Spyware Test Summary -What We Learned.

(A) Conclusions:

• The best performing scanner was CounterSpy, removing a total 85% of all critical detections across all tests.
  AdAware being the least effective at 31%.
  For full results of all products in tabular form, click here.

• No one scanner can be expected to remove all spyware at all times.
  We propose a better way to remove spyware by combining free program features, based on these test results and our real life experience.

• Some proprietors of anti-spyware clearly pick and choose which threats their software will remove AND/OR spyware vendors change the code of their products over time.
  e.g. Back in 2004, as tested on SpywareWarrior, AdAware removed 80% of WhenUSave detections. Disregarding a possible difference in classifying 'critical detections'(minimal), AdAware now removes 0% of the current WhenUSave detections.
  This could be on account of WhenUSave being bundled with the (now decommissioned) Edonkey P2P client.
  Removal of WhenUSave would disable this product as alluded to below in point 3 of 'Limitations.'

• A few of the spyware tested infected Mozilla Firefox. Mozilla is generally a safer and more secure browser as outlined in our discussion on Online Security.
  However, few programs removed the entries associated with 'WhenUSave' and 'ZangoMedia' infection.
  SpySweeper was the only program to remove all 'detections.'
  Using SpySweeper as your main anti-spyware program may be a better option if you choose to use Mozilla Firefox as your default web browser.

• Preventing initial infection is a preferred way to remain safe from these threats, as often their presence can 'open doors' to infection from even more spyware.
  A better method to protect your system with free real-time scanning is discussed here.

• And preventing spyware is only 1 aspect of online security- Safe computing habits should be adopted and include avoiding unknown or dubious websites and emails.
  

(B) Limitations:

There are many limitations imposed on these tests and every effort is made to minimize their effects.

• The sample size of spyware products tested.
  Only 7 'pieces' of spyware were tested from a pool of thousands. To minimize some of this effect, spyware were selected from a list of current, wide-spread threats as outlined in the Test Methology.
  No doubt the anti-spyware programs tested would produce different results with another set of spyware.
  

• Human error in testing, collating and reporting results.
  All processes are duplicable and conducted with reasonable care to produce as accurate a result as possible in this spyware test summary.
  

• Tests results show removal of 'critical detections' (files and registry keys) only. No account is taken of overall removals performed by each anti-spyware scanner, nor memory processes killed as a result.
  A scanner could rate high in removal of critical detections, but may leave a plethora of other registry entries and files behind. These combined with other 'orphan' entries can accumulate over time and contribute to loss of system performance and crashes.
  Additionally, some files and registry entries could be shared components and removing them would result in instability, often disabling the parent application. This is common amongst free, advertising-supported programs.
  

Given limitations of any test, the spyware test summary reported herein provides only a simple basis for any conclusions drawn in (A) above.

Finally, much kudos must go to Eric Howes who paved much of the way for the framework from within which testing spyware should take place.
His site "SpywareWarrior.com", was an inspiration to my own exploration into this realm.
Unfortunately the site is no longer current as Eric has moved on, and any testing he performed in 2004 can no longer be considered current.