Monthly Archives: October 2017

Phil 10.31.17

7:00 – 4:30 ASRC MKT

    • Wrote up notes from yesterday’s meeting
    • Look for JCMC requirements
    • Change the rest of the “we” to “I” in the DC, then submit. Done, did a spell check because I had forgotten to integrate a spell checker!
    • Saw this today on the Google Research Blog: Closing the Simulation-to-Reality Gap for Deep Robotic Learning. In it they show how simulation can be used to improve deep learning because of the vast increase in conditions that can be simulated rather than found or built in the real world. The reason that it’s important in my work is that the simulation can feed and support the training of the classifiers once the simulation becomes sufficiently realistic.
    • Because I can’t stop reading horrible things, ordered Totalitarianism, Terrorism and Supreme Values: History and Theory, by  Peter Bernholz
    • Not the most exciting thing, but yay!
      ID	posted		message					playerID	parentID
      1	1509458541	message 0 of 20 by Abbe, Karleen	5	6	
      2	1509458541	message 1 of 20 by Abbey, Abbi	7	6	
      3	1509458541	message 2 of 20 by Abbey, Abbi, responding to message 1	7	6	2
      4	1509458542	message 3 of 20 by Abbe, Karleen, responding to message 2	5	6	3
      5	1509458542	message 4 of 20 by Abbe, Karleen, responding to message 1	5	6	2
      6	1509458542	message 5 of 20 by Abbe, Karleen, responding to message 4	5	6	5
      7	1509458542	message 6 of 20 by Abbe, Karleen, responding to message 3	5	6	4
      8	1509458542	message 7 of 20 by Abbe, Karleen, responding to message 1	5	6	2
      9	1509458542	message 8 of 20 by Abbe, Karleen, responding to message 1	5	6	2
      10	1509458542	message 9 of 20 by Aaren, Abbie, responding to message 2	3	6	3
      11	1509458542	message 10 of 20 by Abbey, Abbi, responding to message 5	7	6	6
      12	1509458542	message 11 of 20 by Abbe, Karleen, responding to message 10	5	6	11
      13	1509458542	message 12 of 20 by Abbey, Abbi, responding to message 7	7	6	8
      14	1509458542	message 13 of 20 by Aaren, Abbie	3	6	
      15	1509458542	message 14 of 20 by Abbe, Karleen, responding to message 8	5	6	9
      16	1509458542	message 15 of 20 by Abbe, Karleen, responding to message 11	5	6	12
      17	1509458542	message 16 of 20 by Abbe, Karleen	5	6	
      18	1509458542	message 17 of 20 by Abbe, Karleen, responding to message 4	5	6	5
      19	1509458542	message 18 of 20 by Aaren, Abbie, responding to message 14	3	6	15
      20	1509458542	message 19 of 20 by Aaren, Abbie, responding to message 2	3	6	3
    • cleaning up some cases where scenario is set to null. Fixed. It’s the first array index problem. Grrrrr. Ok, broke some things trying to make things better….
    • Then it’s time to make some REST interfaces
    • Meeting with Cindy. Much progress!
      • User-specified scenarios, seeded with some fun topics like conspiracy theories
      • Private deliberations.
      • Esperanto for verdict: verdikto
      • Lobbies for collecting users
      • Game starts when an DM-specified minimum is met, though there may be time to accumulate into a max as well
      • Game ‘dies’ if no contribution (by all players?) in a certain window
      • One user can kill a game by withdrawing. This can be attached to a user (troll), so the player can anonymously block in the future
      • Games can be respawned, optionally without a triggering troll from the last time
      • Games/Scenarios can be cloned
      • Highest-quality games that reach a verdict are featured on the site. Quality could be determined by tagging or NLP+heuristics.


Phil 10.30.17

7:00 – 4:30 ASRC MKT

  • The discussion and conclusion
  • Tweaked the “Future Work” section of the CHIIR DC proposal to reflect the herding work. More words means less bullet points!
  • Updated Java and XAMMP on my home machine
  • Pointed the IDE at the correct places
  • I don’t think I have PhpInspections (EA Extended) installed at work? It does nice things – Have it now
  • Working through creating a strawman game. Having some issues with a one-to-many relationship with RedBeanPHP. Ah – it’s because you have to sync the beans. I think rather than have a game point at all the players, I’ll have the players point at the scenario, and the chat messages point at the game and players.
  • Got that mostly working, but having a null player issues
  • Important PHP issue – arrays don’t need to start at zero! The bean arrays are indexed with respect to their db id!
  • Meeting with Wayne
  • The DC is good to submit
  • Start working on a JCMC article that connects the flocking model to qualitative theory.
  • Keep on working on the game. Possible project for a class/group in either 729 – design and evaluate class (Komlodi) or 728 – Online Communities & Social Media (Branham)

