What statistics are we collecting? How does that corelate to what a user wants the company to do in terms of allocating resources? With this proposal users can express in broad terms what they want from the system while those who vote directly can pick the best delegates to perform those functions.
When you make decisions it has to be data driven. You need to collect data first and then analyze it. So for example how do you know what the problems are if you don't have some data showing it? How do you know what the risks are to the Bitshares ecosystem if you don't have some data backing that up?
If you just ask people to allocate resources based on their feelings then a psychological marketing campaign could convince the community to burn money by focusing on improvements which might have a very low impact for the users of Bitshares. In fact some improvements might actually have a very negative impact and could hurt the community.
So when you're talking about resource allocation you have to collect statistics prior to and during the decision making process. How do people use Bitshares? How secure do people feel while using Bitshares and what types of security do people use the most? How often is multisig used and how much do people value transparency/privacy based on actual use of these features?
If you know the most commonly used features then you can then ask "how important are these features?" within the app itself. Like for example the user could use a particular feature in Bitshares itself and within the app submit data indicating they rate that particular feature 5 stars while another feature might get 1 star. They could discuss the feature within the app anonymously like how purchasers of iTunes music discuss albums in the review section.
This would produce statistics which show the actual usage patterns of Bitshares without compromising anyone's privacy. This data could be used to show that most people will need certain features to support their usage patterns. When you combine these statistics with a poll then you can look at the poll and the statistics and if the data matches you know something has to be done.
There could be times where the statistics don't match what is going on in the polls. Such as if there is a FUD campaign not really based on any change in use statistics. Say if somehow people become really paranoid about privacy and are loud about it in the polls yet no one uses the feature. You would have to then question why the use statistics don't match what is in the polls.
If we are talking about governments or corporations then they direct resources based on statistics. They collect statistics to determine what is needed to keep the business or state running. Businesses collect statistics on what kind of chairs and office designs improve or reduce productivity and governments collect crime rate statistics to figure out where to direct resources.
So in our case we would need usability statistics, we would need use statistics that the user themselves can review (such as you spend X amount of money on security features you never use). As long as the proposal you describe is focused on being data driven then I can support it.