Insufficient Information: Too Many Unknowns !

Agree with the first part Sandilyan - We are all on the same page in this regards.

However, if there are too many unknowns, you can never tell if they are threats or opportunities or threats that can be changed into opportunities. It rarely happens that a project is initiated and then terminated because there are too many unknowns (At least I never encountered this). It happens a lot that a WP has too many unknowns and then resolved with progressive elaboration. Moreover, for the Unknown-Unknowns, there is always a management reserve.

This is a good point that you raised and would like to know your opinion with regards to the above mentioned in addition to Suhail's.

Agree with your pt. of view Rami.
Reg. second part from my previous comment, in my opinion - If too many unknown-unknowns then even the management reserves we cannot set it, since this not part of your schedule baseline anyhow, we may eventually end up utilizing and exhausting / over project budget very fast. If the project has some known-unknowns and is manageable then it’s good to go-ahead with this project. We might have already done feasibility study by which we would already come to know about this position – is of having of too many unknown-unknowns in this project. If we come to know too-many unknowns at later stage of the project, then it’s no wonder that it’s a good chance for failure. Only having too many unknown-unknowns, it is not advisable to proceed on this project, and we’re taking a very high risk of executing this project, without knowing what we’re doing. Management reserve is only small portion, 10 – 15% of your total project budget typically, we will need to guestimate & try to evaluate max possible at least to see how much we could accommodate if too-many unknown-unknowns. Preliminary estimates (ROM) and feasibility check / study should really help you to go or no go into this project. Like others had mentioned, if we have already had the SMEs / expertise in this project, couple of unknowns could have been unearthed in the initial stage itself, being done this activity, taking into further stages with “too many unknown-unknowns” is a great risk towards, and we’re working towards the failure of the project and its resources, also we know it is going to overshoot your project budget for sure. The best is execute only if and when the project is manageable and we know what we’re doing and going to do in the project. It is as equivalent as having too many unknown-unknowns in the project is that we do not know what the project is all about. As simple as it is, better stay away, so that we can cut down the cost and resource time at least! ?

I got your point now, I agree with you in this sense when the UN-UN are for a project as a whole but please note that I was talking about too many unknowns in a certain WP or Activity, not about the project as a whole. If the project as a whole has too many unknown-unknowns then the limit to management reserve is the sky and this is a very high risk and would lead to project failure. I agree with you that we should not proceed.

If it was a certain WP or activity then Risk Analysis, Rolling Wave Planning and Progressive Elaboration would do the job.

Yes, exactly. We're on same page.

You can't use Delphi Technique when you have unknowns. Delphi technique is used when you have too many opinions and you want unanimity (Best and One Opinion).

You can't provide good estimate if you have too many unknowns even if you have experts. The only way to do that is Roll Waving, Risk Assessment and Progressive Elaboration.

This is my humble opinion.

Very pertinent argument by Rami on Bernard Gore's comment about use of Delphi technique. You can only use Delphi technique in trying to identify as many risks as possible but that is not a guarantee that no risk will be left to be found. There still will be unknowns, and providing estimates for something which is unknown is UNKNOWN to me.

Interesting discussion, just catching it.
Rami, Suhail and Stéphane you cover it pretty well.

When too much information is missing, it is more than a contingency you need, risk approach is a better option.

And thought a series of iteration (rolling waves) you should be able to progressively reduce the unknown part, then it could be remove from the risk and handle back as contingency. The margin of error would progressively diminish from +-100% to +-50% to +-25% to less and less, often presented graphically has an horizontal Tornado, left with a big opening to eventually 0 (Barry Boehm's "Cone of Uncertainty").

In a project there should always be a single risk that is unknown-unknown, a catch all risk.

The only other option I've seen is using expert advice, that would add his/her experience and knowledge.