Phil 10.27.17

7:00 – 5:00 ASRC MKT

  • Nicely written paper on GANs:
    • Abstract: We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, we add new layers that model increasingly fine details as training progresses. This both speeds the training up and greatly stabilizes it, allowing us to produce images of unprecedented quality, e.g., CELEBA images at 10242. We also propose a simple way to increase the variation in generated images, and achieve a record inception score of 8.80 in unsupervised CIFAR10. Additionally, we describe several implementation details that are important for discouraging unhealthy competition between the generator and discriminator. Finally,we suggest a new metric for evaluating GAN results, both in terms of image quality and variation. As an additional contribution, we construct a higher-quality version of the CELEBA dataset.
    • With cool video
    • And code
  • Working on adding UI and batch interaction for the adversarial herding
    • Enable/disable switch – Done
    • Field for power – don’t know what the scale should be so no slider yet – Done
    • Set<String, Set<Flockingshape, weight>> If this doesn’t work, make shape comparable by name. Done!
      HashMap<FlockingShape, Double> alignedShapeMap;
      if(flock.size() > 0 && !alignedFlockMap.containsKey(flockName)){
          alignedShapeMap = new HashMap<>();
          alignedFlockMap.put(flockName, alignedShapeMap);
          alignedShapeMap = alignedFlockMap.get(flockName);
    • Do I want to delay the triggering of the herding on a separate timer? Waiting on this.
    • It’s done, and the results are kind of scary. If I set the weight of the herder to 15, I can change the change the flocking behavior of the default to echo chamber.
    • Normal: No Herding
    • Herding weight set to 15, other options the same: HerdingWeight15
  • Did some additional tweaking to see if having highly-weighted herders ignore each other (they would be coordinated through C&C) would have any effect. It doesn’t. There is enough interaction through the regular populations to keep the alignment space reduced.
  • It looks like there is a ‘sick echo chamber’ pattern. If the borders are reflective, and the herding weight + influence radius is great enough, then a wall-hugging pattern will emerge.
    • The influence weight is sort of a credibility score. An agent that has a lot of followers, or says a lot of the things that I agree with has a lot of influence weight The range weight is reach.
    • Since a troll farm or botnet can be regarded as a single organization,  interacting with any one of the agents is really interacting with the root entity.  So a herding agent has high influence and high reach. The high reach explains the border hugging behavior.
    • It’s like there’s someone at the back of the stampede yelling YOUR’E GOING THE RIGHT WAY! KEEP AT IT! And they never go off the cliff because they are a swarm Or, it never goes of the cliff, because it manifests as a swarm.
    • A loud, distributed voice pointing in a bad direction means wall hugging. Note that there is some kind of floating point error that lets wall huggers creep off the edge.Edgecrawling
    • With a respawn border, we get the situation where the overall heading of the flock doesn’t change even as it gets destroyed as it goes over the border. Again, since the herding algorithm is looking at the overall population, it never crosses the border but influences all the respawned agents to head towards the same edge: DirectionPreserving
  • Paper thoughts:
    • Armys have different patterns from emergent groups. They are imposed formations and reflect a commander’s will
    • From a distance, they look different, but close up, they may look the same. One of the reasons for the success of the Roman Legion was the use of formations against the less sophisticated structures of their adversaries [ref]

Phil 10.26.17

7:00 – 3:30 ASRC MKT

  • Listening to BBC Business Daily this morning on Facebook vs Democracy:
    • Presenter Ed Butler hears a range of voices raising concern about the existential threat that social media could pose to democracy, including Ukrainian government official Dmytro Shymkiv, journalist Berit Anderson, tech investor Roger McNamee and internet pioneer Larry Smarr.
  • …and had some thoughts on adversarial herding in information space
    • Herders can teleport, since they are not emotionally invested in their belief space position and orientation
    • Herders appear like multiple individuals that may seem close and trustworthy, but they are actually a distant monolithic entity that is aware of a much larger belief space.
    • Herders amplify the most extreme positions and may also amplify opposition. The insight is that they are not herding in a direction, but to increase polarization
    • To add this to the model, I need to do the following:
      • Make the size of the agent a function of the weight
      • When in ‘herding mode’ the overall heading of the group is calculated, and the agent that is furthest in that direction is selected
      • The weight is increased to X, and the radius is increased to Y.
        • X represents amplification of the concept, by trolls, bots, etc.
        • A large Y means that the bots can swamp other, normally closer signals. This models the effect of a monolithic entity controlling thousands of bots across the belief space
    • Got it! Note that the influence radius is 1/3 of the range normally needed to polarize. Also note that the amplified agent is not leading. This reflects Arendt’s insight that totalitarian rulers follow and amplify the mobTrollfarm I need to make it so that there is UI support (on/off, amount), and make it so that each flock can have its own bots.
    • I expect this to produce extreme polarization (low time to border) more quickly than emergent, ‘organic’, echo chambers
    • This may describe some applicable group hunting behavior: Predator-prey interactions in two schooling fishes, Caranx ignobilis and Stolephorus purpureus
      • Interactions between the jack, Caranx ignobilis, a facultative schooling species, and the Hawaiian anchovy, Stolephorus purpureus, an obligate schooler were studied within an enclosure in the field in Hawaii. Single predators were the most successful at capturing isolated (individual) prey, and relatively unsuccessful at capturing individuals in schools. Grouped (schooled) predators were the most successful at capturing schooled prey. The leading, or first, predator was the most successful member of a group or school at capturing isolated or schooled prey. Following predators tended to make it possible to catch more prey earlier in the experiments. Larger predator groups were able to break up schools of prey quickly, resulting in increased numbers of prey becoming isolated. These prey were captured before they could reform or join other schools. As the initial size of the prey school increased, the per cent of individuals captured declined. Schooling in prey reduces the time a visually orienting predator has to align himself with an individual prey. Schooling in predators may have co-evolved as an adaptation, making it possible for predators to break up and isolate schooled prey. Larger prey schools may have co-evolved to satiate or swamp the feeding capacity of a finite number of schooled predators and decrease the probability that a specific given individual would be captured.
  • Finishing up Suppressing the Search Engine Manipulation Effect

