Issue Results Vector Plots

[Carkosa]

The NIP Collaboration has gathered data on the changes in nation stats (census and policies) caused by answering issues, as a function of those nation stats. This can be described as: each answer for each issue being associated with a 162 dimensional vector field (or flow, or section of a vector bundle - whatever terminology you prefer) over a 162 dimensional manifold (162 = number of census items + number of policies). One does not expect the vector associated with every answer of every issue to have a non-zero component in all of these dimensions, or to vary over all of these dimensions. However, all of this data must be captured for every issue answered in order for how these vector fields vary to be determined empirically. This is what the NIP Collaboration has been doing - capturing the position and vector of change for as many answers of as many issues as possible, as many times as possible. Attached below are several vector plots for the various possible answers of issue 23 (Uranium Deposit Promises to Enrich @@NAME@@). Our work has only just begun, and there are painfully few points so far, but we hope people find this 'first light' data as interesting as we have. Needless to say, these should be taken as a first indicator of possible results of a work in progress, not as a finished product.

In the plots, two positional dimensions are displayed, as well as the vector of change in those two dimensions. All of the other dimensions are ignored (integrated over). Note that in the plots the length of the displayed arrow is not the magnitude of the change vector, but it is proportional to it. Thus there is an overall arbitrary scaling factor in the lengths of the displayed arrows. However, that factor is the same across all datasets in a plot.


Below is for a full-bulldozing, in a capitalist nation:


Below is for a full-bulldozing, in a socialist nation:


Below is for no-bulldozing - ie: full environmental protection:


Below is for a compromise - partial bulldozing of the rainforest:


Below are the results of all four answers plotted together.


In conclusion, with respect to these two stats the issue behaves pretty much as one would expect. Notice, however, for the "partial bulldoze" answer there seems to be an economy- attractor somewhere around 60. The vector flow direction is not as perfectly uniform as one might hope. This might be due to dependence on the other dimensions that are suppressed in these plots, or could be random noise. There are probably not enough data points to draw any more precise conclusions from this data.

[Noldoria]

The results for ecological friendliness vs economy for the same issue is not at all at intuitively sensible as it was with environmental beauty vs economy above. As you can see from these plots, both of the 'full bulldoze' answer-options can increase ecological friendliness. That seems somewhat counter-intuitive, as the nation is presumably replacing a rain forest with a uranium strip/pit mine. After first sight there was concern that our data was corrupted or mis-aligned. However, double-checking the provided parameters values and deltas in the preserved webpage output confirmed that this data is correct. A couple of effects are visible that we presently have no explanation for. Considering the 'full bulldoze' answer-options first, it seems higher pre-existing ecological friendliness tends to increase the delta of the same, as does either a lower economy or socialist policy (due to the strong correlation we don't have enough data presently to determine which) - though this is clearly not a universal trend. The 'zero bulldoze' option increases the ecological friendliness more for higher economies. The 'partial bulldoze' option increases the ecological friendliness more for either lower economies or the socialist policy. The is also something odd happening at zero ecological friendliness for all answer-options except 'zero bulldoze' - this seems to be (within allowances for noise) a stable point in the ecological friendliness direction.

The same caveat concerning global normalization of the vector arrow (length) applies in this case also. This is a limitation of our plotting software, and not the data. We are using XmGrace with XYVMAP at the present, and there the arrow orientation and length must be provided in a viewing coordinate system, not the underlying parameter coordinates. Even if one did normalize the arrows properly for one plotting, auto-scaling, shifting or zooming the viewing coordinate system would ruin it. Eventually we provide vector plots in which the lengths of the arrows directly match the magnitude of the delta in the underlying parameter coordinates.

[Trotterdam]

The results for ecological friendliness vs economy for the same issue is not at all at intuitively sensible as it was with environmental beauty vs economy above. As you can see from these plots, both of the 'full bulldoze' answer-options can increase ecological friendliness. That seems somewhat counter-intuitive, as the nation is presumably replacing a rain forest with a uranium strip/pit mine.

Basically, what this means is that the money from the uranium mine is being used to fund better ecological protection in other parts of the nation.

[Merconitonitopia]

What Trot said. It's a function of rising income. Ecofriendliness is based on how much money the government spends on environment. Assuming a fixed proportion of GDP is spent, a higher income means the total rises, as the share is taken out of a higher sum.

The stat's description is telling: 'The following governments spend the greatest amounts on environmental issues. This may not always be reflected in the quality of that nation’s environment.'

A higher delta for higher preexisting EF values is because raising two values by the same rate yields a higher delta for the higher value. Socialist countries and/or low-economy countries (>90% of the latter are the former) seeing higher rises from bulldozing will be because most of these nations are recently-minted socialist nations, which tend to start off with low economies and low incomes. Income sees higher gains at lower values.
Higher economies seeing a higher rise in EF from zero bulldoze is probably confounded by the correlation between economy and income.
Zero EF nations seeing no change in EF from non-zero-bulldoze options will be because they're already spending none of their GDP on EF, so a higher income makes no difference.

[Eldritch-0027]

Unfortunately, the sparse datapoint coverage of the above plots will never be improved. It seems the NationStates admins have decided, despite our clear explanation to the contrary, that our data collection was being performed by means of an "illegal script". We were sent this accusation about a week ago, and I explained then that the issue data is collected via a modified browser setup. The browser automatically logs into a nation, randomly selects an issue, and randomly selects an answer-option, collects current nation stat data (it can either be cached from before or read anew) and then tries to match the provided text data of the issue to the text of the known issue answer-options in a database. A user - person manning this data-collection browser - is responsible for observing if the match of the text is correct, and then if it is, they click on a button that submits the answer-option to the website. The browser then collect new current nation stat data (either using the provided stat deltas or reading the current state vector again). Thus, clearly, the issue is answered with human interaction. This is required by the rules, and we were obeying this rule. Another rule is that actions can not be submitted too quickly to the website. The modified web browser was set up to ensure this rule was obeyed also. Thus even if the user were able to click the button more rapidly, the browser would only interact at a maximum rate - it's operation was rate-limited. Thus this rule was also obeyed. All of the above was clearly explained to the moderator(s) who accused our collaboration of illegal actively. I have asked them to either show the above statements are untrue and that we violated a rule, or to show that there was a different rule of which we were unaware, and to show we violated that rule. They have done neither of these. They instead deleted most of the nations used by this collaboration. For example, of the 800+ puppets we used for data collection, only this one remains. It seems that some members of our collaboration were interested by the trading cards that answering issues generated, and had begun to collect them. Since the issue answering was done within the rules, this should not have presented a problem. However, it seems that the NationStates administrators have ignored that, as I explained above, the issue-answering was done legally, and simply stated (without proof or counter-evidence) that our data collection was performed illegally, and used this as a justification to delete most of our nations. They also asserted that the intent of our activity was to create these trading cards. Though this creation of trading cards was a side effect that some of our collaboration members LEGALLY enjoyed, it was not the primary purpose of our work. Given that we can - as far as I can tell - obey all the rules and still have our nations deleted and our data collection halted on an apparent ill-founded whim, I personally have no interest in continuing with this project. Others in the NIP collaboration might - I don't know at this time.

Though no new datapoints will be collected, it would still be possible to create more vector plots with the existing data, or to share our raw data with other people. So, anyone wants either, please post a request in this thread - I will ask someone in our collaboration to check it regularly, atleast for a while.