, at which point HTTPS is completely broken.
Sometimes, there are good reasons to make unusual things hard.
No, the right answer is for somebody to come up with a sensible standard for .local certificates in which they are accepted with SSH-like behavior — ask once, and never ask again (with no expiration), but accepted only for that specific hostname, never allowed to be treated as any sort of root cert, etc.
Check out my sci-fi/humor trilogy at PatriotsBooks.
On the other hand, it will put the power of censorship in the hands of domain name registrars, TLS certificate providers, and whomever has the power to decide which certificates are "not trusted" (Google).
1. Privacy, so that ISP's and other companies don't get to record which old files you access and when
This is bullshit. It's been proven to be bullshit. Creeps in the wires know where you are going. They see IP headers, SNI indications, public key identities and TLS session keys. They know size, timing and length of transfers.
This is sufficient information to deduce exactly what you are doing on a publically accessible website with high degree of accuracy regardless of encryption.
Have a look at the CAs accepted by your browser. Do you actually trust each and every one of those entities to never issue a cert in error? Have you even heard of most of them?
about HTTPS. You just answered my question. They don't want the ISPs to have the detailed data google has (they still have URLs but no page content) and they can't replace google's ads with their own. Now it makes sense.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
Wait until you find wire-sniffing apps inside your (expletives deleted) routers, or someone that's programmed a router port mirror to a tor listener. Security isn't that tough, but it eludes thousands of organizations. Look at this weeks, largest-ever breach in Florida, where most all of the living population of the United States had their names, addresses, and a few other juicy fields snarfed because of stupidity. The basics should include TLS 1.3.
Then you are already fucked. Period. There is nothing stopping the attacker from doing the exact same thing, but easier on your computer, all while being able to read the information in the decrypted form. That means the attacker is already in your network and can chain exploits until they own everything.
Not to mention - why the FUCK would I need HTTPS to view a page that has been sitting around since 1998, is static HTML, likely has no ads plastered all over its face, and contains information on something obscure and random that newer pages don't have anymore? There's no reason for encryption for these older pages. Ever. There is no login information, user credentials, or even scripts being executed. It's fucking HTML, if the browser manage to fuck it up enough to be an exploit maybe, just maybe we should be looking at securing the browser instead of the transfer at that point.
To err is human; effective mayhem requires the root password!