Phil 10.25.17


7:00 – 1:00 ASRC MKT

  • Collective Agency and Cooperation in Natural and Artificial Systems (some chapters:)
    • The Participatory Turn: A Multidimensional Gradual Agency Concept for Human and Non-human Actors
    • Collective Agency and Cooperation in Natural and Artificial Systems
    • Planning for Collective Agency
    • Simulation as Research Method: Modeling Social Interactions in Management Science
    • How Models Fail
    • Requested the library copy. Request #426721
  • Continuing with Suppressing the Search Engine Manipulation Effect
    • One quick thought on Likert scales: If the selection is a slider, then we can see how the user interacts with the slider. This could let us see how decisive the users are. This could also be dome by looking at the area surrounding the radio buttons and tracking mouse motion and number of clicks in the area
    • Introduction
      • Recent research has shown that society’s growing dependence on ranking algorithms leaves our psychological heuristics and vulnerabilities susceptible to their influence on an unprecedented scale and in unexpected ways
      • Experiments conducted on Facebook’s Newsfeed have demonstrated that subtle ranking manipulations can influence the emotional language people use
      • Similarly, experiments on web search have shown that manipulating election-related search engine rankings can shift the voting preferences of undecided voters by 20% or more after a single search
      • While “bias” can be ambiguous, our focus is on the ranking bias recently quantified by Kulshrestha et al. with Twitter rankings
      • Our results provide support for the robustness of SEME and create a foundation for future efforts to mitigate ranking bias. More broadly, our work adds to the growing literature that provides an empirical basis to calls for algorithm accountability and transparency [24, 25, 90, 91] and contributes a quantitative approach that complements the qualitative literature on designing interventions for ranking algorithms
      • Our results also suggest that proactive strategies that prevent ranking bias (e.g., alternating rankings) are more effective than reactive strategies that suppress the effect through design interventions like bias alerts. Given the accumulating evidence, we speculate that SEME may be impacting a wide range of decision-making, not just voting
    • Related Work
      • Order effects are among the strongest and most reliable effects ever discovered in the psychological sciences [29, 88]. These effects favorably affect the recall and evaluation of items at the beginning of a list (primacy) and at the end of a list (recency).
        • There does not seem to be an equivalent primacy effect in maps that I can find
      • online systems can: (1) provide a platform for constant, large-scale, rapid experimentation, (2) tailor their persuasive strategies by mining detailed demographic and behavioral profiles of users [1, 6, 9, 18, 121], and (3) provide users with a sense of control over the system that enhances their susceptibility to influence
        • Is this flocking from the flock’s perspective? Sort of an Ur-flock?
        • This is that Trust/Awareness equation again
      • A recent report involving 33,000 people found that search engines were the most trusted source of news, with 64% of people reporting that they trust search engines, compared to 57% for traditional media, 51% for online media, and 41% for social media [10]. Similarly, a 2012 survey by Pew found that 73% of search engine users report that “all or most of the information they find is accurate and trustworthy,” and 66% report that “search engines are a fair and unbiased source of information” [105].
      • Suggestions for fostering resistance can be broken down into two primary strategies: (1) providing forewarnings [43, 49] and (2) training and motivating people to resist [79, 120].
        • Interesting that alternate, non-ordered design approaches aren’t even mentioned
      • Part of the reason that forewarnings work is explained by psychological reactance theory [12], which posits that when people believe their intellectual freedom is threatened – by exposing an attempt to persuade, for example – they react in the direction opposite that of the intended one
      • In the context of online media bias, researchers have primarily explored methods for curbing the effects of algorithmic filtering and selective exposure [87, 96] rather than ranking bias [71]. In this vein, researchers have developed services that encourage users to explore multiple perspectives [97, 98] and browser extensions that gamify and encourage balanced political news consumption [19, 20, 86]. However, these solutions are somewhat impractical because they require users to adopt new services or exert additional effort.
    • Methods – Experiment Design
      • To construct biased search rankings we asked four independent raters to provide bias ratings of the webpages we collected on an 11-point Likert scale ranging from -5 “favors Cameron” to +5 “favors Miliband”. We then selected the 15 webpages that most strongly favored Cameron and the 15 that most strongly favored Miliband to create three bias groups
      • The query in the search engine was fixed as “UK Politics ‘David Cameron’ OR ‘Ed Miliband’”, and subjects could not reformulate it.
      • On top of assignment to a bias group, subjects were randomly assigned to one of three alert experiments.We drew from the literature on decision-making and design intervention to implement so-called debiasing strategies for improving decision-making in the presence of biased information [39, 78, 82]. Specifically, we constructed and placed alerts in the search results produced by our mock search engine that provided forewarnings with salient graphics, autonomony-supportive language, and details on the persuasive threat
    • Methods – Procedure
      • After providing informed consent and answering basic demographic questions
        • Do this and use this phrase!
      • Subjects then rated the two candidates on 10-point Likert scales with respect to their overall impression of each candidate, how much they trusted each candidate, and how much they liked each candidate. Subjects also indicated their likelihood of voting for one candidate or the other on an 11-point Likert scale where the candidates’ names appeared at opposite ends of the scale and 0 indicated no preference, as well as on a binary choice question where subjects indicated who they would vote for if the election were held today.
