The collection of data for this project ended some time ago. The potential use of the data is expansive, my time is quite limited until my current project is completed. I have a few notes about things that we might want to consider:

Sockpuppets. There has been a great deal of traffic in claims of sockpuppet deployment recently. Yes, there are bots. What we need to think about is how bots figure into our existing models. Comment sections largely went extinct when editors of sites realized that their own comment sections undermined their arguments. It will be important to understand the centrality of bots to the network to see if they really have any impact.

Do bots matter in a transmit or event centered eco-system? During media events social networks tend to focus on recirculating information and most nodes tend to go dormant. This is an interesting phenomena that I have documented extensively in this project. In any social media event network, most nodes have eigenvector centralities of zero.

Do swarms of unimportant bots have an ambient agenda setting effect? Meraz found that the legacy media tended to maintain agenda setting influence via the blogosphere. Without getting too much into the literature it is clear that the current consensus position is that agenda setting functions of elites simply transitioned onto the internet. Given the utility of this argument in many other forums, it would only make sense that it is true here too. Much like new media is old, or old is new, this is a fairly safe argument to be making. After all, break, rupture, and discontinuity are are delta to continuity. Media theorists love to endlessly chew through this debate in the context of the “digital.” My concern with these debates comes on the level of practice. Debunking claims to novelty can be an important move, I am not sure what else that debate gets for scholars. Ontological critique will never be really complete particularly when there is an aporia involved. As a dialectical engine these debates can power interesting reflection that might inform semantics, pragmatics, and alterity but should not be allowed to supplant them.

Propaganda is bad. I am not sure if it is bad because it is false or even because it is, but because the public culture into which propaganda is diffused is not able to cleave truth from falsity or I believe more accurately, is entirely willing to support falsity as a weapon.

What could I possibly mean by that? People are not stupid. In a class on code and small screen design we were discussing a study of Instagram which connected IG with depression. The mechanism? The confusion that sets in when people see the highly curated feeds. Apparently they can’t distinguish this from reality. This is an old canard in media studies and cultural critique in general. Just as novels would make it impossible for an eighteenth century mother to respond to her child (due to confusion) the media now does the same thing. Of course every student already knew the difference between social network profile life and regular physical reality. Watch Doug Rushkoff’s Digital Nation. Advertising agencies have nothing on these kids. The buzzword in social media research a decade ago was context collapse, the potential victims: children. Now the average college frosh is incredibly sophisticated at parsing contexts. Older people, the adults who once would have protected them from such misunderstanding, now the victims. This meme builds out the argument:

Your parents in 1996: Don’t trust ANYONE on the Internet. Your parents in 2016: Freedom Eagle dot Facebook says Hillary invented AIDS.

Where the rubber meets the road is the understanding of ideology. Ideology in this formulation is a form of enlightened self-consciousness. This formulation has been well developed by Sloterdjik, Zizek, and many others. Regardless of what you might think of this research trajectory, it is a really helpful argument.

It seems possible then that the bulk of users might have been communicating in conjunction with the bots. Instead of deceiving users into believing something false, the agenda setting power of bots might be unified with a public interested in distributing propaganda. A web of publics interested in circulating and recirculating false content for strategic purposes is much more powerful than a robot. Too many activists believe that they are strategic and the publics they attempt to persuade are not. They are all strategic. Robots may just not be that important.

As this election winds to a close the locations of the final stops are predictable, Ohio, Pennsylvania, Florida, and a smattering of Missouri and North Carolina.

In the late sixth party system, elections are contested on a very predictable axis that runs somewhere between Columbus and Orlando. Cover Pittsburgh for a little drama, perhaps Raleigh, and call it a campaign. This will not due in the future. At the opening of the mid-sixth Ohio and Florida were equal prizes. Now Florida dwarfs Ohio. After the next reapportionment, Florida will nearly double Ohio in size. Of course this doesn’t matter, right? They are still the same size. Florida is getting more liberal, Ohio is more conservative, also the rest of the country exists.

The base of the early sixth party system supposed that a candidate could align the south and the west. For Nixon and Regan, this included California. Aside from Carter in 76, who dealigned the South, the base map made it difficult if not impossible to elect a democratic President. Clinton was, in many ways, Carter 2.

California, Oregon, and Washington slipped out of alignment with the mountain west and Texas while substantial the rest of the country became more consistent. A southern Democrat could pick off enough votes to win with a base of the upper midwest, pacific, and north east plus a few more.

This was a reporters dream. A stable, consistent map with a tie almost built in and the key swing states a quick flight from New York or Washington. After two close Bush elections and two not so close Obama elections the map has changed. Safe Republican states like Colorado, New Mexico, Nevada, and Virginia have shifted alignment. Arizona is a battle ground, the great city of Atlanta suddenly matters. The basic alignment of the South and the West is breaking down and we simply don’t know how to report it.

