Why client-side verification is bad(blog.dryft.net)
blog.dryft.net
Why client-side verification is bad
http://blog.dryft.net/2011/07/why-client-side-verification-is-bad.html
17 comments
Yes absolutely. Programmers/devs often get way too wrapped in doing everything _right_. But doing it right takes time and money. Commercially it makes sense to get away with the bare minimum. The holes will only be an issue if the product actually takes off.
This is a very good point. After all, it is just a game, not a banking application, and a pretty simple game at that.
This attitude throws the baby out with the bathwater.
Client-side verification is Good, it saves bandwidth and reduces annoyance to the user. Client-side verification without also checking on the server is bad.
Given the example of a word game to be played on a phone. I would argue that client-side verification of valid words is vital.
Client-side verification is Good, it saves bandwidth and reduces annoyance to the user. Client-side verification without also checking on the server is bad.
Given the example of a word game to be played on a phone. I would argue that client-side verification of valid words is vital.
The title should probably read "Why relying only on client-side verification is bad".
I don't think the author is suggesting that doing client-side verification is a bad thing in itself.
That's probably what's in the authors mind, but not what's in the article. The author is not just suggesting that client side verification is bad in itself, but stating that position quite precisely.
"Why client-side verification is bad.."
(rather than the title you suggest)
"to my surprise discovered that it did client-side verification of the words"
(rather than "...that it didn't do server side verification")
However, in support of your position; in his opening paragraph, he does say "you can't rely on it to do authentication", rather than "you shouldn't use it to do authentication".
"Why client-side verification is bad.."
(rather than the title you suggest)
"to my surprise discovered that it did client-side verification of the words"
(rather than "...that it didn't do server side verification")
However, in support of your position; in his opening paragraph, he does say "you can't rely on it to do authentication", rather than "you shouldn't use it to do authentication".
The author wrote "A common mistake is the web site that relies on javascript". I think the key word is "rely". You either rely on something or you don't. It's not a partial measure.
Author's comment on his site:
"It's not wrong to use client-side methods to do initial validation, but it's wrong to trust that validation on the server side."
"It's not wrong to use client-side methods to do initial validation, but it's wrong to trust that validation on the server side."
The comment you cite (which is new since my comment above) confirms my point that it might be what he thinks, but not what he says. The comment continues:
"I probably wasn't very clear on my feelings though."
"I probably wasn't very clear on my feelings though."
In this specific instance there is also validation by the other player. Hacking the game is obvious and the other party will refuse to pay.
In more competitive games like Starcraft you can't trust the client (maphack for example).
In more competitive games like Starcraft you can't trust the client (maphack for example).
It's also not exactly the only way to cheat (low-tech solution: spend a day with your grandma, who is an absolute Scrabble wizard.)
There was a competition for university students in the UK a few months ago from some investment company, trying to raise interest for post-grad careers with them, basically a several-stage speed test, top five speeds at the end of the time period would get to choose a prize, iPhone, cash, few other things.
A friend of mine (a student) asked me if it was possible to create macros in a browser to help go faster, and since I was bored and lack morals in this area... just took a look through the JS, and fifteen minutes later and an injection for him to paste into the URL bar and skip through each stage, making the game think he had completed it, as fast or slow as he wanted. So he got a free iPhone.
A friend of mine (a student) asked me if it was possible to create macros in a browser to help go faster, and since I was bored and lack morals in this area... just took a look through the JS, and fifteen minutes later and an injection for him to paste into the URL bar and skip through each stage, making the game think he had completed it, as fast or slow as he wanted. So he got a free iPhone.
(thinking more about web pages than apps here)
are there any frameworks that let you re-use the same verification code on both client and server? i'm imagining some way to decouple the validation from the presentation and then run the validation part on both sides - i don't see why this wouldn't be possible with something like Node.js.
related, i guess that some frameworks (particularly ajax) support duplicating validation on client and server by automatically generating (different) client (javascript) and server-side code (python/ruby/java/...) from simple specifications (eg regexps). does anyone have good experiences with any of these?
are there any frameworks that let you re-use the same verification code on both client and server? i'm imagining some way to decouple the validation from the presentation and then run the validation part on both sides - i don't see why this wouldn't be possible with something like Node.js.
related, i guess that some frameworks (particularly ajax) support duplicating validation on client and server by automatically generating (different) client (javascript) and server-side code (python/ruby/java/...) from simple specifications (eg regexps). does anyone have good experiences with any of these?
Well it's not a framework but asp.net mvc can actually do pretty much exactly this with data annotations.
http://www.asp.net/mvc/tutorials/mvc-music-store-part-6
http://www.asp.net/mvc/tutorials/mvc-music-store-part-6
Client side validation is good for speed, however should not be trusted when sent to server. Server side validation is for validating such information to prevent issues. eg SQL injections.
But you can't trust client-side validation, unless you run the client in a DRM container (client then can just sign the request and you can assume that it executed in a secure memory, unaltered).
Words with friends really needs both, without client side it would be tedious for attempting a word, of course this could also cut down on people trying their letters just about every way they can until they get a word.
This is obvious, isn't it?!
Everything the user has access to can be manipulated.
You'd be surprised how many applications and web sites still rely solely on client-side validation, even these days. Either don't think at all about threat scenarios ,or they trust their "Do not reverse engineer" EULA to protect them :-)
wow, I suppose we shouldn't begin to talk about sql sanitation...
Sorry, what? Client-side verification is categorically not bad. It’s when you don’t combine it with a round-trip to server-side verification that you have problems.
Client side verification is good to prevent unnecessary hits on the server when the input is incorrect/incomplete , its not.. in any way.. a real input validation.
"It wasn't very hard to circumvent this, resulting in a version which accepts moves like the above one.."
I'm more curious as to how you go about circumventing this. I would imagine that it probably IS very hard.
I'm more curious as to how you go about circumventing this. I would imagine that it probably IS very hard.
Surely this is just common sense at this point?
Validation client side is just a way to make user interaction faster. Nothing more, but quite useful, indeed.
Wanted to comment on the article but all I could think about was "READ THAT EMAIL IN YOUR STATUS BAR"
Does HN use client side verification for votes? Otherwise how did this get here?
Not bad, insufficient.
it is very annoying to have to duplicate validation logic though.
Fundamental truth = no code should ever trust its input. Break this rule never. If you have optimization problems, then don't worry because they're better than security problems and the hardware will likely solve it for you by the time you release. Trust relationships between bits of any codebase mean 'not loose coupled' = fundamental design flaw for simple, maintainable and secure code. Unix design philosophy: "do one thing and do it well" (and that means checking all your input religiously for every function)
But think about it for a moment. Assuming the ideal product would have both client + server verification, it's clearly less work to only do client-side verification. The product can be released quicker.
Maybe they intentionally forgo the server-side verification code until they actually get cheaters. Then they implement it before it begins to hurt their bottom line.
Because what if the game was an utter failure, and never got any users? How glad would you be that you spent all that time writing the server checks for every game action?