        • This is a good way to set up the game. People read the dilemma, formulate an initial solution and their level of commitment to it. They can choose to make it “public” as their first statement or to keep it private and display a “no opinion” initial statement
      • We asked: “While you were doing your online research on the candidates, did you notice anything about the search results that bothered you in any way?” and prompted subjects to explain what had bothered them in a free response format: “If you answered “yes,” please tell us what bothered you.” We did not directly ask subjects whether they had “noticed bias” to avoid the inflation of false positive rates that leading questions can cause
    • Methods – Participants
      • We recruited 3,883 subjects between April 28, 2015 and May 6, 2015 on Amazon’s Mechanical Turk (AMT;, a subject pool frequently used by behavioral, economic, and social science researchers [8, 13, 102]. We excluded from our analysis subjects who reported an English fluency level of 5 or less (on a scale of 1 to 10) (n=26)
        • MTurk would be a good source of participants as well
    • Analysys – Search metrics
      • Utilizing Kolmogorov-Smirnov (K-S) tests of differences in distributions, we found significant differences in the patterns of time spent on the 30 webpages between subjects in the no alert experiment (correlation with ranking: Spearman’s ρ = -0.836, P <0.001) and the high alert experiment (ρ = -0.654, P <0.001) (K-S D = 0.467, P <0.01), and between subjects in the low alert experiment (ρ = -0.719, P <0.001) and the high alert experiment (K-S D = 0.400, P <0.01)
        • A way of looking for explore/exploit populations? And how fast can it be determined? Google uses a mechanism to stop an experiment once a confidence level is reached. Also, bootstrap would be good here
      • Similarly, we also found significant differences in the patterns of clicks that subjects made on the 30 webpages between subjects in the no alert experiment (ρ = -0.865, P <0.001) and the high alert experiment (ρ = -0.795, P <0.001) (K-S D = 0.500, P <0.001), and between subjects in the low alert experiment (ρ = -0.876, P <0.001) and the high alert experiment (K-S D = 0.367, P <0.05)
      • Among all conditions,we found that differences in the patterns of time and clicks on the individual rankings primarily emerged on the first SERP, but less so on the second, fourth, and fifth SERPs
    • Analysys – Attitude Shifts
      • we found that the mean shifts in candidate ratings for the bias groups significantly converged on the mean shift found in the neutral group as the level of detail in the alerts increased, with high alerts creating higher convergence than low alerts
        • As more diverse information is injected, populations compromise
    • Analysys – Vote Shifts
      • Vote Manipulation Power (VMP)is the percent change in the number of subjects, in the two bias groups combined, who indicated that they would vote for the candidate who was favored by their search rankings. That is, if x and x ′ subjects in the bias groups said they would vote for the favored candidate before and after conducting the search, respectively, then VMP = (x ′ − x)/x.
        • This could also be applied to the game to watch how votes for an outcome change over time. In the case of the game, new candidates can come into existence, so we need to watch for that.
    • Analysys – Bias Awareness
      • We found 8.1% of subjects that showed awareness of the bias in the no alert experiment, a figure identical to the 8.1% awareness rate found by Eslami et al. in their audit of [37], and similar to the 8.6% of subjects who showed awareness in the original study [30]. The percentage of subjects showing bias awareness increased to 21.5% in the low alert experiment, and 23.4% in the high alert experiment.
    • Discussion
      • However, despite the additional suppression of the high alerts, the lowest VMP was found among the neutral group subjects: rankings alternating between favoring the two candidates prevented SEME.
        • This configuration forces users to “explore” more, within the context of a list affordance.
      • As with previous research on SEME [30], and with research on attitude change and influence more generally [3, 72, 120], we found that subjects vary in their susceptibility to SEME, as well as in their responsiveness to the alerts, based on their personal characteristics (Figure 6 and Figure 7 in the Appendix).
        • Explorer and exploiter populations?
      • As more people turn to the internet for political news [85, 115], designing systems that can monitor and suppress the effects of algorithm biases, like ranking bias, will play an increasingly important role in protecting the public’s psychological vulnerabilities.
        • And one of the big issues is finding bias at scale with domain independence
      • Real-time automated bias detection could potentially be achieved by utilizing a Natural Language Processing (NLP) approach. One could utilize opinions [75], sentiment [99], linguistic patterns [109], word associations [14], or recursive neural networks [59] with human-coded data to classify biased language.
        • Scale and domain problems.
      • Discussion – Awareness of bias
        • Awareness of ranking bias appears to suppress SEME only when it occurs in conjunction with a bias alert, perhaps because an alert is a kind of warning–inherently negative in nature.
          • According to Moscovici, an inherently negative construct should reduce polarization movement.
        • Awareness of ranking bias in the absence of bias alerts might increase VMP because people perceive the bias as a kind of social proof [111, 112], made all the more powerful because of the disproportionate trust people have in search rankings [10, 95, 105]. The user’s interpretation might be, “This candidate MUST be good, because even the search results say so.”
  • Polarization Game
    • Upgraded to PHP7. I went with the threadsafe version, which meant that I had to upgrade xdebug as well. and for some reason, I had to put the php_xdebug.dll file in the ext directory.
    • And now I have typing!
  • Proposal Work (1:00 – 5:00)
    • Skimmed the RFI, thought of Rick Satava
    • Building out template
    • Wrote the first pass with Aaron