The 2020 reapportionment won’t take effect until the 2024 Presidential election. This won’t be a dramatic shift, just a few seats moving in either direction. Density models suggest that this will make big states bigger and continue the trend of flattening the rest of the country into a Presidential desert. Likely losers, Ohio, Pennsylvania, Michigan, Illinois, West Virginia, Rhode Island, Alabama, and possibly New York and Minnesota. Gains likely in Texas, Florida, North Carolina, Virginia, Colorado, California, Arizona and Oregon.

For the most part this shift does little to the overall dynamics of the election. What it does is amplify the importance of states shifting party alignment. A swing campaign in 12 vote Arizona is very different than a 17 vote Ohio. A safe, Democratic, 14 vote Virginia transforms the map, as does a Washington-Oregon axis that finally outweighs Pennsylvania.

The implications?

Coverage will be shifting west. The national media will need to learn to cover Arizona. The campaigns, at least those paying attention, will spend less time in Ohio, moving toward North Carolina, Atlanta, and Denver. The issues covered will be shifting, people in the Southwest have different environmental needs than those in the cold Rust Belt. Replace that axis in the eastern timezone with a line between PHX and ATL and you see America’s future.

Clinton is now passing 160k hashtag uses per day, Trump has recovered to the mid-20k region. This could of course be an indicator of dislike for Clinton, but what it does indicate is that Clinton is the focal point as we arrive at the final calculation.

In the larger scheme of things, if Clinton wins the Romney curse (more twitter activity being negative) will be broken.

Yesterday, Silver’s basic model in Florida went haywire. Clinton had gained in the previous four polls and his probability of Clinton win decreased. Poll Delta was positive for Clinton with negative Model Alpha (coefficient of victory). This means that other factors were outweighing new polling information. These could legitimately include lagged weighting, although any world where positive polls for one candidate could decrease their probability is quite rare. When combined with earlier misreads on the election and process, such as arguing that positive polls for Trump were a bad sign for his candidacy people started to question the model. My best guess would be that this was some combination of lag-weighting and salting state polls with national data. Low Clinton delta combined with an artificial salt score and a lag weight for some older state polls could explain it.

By morning his models had reset to track closer with actual polling data. Clinton appears to be well positioned to win the election. Sam Wang, a political scientist and aggregator, has been clear that this is not a particularly turbulent election. The polls have been remarkably consistent as has his model.

Here are a few important take aways:

A. Aggregated poll methods were an improvement over the simplistic horse-race, especially when those polls are matched to the actual unit of measurement, state votes.

B. Good modeling methods are not dramatic. No. The model won’t wildly swing based on a new poll. If 100 polls show Timmy winning, one poll with Billy winning won’t switch it back to 50/50.

C. This process is out of phase with the news cycle. Once the news cycle starts driving model runs, the model likely will be pushed to the breaking point.

Poll aggregation, like many predictive methods, have instead become the fodder for news the day of the event. This is not optimal. The domain of near future news is interesting, but should be restricted to the best methods. Sometimes there is no news. That is perfectly alright. The Presidential election algorithm is pretty straight forward. Basically, this election has not been particularly volatile, so why run the model after every poll?

Another Note about Policy Debate

One of the factors that keeps my peer group returning to Silver is his history as a high school debater. I debated in the same era and back then a popular argument was called the politics disadvantage. You would argue that the work of Clinton to pass your bill would preclude the passage of some other bill, like the African Growth and Opportunity Act, Debt Relief, KEDO, PNTR with China, or the CTBT. These are all important weighty matters. To get to this point one needs to win a few specific arguments: uniqueness – X will pass now, link – your plan would require substantial capital to pass, internal link – decreased capital means no passage of the other issue, impact – X is really good.

Of course one could argue defensively that the disadvantage is silly for any number of reasons, offensively one can turn the initial link (the plan increases capital/is popular) the impact (CTBT is bad) or the internal link by answering the theory of politics posed by the disadvantage. Some teams carried entire files devoted to responding exclusively on the internal link level. So you would say, political capital (or losers-lose) they could then read cards about Presidential losses causing policy wins (losers-win) or if the disadvantage was reversed they would read (winners-lose) or (winners-win) to argue that the act of spending capital makes more capital. Teams could plug and play the different internal link scenarios. A team with great internal link blocks could mange their research burden and exposure to risk by debating a known quantity.

Unlike an affirmative response to a politics disadvantage, there are not plug and play political theories that can make polls go backwards.

