Google and Spam

CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart)

I suggested this to them back when things hit the fan.

Neither do I.

I tried it on alt.test and that was NOT the case. I think it's clear that Google won't do anthing that costs them traffic

--of ANY kind.

That explains why there's been a typical amount of spam over the last few days.

Filters, relax!

Tim

-- Deep Fryer: A very philosophical monk. Website @

I went and had a look at:

as I have CAPTCHA on my web site contact form, but wondered what the letters really meant, and see Alan Turing gets a mention.

But, have a look at the third series of alpha-numeric pictures on the above page. Maybe I am getting old, but some of these are impossible to read, and when they decide to enhance them with back ground drawings, well, how does a human get to actually contact anyone in future?

I would find a mathematical question a lot easier, even if it was 3 or 4 levels of operations, with decimal point answers.

Cheers Don...

--
Don McKenzie

Site Map:            http://www.dontronics.com/sitemap
E-Mail Contact Page: http://www.dontronics.com/email

Intelligent 2.83" AMOLED with touch screen for micros:
http://www.dontronics-shop.com/product.php?productid=16699

Spam? What is that ?:-) I _never_ see any post from Google, unless you're white-listed.

...Jim Thompson

-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: "skypeanalog" | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | America: Land of the Free, Because of the Brave

If they don't use it for original posts, it's useless. The countless spam postings that pollute the busy newsgroups will continue to get worse. The CAPTCHA won't make any difference at all. Total bullshit. "Don't be evil," my ass.

The problem, of course, is that computers are so damn good at maths.

--
RoRo

I've got an idea on my brain's back burner - to do something much like that, but instead of distorted letters, it would show a picture of a random animal - really obvious ones, like "cat" and "dog" and "lion" and "horse" and so on, and say, "enter the N-letter name of this animal", and give them a hint like the first letter:

[pic of lion] Please enter the 4-letter name etc: [L___]

But, I haven't got a round tuit yet. ;-)

Cheers! Rich

D>I went and had a look at:

Yup. Telling humans from computers is the essence of The Turing Test.

Now throw into the mix someone with a true vision problem. 8-(

Google's are often similarly difficult. When Google uses a 7- or 8-letter CAPTCHA, they are almost always a pain (the lines get thin and wispy or just make a clump). I usually click for another until I get a 5- or 6-letter one.

I still say the OP didn't get a CAPTCHA the way he thinks he did; the only way I have found is to click an obfuscated email address. I believe Google inherited this from Deja and it appears they haven't put any more effort into it.

As has been pointed out, Google needs to challenge EVERY action with a test if they have any interest in stopping spam. (I'm not holding my breath on that. As has also been mentioned in this thread, the *Don't Be Evil* thing WRT Google is so pre-IPO.)

Slashdot and Network Solutions use a black & white CAPTCHA that is a monospaced font with lines running thru it. Those are easier on humans and I haven't heard that a machine has cracked it.

Wikipedia has a black & white CAPTCHA where the borders of the letters are like a thorn bush. That is also very easy for humans; not cracked yet either.

Since the CAPTCHAs of multiple Web mail services got cracked in recent(?) weeks, there has been a lot of chatter about this on Slashdot.

Puzzles do seem to be the logical evolution. Rotating questions like *Click the picture of the kitty* (when a leopard and tiger are the other pictures) would be easy enough that kids could do it. Arithmetic problems would be easier for blind people than anything currently in use.

I think inclusion of another level of

*random selection of a CAPTCHA #method#* will be necessary to leapfrog the progress being made in cracks.

Bank of America does something like that for their online banking, as a security measure.

-mpm

One of my banks uses an interesting PIN entry system for online access. You select the numbers in sequence from an on-screen numkey pad. The keypad layout is randomised.

Too easy; a look-up table could solve that...

Best learn how to read them or you may be stuck. You could always revert to snail mail.

It would fail to a fairly simple keyword detection and dictionary attack. Computers are also getting pretty good at recognising simple objects in pictures - distortion is about the most robust (ie computationally intensive) defense against automated abuse of mail.

The sheer beauty of CAPTCHA is that with a few exceptions the target pattern can be hidden by distortions from all but the most sophisticated AI programs (which would then take an age to read it) and still be relatively easy for a human. I have seen a few duds too.

Regards, Martin Brown

** Posted from **

...

Oh, well. Well, it's not the first time one of my ideas has "died on the vine." :-)

Thanks, Rich

Only problem with captcha is that humans can be smart enough to solve them, but not smart enough to work out when some spam site is fetching pages with captcha from somewhere, showing the captcha to said moron, who solves it, thereby letting the spammer abuse the originating site, while rewarding said moron with a juicy dog biscuit, er, image.

Detecting human intelligence isn't always enough...

Clifford Heath.

Unfortunately, you are so right. Social engineering attacks which either persuade the target or someone who knows the target to give away password or other critical secure data are all too easy these days.

Top phishing sites are extremely good imitations of real banks too!

Regards, Martin Brown

** Posted from **

No, it will just drive the posters from email to usenet news. It is a number of recipients thing. The supposed "breaking" of CAPTCHA tests is due to paying desperate twits to hand over gmail accounts for a few bux/rupees/geld.

This is already being tested out. I tried it out and took 4 tries to figure out what various images were because they distort it and add background noise. It was multiple choice answers and I still couldn't figure out what the image was.