Phil 10.24.17

7:00 – 5:00 ASRC MKT

  • Overview (Research Browser-ish) and their blog. Here it is working with the serendipity corpus: OverviewText
  • Google is doing a lot to map art with ML, but it lacks a sense of meaning
  • Found Visualization of Topic-Sentiment Dynamics in Crowdfunding Projects. Put it in the phase 2 lit review
  • It shouldn’t be called the Polarization Game. Need a title. Maybe something from myth?
  • Continuing with Suppressing the Search Engine Manipulation Effect
    • In the discussion of Order effects, they talk about primacy effects of lists. A quick Scholar search didn’t turn up any studies of primacy effects of maps (like maybe preference for the local area), but some poking around in this space turned up this: Maps of Bounded Rationality: Psychology for Behavioral Economics
  • Back to game design. Having a problem with integrating PHPUnit and RedbeanPHP. Getting this message when I have two or more tests to run: RedBeanPHP\RedException : A database has already been specified for this key.
    • The answer was to add –process-isolation to the TestRunner args:processIsolation
    • BTW, clicking on the blue “info” box at the right of the field will bring up the args allowed
    • I went looking for a way to #ifdef some basic exercising code that I like to add at the bottom of php files. Couldn’t find anything, but I figured out this pattern for file SomeClass.php:
      class SomeClass
          public function doStuff(){echo "stuff done!;}
      if(strpos($_SERVER['PHP_SELF'], "SomeClass.php" )){
          printf("running from %s\n", $_SERVER['PHP_SELF']);
          $sc = new SomeClass();
    • This will only run if the file containing the code is executed directly. It won’t run when called by phpUNIT, for example

4:00 – 4:30

  • Chat with Stan about ML to recognize signal outliers. We talked about GAs for a while, and I sent him Zhenping’s ppt.


