June 19, 2007 at 13:05Protection against bots, fewer digits and beeches
To read this article, enter the sequence of characters from the image.
Attention! Be case sensitive and character language.
With the improvement of spam bots, defense mechanisms are becoming more complicated, but a phase has already been reached when many people cannot even make out the pictures.
Especially if these tests were not done by quite adequate people.
I personally met a lot of test pictures in which there are letters and numbers, to separate some letters from some numbers is a more difficult task than decoding the human genome.
Especially when authors use a variety of "stylish" fonts.
Latin lowercase "l" is indistinguishable from one, and the letter "o" from zero. Also, there may be difficulties with the register, because by complicating the work of bots, letters are scaled, and, for example, the lowercase “w” is not distinguishable from the capital “W” when scaling.
Protection options using mathematical operations are of little use - it is hardly harder to recognize numbers on them than in ordinary pictures, and bots can add and multiply bots much better than most visitors.
I will not say that I am reinventing the two-wheeled pedal car for driving on the ground, but here are a few alternative suggestions:
1. Use people's knowledge of basic truths, for example:
Nowadays, riddles are often used, but in addition to the fact that not everyone knows riddles (many of them are highly specific, and can only be known to the environment of the creator of the resource), in almost all the resources that I have encountered, they ask questions in plain text. If you combine the method of elementary truths and the "picture" method, the protection will be noticeably more reliable.
2. The ability of people to separate some reality objects from others, for example:
Indicate the number of circles in the picture:
In this example, numbers are also added to completely bring down bots.
How many bitten apples do you see in the picture?
If the questions themselves are also posted in pictures, the task for bots is complicated by an order of magnitude.
In this way, you can come up with many tests that are much more understandable to humans and more difficult for the bot than the most popular “alphanumeric” ones.
Attention! Be case sensitive and character language.
With the improvement of spam bots, defense mechanisms are becoming more complicated, but a phase has already been reached when many people cannot even make out the pictures.
Especially if these tests were not done by quite adequate people.
I personally met a lot of test pictures in which there are letters and numbers, to separate some letters from some numbers is a more difficult task than decoding the human genome.
Especially when authors use a variety of "stylish" fonts.
Latin lowercase "l" is indistinguishable from one, and the letter "o" from zero. Also, there may be difficulties with the register, because by complicating the work of bots, letters are scaled, and, for example, the lowercase “w” is not distinguishable from the capital “W” when scaling.
Protection options using mathematical operations are of little use - it is hardly harder to recognize numbers on them than in ordinary pictures, and bots can add and multiply bots much better than most visitors.
I will not say that I am reinventing the two-wheeled pedal car for driving on the ground, but here are a few alternative suggestions:
1. Use people's knowledge of basic truths, for example:
Nowadays, riddles are often used, but in addition to the fact that not everyone knows riddles (many of them are highly specific, and can only be known to the environment of the creator of the resource), in almost all the resources that I have encountered, they ask questions in plain text. If you combine the method of elementary truths and the "picture" method, the protection will be noticeably more reliable.
2. The ability of people to separate some reality objects from others, for example:
Indicate the number of circles in the picture:
In this example, numbers are also added to completely bring down bots.
How many bitten apples do you see in the picture?
If the questions themselves are also posted in pictures, the task for bots is complicated by an order of magnitude.
In this way, you can come up with many tests that are much more understandable to humans and more difficult for the bot than the most popular “alphanumeric” ones.