Lee Oddens Poll and WPW

[kgun]

I post this post here, since it is related to programming, making poll tools.

Background: What is your favorite SEO/PPC Forum?

Result:

• WebProWorld (25%)
• High Rankings (19%)
• WebmasterWorld (18%)
• Digital Point (13%)
• SEO Chat (7%)
• Search Engine Watch (6%)
• Syndk8 (5%)
• Search Engine Roundtable (4%)
• V7 Network (2%)

I made this end post:

"So what was the conclusion from the poll?

Lee Odden wrote:

“We may need to find another tool for conducting polls”.

Is there any reliable automatic (computer based) method which can be used to make an unbiased pool on the internet? If so, with education on the master LEVEL (not degree) in Statistics and on the Dr. Scientlevel (no degree) in mathematical finance and numerical analysis, I am very interested in being drawn attention to such a method (see my comments below).

Note:
No ranking method is also a method and the only I know that can not be manipulated.

My conclusion so long:
The best ranking methods are those made by professionals.

Example: My own site

RedCarpetRank.com"

Any comments?

[kgun]

I think there is one way that it is difficult to be "many persons". The poll must be related to a digital ID like openID that gives every person a unique identity. In theory a digital signature shall be more secure than a nondigital.

How easy it is and when / if it will ble implemented is another story. Will giving away your ID to a poll restrict somebody from participating? Then the result will be biased. That is elementary statistical theory, even if there are methods to compensate for this bias if it is systematic.

They call it Advanced Poll 2.0.8. How advanced is it?

[Dinghus]

It would seem that this is the reason why internet polls are not scientific

[kgun]

They can not be scientific as long as it is easy to be "many" persons. You can visit a library, a class room and move from Pc to Pc to vote. You can use a dynamic or a random Ip and you can delete cookies etc.

In my view, you can not rely on an online poll before there is an unique way to identify each voter.

The problem is that if such a system is launched, the poll will be biased, if some people are more reluctant to give away their ID. Who will give away their digital ID at all on an internet page?

Conclusion:
Do not rely on an online poll.

[kgun]

Here is one of the most famous Norwegian fashion sites, DetNye (Don't ask me about design ).

Today, our family's greatest lobbyist, my oldest daughter, called me and asked me to vote for my youngest daughter in the competition for "the face of the year."

There, I noted a new way to conduct online polls:

1. Fill out name. (Will it be checked)?
2. Adress. (Will it be checked?)
3. Phone number. (What about people without a phone? What about people with access to x phones?)

It is still possible to manipulate the poll, like it is with any poll that is the target of lobbyists. Not even a digital signature is immune against lobbying.

Conclusion:
The only poll that can not be manipulated, is no poll. Finally the ability to manipulate online polls has entered our largest news channel.

[wige]

Any online system is subject to hacking or manipulation. The typical issue is one of diminishing returns. The more secure you make the system, the higher the cost to implement, either a real cost or a cost in terms of time and resources. The more secure the system becomes, the more costly it becomes for someone to manipulate the system.

Take the example of the poll listed, an informal poll to get opinions on a preferred forum, compared to a more important poll, such as running an election online. For the informal poll, there is a limit to how much time and resources can be devoted to checking: this could be done by restricting the time in which people can vote to prevent changes in dynamic IPs, restricting votes to certain countries or regions to prevent spam proxies, and a few other simple steps. This lets the poll be implented fairly easily, and makes the cost higher than the potential payoff for someone attempting to manipulate the system - too much work to bump one forum a spot or two in an informal poll.

Now what if your country were going to use an online poll to elect the next president? Obviously, if the same protections I mentioned above were implemented, they would no longer be sufficient, because the cost/payoff ratio is reversed - not all that much work considering you could change the outcome of an election. This is where a more intense digital signature system would come into play, high level encryption, very short period of time to vote, etc. along with the promise of prosecution for anyone attempting to manipulate it.

Security is not about making the system absolutely secure - thats an impossible task. Security is about making the system so secure, the cost to break it is higher than the rewards.