Phil 10.23.17

7:00 – 5:00 ASRC MKT

  • Suppressing the Search Engine Manipulation Effect (SEME)
    • Robert Epstein, (American Institute for Behavioral Research and Technology) Epstein and Robertson have found in multiple studies that search rankings that favor a political candidate drive the votes of undecided voters toward that candidate, an effect they call SEME (“seem”), the Search Engine Manipulation Effect.
    • Ronald Robertson (Northeastern University) I design experiments and technologies to explore the ways in which online platforms can influence the attitudes, beliefs, and behavior of individuals and groups. Currently, I am a PhD student in the world’s first Network Science PhD program at Northeastern University and am advised by Christo Wilson and David Lazer.
    • David Lazer (Northeastern University) professor of political science and computer and information science and the co-director of the NULab for Texts, Maps, and Networks
    • Christo Wilson (Northeastern University) Assistant Professor in the College of Computer and Information Science atNortheastern University. I am a member of the Cybersecurity and Privacy Institute and the Director of the BS in Cybersecurity Program in the College.
    • Abstract: A recent series of experiments demonstrated that introducing ranking bias to election-related search engine results can have a strong and undetectable influence on the preferences of undecided voters. This phenomenon, called the Search Engine Manipulation Effect (SEME), exerts influence largely through order effects that are enhanced in a digital context. We present data from three new experiments involving 3,600 subjects in 39 countries in which we replicate SEME and test design interventions for suppressing the effect. In the replication, voting preferences shifted by 39.0%, a number almost identical to the shift found in a previously published experiment (37.1%). Alerting users to the ranking bias reduced the shift to 22.1%, and more detailed alerts reduced it to 13.8%. Users’ browsing behaviors were also significantly altered by the alerts, with more clicks and time going to lower-ranked search results. Although bias alerts were effective in suppressing SEME, we found that SEME could be completely eliminated only by alternating search results – in effect, with an equal-time rule. We propose a browser extension capable of deploying bias alerts in real-time and speculate that SEME might be impacting a wide range of decision-making, not just voting, in which case search engines might need to be strictly regulated.
  • The search engine manipulation effect (SEME) and its possible impact on the outcomes of elections
  • Quantifying search bias: Investigating sources of bias for political searches in social media
    • From the Abstract: It is important to distinguish between the bias that arises from the data that serves as the input to the ranking system and the bias that arises from the ranking system itself. In this paper, we propose a framework to quantify these distinct biases and apply this framework to politics-related queries on Twitter.
  • Making Sense of Conflicting Science Information: Exploring Bias in the Search Engine Result Page
    • Abstract: Currently, there is widespread media coverage about the problems with ‘fake news’ that appears in social media, but the effects of biased information that appears in search engine results is also increasing. The authors argue that the search engine results page (SERP) exposes three important types of bias: source bias, algorithmic bias, and cognitive bias. To explore the relationship between these three types of bias, we conducted a mixed methods study with sixty participants (plus fourteen in a pilot to make a total of seventy-four participants). Within a library setting, participants were provided with mock search engine pages that presented order-controlled sources on a science controversy. Participants were then asked to rank the sources’ usefulness and then summarize the controversy. We found that participants ranked the usefulness of sources depending on its presentation within a SERP. In turn, this also influenced how the participants summarized the topic. We attribute the differences in the participants’ writings to the cognitive biases that affect a user’s judgment when selecting sources on a SERP. We identify four main cognitive biases that a SERP can evoke in students: Priming, Anchoring, Framing, and the Availability Heuristic. While policing information quality is a quixotic task, changes can be made to both SERPs and a user’s decision-making when selecting sources. As bias emerges both on the system side and the user side of search, we suggest a two-fold solution is required to address these challenges.
  • The Network Structure of Exploration and Exploitation
    • David Lazer (Northeastern University)
    • Abstract: Whether as team members brainstorming or cultures experimenting with new technologies, problem solvers communicate and share ideas. This paper examines how the structure of communication networks among actors can affect system-level performance. We present an agent-based computer simulation model of information sharing in which the less successful emulate the more successful. Results suggest that when agents are dealing with a complex problem, the more efficient the network at disseminating information, the better the short-run but the lower the long-run performance of the system. The dynamic underlying this result is that an inefficient network maintains diversity in the system and is thus better for exploration than an efficient network, supporting a more thorough search for solutions in the long run. For intermediate time frames, there is an inverted-U relationship between connectedness and performance, in which both poorly and well-connected systems perform badly, and moderately connected systems perform best. This curvilinear relationship between connectivity and group performance can be seen in several diverse instances of organizational and social behavior.
  • Polarization Game
  • Fika – Not an official one, so Wanajanat, Julie, May(?) and I went over CM and LMN

Phil 10.20.17

7:00 – 4:30 ASRC MKT

  • Asked Wayne yesterday if I could have a desk in the HCC lab, since I’m going to Columbia for no good reason these days.
    • In support of this, got my laptop updated with code bases, PHP, database, etc.
  • Working on getting up the motivation to start combining the db and code pieces. So as a way of avoiding this, tweaked the CHIIR DC submission by using part of my abstract from the HCIC poster and using that browser mockup instead of the one that I put together the other day. I’m conflicted in that it has a word cloud, but I could argue that a word cloud is a reasonable boundary object for someone glancing through the papers and looking at the pictures. I also fixed the populations chart so that the terms line up with the simulation screenshots. populations
  • The RedBeanPHP file didn’t make it into subversion, so I downloaded it at home, verified that everything runs and committed it in the right place.
  • Ok, I think the plan will be to write a test harness that produces a threaded discussion with multiple users and votes for a solution that writes into the DB and is then able to retrieve the thread. To do that I’m going to need one-to-many and many-to-one, so I need to read more of the RedBeanPHP docs
    • Note that the name of the list has to match the type of beans it contains. So, the ‘ownProductList’ contains beans of type ‘product‘, a pageList contains pages, an ‘ownCarList’ contains ‘cars‘ and so on. This convention is used to create the database mapping, in case of the shop, every product record will get a ‘shop_id’field.
    • Ok, I think this is what I want to do:
       * Created by IntelliJ IDEA.
       * Date: 10/20/2017
       * Time: 3:10 PM
      require_once 'libs/rb.php';
      R::setup( 'mysql:host=localhost;dbname=polarizationgameone', 'root', 'postgres' );
      // create or get the scenarios
          echo "CREATE_SCENARIOS defined\n";
          echo "CREATE_SCENARIOS not defined\n";
      // create or get a random number of users between $minPlayers and $maxPlayers
      $minPlayers = 3;
      $maxPlayers = 5;
      $numPlayers = rand($minPlayers, $maxPlayers);
          echo "CREATE_PLAYERS defined\n";
          echo "CREATE_PLAYERS not defined\n";
      // create a game using the scenarios and the players
      // for $numTurns, randomly choose a player to add a chat message
      $minTurns = 10;
      $maxTurns = 20;
      $numTurns = rand($minTurns, $maxTurns);
      for ($t = 0; $t < $numTurns; $t++){
          // connect to a previous statement or set thread_parent to zero
          // write the statment
          // maybe vote for a message?. At the end, some games need to have agreement.
          // a player changing votes pulls their vote from any previous message.
          // produce a list of posts with votes. It'll be something between 0 and $numPlayers
          // if a post has $numPlayers votes, its won. Allocate points to all players and some more to the 
          //  player with the winning post
          // print the status of this turn
          echo "$t\n";
      // reconstruct the game from the database to compare and validate

