I blog extensively at WordPress, which uses akismet for spam
filtering.  I am frequently astonished at how well it works.  I
average maybe five to ten real comments per day and perhaps twice as
many spams.  Of those, maybe one or two real comments PER YEAR are
falsely flagged as spam, and perhaps five or ten spams per year not
caught.

Highly recommended.

-- Mike.


On 24 October 2011 15:42, Jonathan Rochkind <[log in to unmask]> wrote:
> Using the free Akismet API used to work fairly well for me, and is
> completely invisible to the user, it basically uses a behind-the-scenes
> maintained blacklist of IP addresses/email addresses, and other heuristics
> to identify spam.
>
> Haven't used it in a while though, don't know how well it's doing lately.
>  Curious if anyone else has recent experiences. In case of false positives,
> might want to maintain the list of 'spammed' messages and review sometimes,
> which is a pain. But I don't really know any other way to do it;  recaptcha
> is just awful, and poses accessibility problems.
>
> Certainly another option, if it's feasible for your use case, is requiring
> authentication to patron account before accessing form at all.
>
> http://akismet.com/development/api/
>
> On 10/24/2011 10:17 AM, Andrew Darby wrote:
>>
>> I've found that a simple skill testing question does wonders for form
>> spam reduction, e.g.,
>>
>> What is five times five?  [input box 2 chars wide]
>>
>> Maybe I've dealt with a dumber class of bots, though . . . .
>>
>> Andrew
>>
>> On Mon, Oct 24, 2011 at 10:12 AM, Andreas Orphanides
>> <[log in to unmask]>  wrote:
>>>
>>> Here's a method that's by no means foolproof but is practically zero cost
>>> (you may be using a version already). Disclaimer -- I have not actually
>>> tested this to any extent:
>>>
>>> Include a text input field in your form that needs to be blank for the
>>> form to validate in the back end. Keep the field hidden with CSS (or
>>> z-indexed behind another element, size set to zero, etc). Users will never
>>> see it, so their forms will validate; I doubt that most spambots are
>>> sophisticated enough to check whether a form field is hidden or obfuscated
>>> before filling it in. Then silently reject submissions with that field
>>> filled.
>>>
>>> I am not sure whether this would cause any problems with tab navigation,
>>> screen readers or other assistive technologies, but you may be able to do
>>> something to sidestep those issues.... On the other hand, captcha brings its
>>> own host of accessibility problems.
>>>
>>> One other disadvantage is that this might be hard to implement in a
>>> CMS-based form plugin. But if you're coding forms the old-fashioned way,
>>> it's worth a shot.
>>>
>>> -dre.
>>>
>>>
>>>
>>>
>>>>>> "Parker, Anson (adp6j)"<[log in to unmask]>  10/24/2011 9:36
>>>>>> AM>>>
>>>
>>> Mollom is pretty decent...
>>> http://mollom.com works with a lot of cms's
>>> It is commercial with 100 free positives per day, and can require
>>> captcha,
>>> but it tries to avoid it with a crowd sourced algorithm approach
>>>
>>>
>>>
>>> On 10/24/11 9:26 AM, "Ken Irwin"<[log in to unmask]>  wrote:
>>>
>>>> Hi folks,
>>>>
>>>> Some of our online forms (contact, archives request, etc.) have been
>>>> getting a bunch of spam lately. I have heretofore avoided using any of
>>>> those obnoxious Captcha things and would rather not start now. (I
>>>> personally loathe them and they keep getting harder, which tells me that
>>>> the spambots are probably better at them than we are...)
>>>>
>>>> Does anyone have some good/easy/free/less-stressful spam-inhibiting
>>>> ideas?
>>>>
>>>> One that occurs to me to try, and I have no idea if this would match
>>>> well
>>>> with actual bot behavior: at the time the form loads, include at hidden
>>>> field with id=[unixtimestamp]. When the form is submitted, ignore any
>>>> forms that took less than (10? 15? 20 seconds?) to fill out on the
>>>> assumption that bots probably do it way faster - or possibly way slower?
>>>> Do they save them up for later? Should I add an upper bound? Is this
>>>> just
>>>> a really dumb idea?
>>>>
>>>> If I try that one, I would start not by eliminating the bad results but
>>>> by marking them as spam and seeing how effective it is.
>>>>
>>>> Other ideas? (PHP-friendly answers would be easiest for me to implement,
>>>> but others may work too.)
>>>>
>>>> What works for you?
>>>>
>>>> Thanks
>>>> Ken
>>
>>
>
>