As we reach the close of the data gathering phase of this research project, day t-7 saw the Clinton-Trump primary activity levels reach an all time high for division at Clinton x70. Historically, high levels of conversation such as this are bad for a candidate, although in 2015 they were very good for Trump. Something to consider, I am juxtaposing #ImWithHer to #Trump2016.

What do I think is happening? Without social network analysis and other approaches I don’t have a definitive answer. My guess is that the comparative evaluation phase of the process is over and that people are taking one, last, hard look at Clinton. Since this point, Trump activity has been rising and Clinton activity may be starting to normalize.

In terms of timing and election modeling, what does this mean? First, early voting is well underway in many states, so connecting anything related to polls with exit polls with actual results is fishy at this point. Second, For horse-race style analysis, this makes the election much less fun as exit polls are better than normal polls. Oregon for example is largely finished voting, if I vote at this point I will need to take my ballot to an official drop site. Third, even without running Cohen’s D on the daily twitter returns, I can tell you that Comey’s letter had a dramatic effect on Twitter traffic. Trump activity decreased to campaign record lows while Clinton surged to campaign highs.

Overall, I am reminded of the problems with producing social facts. Alain Desrosières The Politics of Large Numbers (a must read for computational critical/cultural studies) calls for a consideration of the categories and processes in statistical analysis. In social network analysis we tend to consider the census of a data set rather than a sample. It is nice to know that we have an inflection point to study and that we can understand the entire dynamic system around it, but it is discomforting to know that it can be enacted so quickly.

Today is 59 days before the election. Why does this matter? Section 315 of the communication act provides that if an official candidate the lowest rate normally paid more than sixty days before the election. From here on out advertising on television will be very expensive.

View at Medium.com

I happened to see some CNN this morning, they were reporting on the upcoming presidential election as a dead heat with Trump appearing more trust-worthy than Clinton. This was somewhat shocking to me as the state level polls are fairly consistent. Iowa is looking particularly odd, although Florida, Virginia, Pennsylvania, and Colorado are looking like a firewall. Why would they report this way?

First, they need energy and excitement. I am not a journalist, I study the social networking industry and do social network analysis. My publication rhythm is really slow. Regular blogging is the extent of what I try to write quickly. Unlike a newspaper reporter or worse a cable television journalist, I have the benefit of time. If a week passes with no interesting polls or media events or twitter feuds, I can avoid writing. Cable news can’t go so much as three hours without a new story to report on. Explanation of poor reporting one: the timing of the legacy media requires the production of substantial volumes of inferior quality work.

Second, there is bias on the part of news organizations. This is inevitable. Generally, this does not mean axe grinding. For most folks, trying to do a good job, their bias is expressed through the selection of frames that are available for writing their huge volumes of stories. For Clinton, this means trying to write a 1990s style ‘scandal’ story. For Trump, this means writing about his narcissism, which he might actually enjoy. Both of these frames are then dropped into the horse-race meta-frame and you have a strange, stilted way of covering the election. Just to be clear, there isn’t a truly objective way to cover any of this, and we shouldn’t pretend that there is. At the same time, this does not mean that there are not better and worse ways to cover the election.

Third, the frames that are available for journalists are designed with a false balance in mind. The generic frame that you would fill out to compose a low-quality election story won’t have one side clearly winning. Strong positions don’t make a mild, low-risk story. False balance is so appealing because it should mitigate risk. If none of the stories really say anything, no one will be that upset, and if they are upset it is because they are grinding a partisan axe. The editor gets plausible deniability. The public gets nothing. Harry Frankfurt’s On Bullshit nailed this sort of reason: a liar is morally superior to a bullshitter (someone who simply talks without regards for the truth or reality) as the liar at least knows and holds a regard for the truth.

In general, the future is not bright for cable news. As the carriage transition begins, folks might actually get to decide if they want to pay for CNN. My guess is that they will chose not to. A few years ago the idea of the Weather Channel dropping to the mere value of a toss-in for a sweeter deal for weather tech. At least FNC and MSNBC have the clarity to make clear and consistent arguments. Viewers reward this by tuning in. CNN might be avoiding all the risk of taking a stand, but they will lose in the long run.

I have written a number of election related posts recently on the site Medium.

Home-brewed hot takes is a short piece with advice about how to cook up your own political hot takes at home. Often, you can do a better job than the pundits by thinking carefully about the algorithm of the election. This handy website (270towin) will let you check your projections. If someone is selling a story about New Hampshire tipping the election, you can check the rest of their map assumptions.

Here is some map-work on Romney’s attempt to derail trump. It was a pretty clear case of too-little, too-late.

I have written a number of other stories using my Medium account about other topics, particularly the economics of social networks. If you are interested in such things, check it out.

Just posted at Medium.