Phil 10.19.17

7:00 – 2:00 ASRC MKT

  • Read this in Understanding Ignorance last night in the section The Ethics of Belief: Beliefs are factive; they aspire to truth. It would he absurd, as the British philosopher Moore observed, to say “It is raining, but I don’t believe it is raining.” To believe is to take to be true. Beliefs may be false, however, and they may be false without being morally wrong. Yet there are beliefs we judge to be morally wrong.
  • Thinking about the relationship of trust and awareness. Assume there is some function f() that maps trust and awareness to the same scale of [0.0, 1.0] (LaTex):
    • Where behavior is near 1.0, behavior is healthy, since awareness and trust are aligned. You can have low awareness and low trust or high awareness and high trust.
    • As trust approaches zero, there is a tendency towards skepticism. As trust approaches 1.0, there is a tendency towards gullibility.
    • As awareness approaches 1.0, there is also a tendency towards skepticism. But if awareness nears zero while trust is any larger value, behavior approaches infinity, which is the stampede state.
    • I think I can incorporate this into the polarization model
  • Set up MySQL and the polarizationgameone database. Ooops, the view is slightly out of date. Need to commit the last changes!
  • Starting with RedBeanPHP tutorials.
  • Pretty cool. This creates an entry in the db:
    R::setup( 'mysql:host=localhost;dbname=polarizationgameone', 'xxx', 'yyy' );
    $book = R::dispense( 'book' );
    $book->title = 'Learn to Program2';
    $book->rating = 20;
    $book->price = 29.99;
    $id = R::store($book);
  • This finds and prints:
    $books = R::find( 'book' );
    foreach ($books as $b){
        printf("book[%d] title = %s, price = %.2f, rating = %d\n",
            $b->id, $b['title'], $b['price'], $b['rating']);
  • find() has a second argument that is SQL, so you can do things like $book  = R::find( ‘book’, ‘ rating > 4 ‘). There is a third argument for bound values, e.g. $books = R::find( ‘book’, ‘ title LIKE ? ‘, [ ‘Learn to%’ ] ).
  • find works with views as well, but there has to be an integer ‘id’ field

2:00 – 4:30 IRAD? Working on A2P material

Phil 10.18.17

7:00 – 3:30 ASRC MKT

  • Gotta renew the IRB. For some reason I’m not using my school account for this…
    • Done! Good through 17-Oct-2022
  • Looking for a nice way to have PHP interface with the database using object relational mapping (ORM). Looking at RedBeanPHP and FatFreeFramework. I think I’m going to start with RedBean. It looks like something I might have written.
  • Spent some time with Aaron walking through what’s important in getting into the UMBC HCC MS program
  • Found this conference that might be good to go to, though it sounds intimidating as hell
    • 6th International Conference on Learning Representations
    • April 30 – May 3, 2018
    • Vancouver Convention Center, Vancouver, BC, Canada
    • The performance of machine learning methods is heavily dependent on the choice of data representation (or features) on which they are applied. The rapidly developing field of deep learning is concerned with questions surrounding how we can best learn meaningful and useful representations of data. We take a broad view of the field and include topics such as feature learning, metric learning, compositional modeling, structured prediction, reinforcement learning, and issues regarding large-scale learning and non-convex optimization. The range of domains to which these techniques apply is also very broad, from vision to speech recognition, text understanding, gaming, music, etc.

