Frequently Asked Questions

What's a CAPTCHA, and why do some websites use them?

A CAPTCHA involves a challenge that a website visitor must pass before performing an action (such as posting a comment or creating an email account). A CAPTCHA challenge is designed for humans to pass almost always and for bots (software programs) to fail almost always. This limits the effectiveness of bots from repeatedly performing the action guarded by the CAPTCHA. In this context, the bots are generally being used by spammers to maximize transmission of messages.

How is peoplesign more secure than text-based CAPTCHAs?

Text-based CAPTCHAs are known to be ineffective against sophisticated bots.
These CAPTCHAs have been around for nearly a decade, and spammers have acquired sophisticated bots that can defeat them (achieve a high pass rate against them). Text-based CAPTCHAs used by Google, Microsoft, Yahoo and others have been effectively circumvented by bots--and the attacks seem to be increasing. Techniques for performing distorted optical character recognition are mature and widely available.
Challenges like peoplesign are more difficult for bots to solve.
Humans effortlessly use a highly sophisticated skill set to solve challenges like peoplesign. For peoplesign in particular, this skill set must include idiomatic natural language processing and complex image recognition. Nobody has been able to endow even a supercomputer with equivalent skills, let alone a bot accessible to a spammer. Scientists are still trying to fully understand the depth and breadth of this skill set. When/if they eventually succeed, it may take years for top computer scientists to endow a computer with a matching skill set. Years after that, elite spammers *might* have access to bots with a matching skill set. Until that uncertain future, peoplesign will be available as an effective countermeasure to bots available to spammers.

Where is the source of your images? How many images does peoplesign have?

We snap all of the photos. Our image database is 100% dedicated to peoplesign. The size of the beta database is formidable--large enough to generate 1000's and 1000's of challenges with no image repeats. All images are unique before any distortion is performed. However, the rate of growth of the database relative to the number of challenges served is more important than it's current (beefy:) size. The core technology of peoplesign is a non-trivial set of techniques to rapidly and economically add unique, annotated images to our private database(s). This technology is patent pending. Note we've personally invested a lot to fairly develop the collection of innovative intellectual property comprising peoplesign. We're prepared to spend everything else including our last breaths to defend it if necessary.

Ok, hotshot, if you can generate labeled images so fast, and it's so hard for bots to match the right labels, why do you need to distort the images?

It's an extra precaution. Two pictures, p1 and p2 may contain the same object, x. A highly sophisticated bot/spammer that has seen p1 and figured out the correct label might scan new images for the presence of x using standard computer object recognition techniques. We've anticipated this type of "attack" and have highly effective defenses against it. To develop these defenses, we identified a set of object recognition techniques (texture based and color based) that might be used in an attack. We then evaluated a wide range of image distortions that limit or completely eliminate the effectiveness of these techniques. We eventually converged on a set of non-linear distortions that significantly reduce the effectiveness of multiple object recognition techniques.

Great, you have elaborate countermeasures against object recognition and indexing. But isn't there still like a 1/100 chance of passing a challenge by randomly guessing? Couldn't such a bot simply guess at random all day long?

If such a bot starts guessing randomly and repeatedly, it's likely the bot will fail dozens of challenges before passing even one. That behavior is likely to allow peoplesign to identify the visitor as an antagonist and ignore it. However, if you're especially concerned, you can customize peoplesign for your site to make it even harder for bots to pass.

Yikes, I just failed a challenge. Does that mean peoplesign thinks I'm an antagonist?

No. First of all, you've done nothing wrong at all (assuming you looked at each picture and read each label before making a choice). In these situations we believe it's more accurate to say that the CAPTCHA failed to correctly identify you as a human! Remember, a design ideal for a CAPTCHA is for all humans to pass all the time. (For the moment, please ignore the fact that peoplesign is only available in English, give us some time to translate:) We're sorry if you've "missed" even a single challenge. We regularly review our mistakes and tune the system to reduce future mistakes. But don't worry, even if peoplesign is having a really bad day and you miss 3 (or more) challenges in a row, you won't be marked as an antagonist.

With that being said, we think you'll be able to pass peoplesign at a much higher rate than most of those "annoying squiggly character" tests.

Do you offer an audio based CAPTCHA with peoplesign so blind people are not blocked?

Regrettably, we currently do not offer an audio CAPTCHA. Individuals with visual impairements can use web-browsers with dictation software to listen to the contents of webpages. However, images basically can't be translated. Since text-based CAPTCHAs consist of an image of text, the image can't be understood by visually impaired individuals, they are unable to pass the test, and therefore their access is denied. The current version of peoplesign has the same limitation. We're working on a solution; please be patient. Until then, we sincerely apologize for any inconvenience caused.

Tell me about Myricomp LLC, the company that owns peoplesign.

Myricomp is a small, privately funded company in California. Myricomp's agenda is to provide net value to its social context in exchange for fair profits, without ever compromising our integrity or social responsibilities. You should demand we adhere to this agenda, as you should with all companies in your social context.

Are you serious?

Yes, absolutely.

news  FAQ  terms of use  contact