Phil 10.17.17

6:30 – 11:30 ASRC MKT

  • Google’s AI can create better machine-learning code than the researchers who made it, based on this from Google’s research blogUsing Machine Learning to Explore Neural Network Architecture
  • Working on review – done!
  • What examiners do: what thesis students should
    know The advice for thesis students is: first, treat your examiners as friends who want you to pass, and write calmly without agonizing about getting it perfect. Aim to make your thesis reader-friendly, and do a thorough proofread to remove distracting errors. Identify the field(s) you will contribute to, and make your thesis interesting and convincing for examiners from this field. Write a draft, get feedback and use this to improve your thesis. Help your examiners to follow your train of thought: explain what you are doing and why, especially if your thesis differs from what they would expect. Convince your examiners that you have a sound interpretation of the literature, an important topic and an appropriate method, and that your conclusions make a significant, publishable contribution to your field. When you have submitted, expect lots of examiner comments, most of which can help you improve.

    • Thesis examiners tend to:
      (1) be broadly consistent
      (2) expect a thesis to pass
      (3) judge a thesis by the end of the first or second chapter
      (4) read a thesis as an academic reader and as a normal reader
      (5) be irritated and distracted by presentation errors
      (6) favor a coherent thesis
      (7) favor a thesis that engages with the literature
      (8) favour a thesis with a convincing approach
      (9) favour a thesis that engages with the findings
      (10) require a thesis to be publishable
      (11) give summative and formative feedback

Phil 10.16.17

6:30 – 5:00 ASRC MKT

  • Stochastic Modeling And Analytics In Healthcare Delivery SystemsThis book focuses on the research and best practices in healthcare engineering and technology assessment. With contributions from researchers in the fields of healthcare system stochastic modeling, simulation, optimization and management
  • I just realized that the Research Browser is Augmented Data Discovery, and is about to 2-3 years out from peak hype. Something to think about while writing proposals and pitching to management. 
  • Starting to look through Risk Taking to see if I can find scenarios
    • Add fields for user
      • Age – done
      • Gender – done
      • Game name (change by game so that a small number of players don’t recognize each other easily over repeated games
    • Something to think about is whether one scenario would be to create a scenario which is then used in the game.
    • Found it! Appendix E! Aaaaaaaaaaaand Adobe Acrobat is busted so I can’t scan it. Reinstalling. And it’s better, but still busted. Scanned in Photoshop, which works just fine
  • Tweaking CorpusManger so that the TF-IDF output has counts rather than the floating point value to see if that produces better results in search. This is from the serendipity corpus:
    • TF-IDF raw: information serendipity system encounter discovery serendipitous visualization 
    • TF-IDF normalized: information encounter serendipity system datum discovery serendipitous
    • BOW raw: information serendipity system encounter serendipitous discovery result
    • BOW normalized: visualization computer information search system serendipity encounter
    • That works much better with the wordrank algorithm. Keeping it.
  • Fika, talked a little about CM and LMN. Wajanat and Julie are interested. Maybe as a way of quantitatively ranking the centrality of concepts and people in a qualitative study.

Phil 10.15.17

orco Mutagenesis Causes Loss of Antennal Lobe Glomeruli and Impaired Social Behavior in Ants

  • An example of how group behavior patterns reveal a mediated communication problem – Life inside ant colonies is orchestrated with diverse pheromones, but it is not clear how ants perceive these social signals. It has been proposed that pheromone perception in ants evolved via expansions in the numbers of odorant receptors (ORs) and antennal lobe glomeruli. Here, we generate the first mutant lines in the clonal raider ant, Ooceraea biroi, by disrupting orco, a gene required for the function of all ORs. We find that orco mutants exhibit severe deficiencies in social behavior and fitness, suggesting they are unable to perceive pheromones. Surprisingly, unlike in Drosophila melanogaster, orco mutant ants also lack most of the ∼500 antennal lobe glomeruli found in wild-type ants. These results illustrate that ORs are essential for ant social organization and raise the possibility that, similar to mammals, receptor function is required for the development and/or maintenance of the highly complex olfactory processing areas in the ant brain
  • orco Mutagenesis
  • Some random thoughts while riding.
    • The difference between a ‘large group’ and a ‘small group’ is the threshold at which multiple incremental interactions can happen between all members.
      • A group that has a fully connected trust network is fundamentally different from a group that doesn’t. A ‘large group’ requires transitive trust.
        • From HBR: It’s not just biases inside our heads that skew our judgment. We often rely on trusted third parties to verify the character or reliability of other people. These third parties, in effect, help us “roll over” our positive expectations from one known and trusted party to another who is less known and trusted. In such situations, trust becomes, quite literally, transitive. Unfortunately, as the Bernie Madoff case illustrates, transitive trust can lull people into a false sense of security. The evidence suggests that Madoff was a master at cultivating and exploiting social connections. One of his hunting grounds was the Orthodox Jewish community, a tight-knit social group.
      • This can be affected by communications technology in many ways, which the above study points to.
      • Again, the relationship between awareness and trust becomes an issue.

Phil 10.13.17

7:15 – 8:15, 1:00 – 4:00 ASRC MKT 9:00 – 1:00 IRAD

Phil 10.12.17

ASRC MKT 7:00 – 8:30, 2:30 – 3:00, IRAD 8:30 – 2:30