Content on Twitter’s home timeline is selected and ordered by personalization algorithms. By consistently ranking certain content higher, these algorithms may amplify some messages while reducing the visibility of others. There’s been intense public and scholarly debate about the possibility that some political groups benefit more from algorithmic amplification than others. We provide quantitative evidence from a long-running, massive-scale randomized experiment on the Twitter platform that committed a randomized control group including nearly 2 million daily active accounts to a reverse-chronological content feed free of algorithmic personalization. We present two sets of findings. First, we studied tweets by elected legislators from major political parties in seven countries. Our results reveal a remarkably consistent trend: In six out of seven countries studied, the mainstream political right enjoys higher algorithmic amplification than the mainstream political left. Consistent with this overall trend, our second set of findings studying the US media landscape revealed that algorithmic amplification favors right-leaning news sources. We further looked at whether algorithms amplify far-left and far-right political groups more than moderate ones; contrary to prevailing public belief, we did not find evidence to support this hypothesis. We hope our findings will contribute to an evidence-based debate on the role personalization algorithms play in shaping political content consumption.

Two n-type semiconducting polymers with alternating arylene (thiophene or selenophene)−tetraazabenzodifluoranthene diimide (BFI) donor−acceptor architecture have been investigated as new electron acceptors in polymer/ polymer blend solar cells. The new selenophene-linked polymer, PBFI-S, has a significantly smaller optical band gap (1.13 eV) than the thiophene-linked PBFI-T (1.38 eV); however, both polymers have similar HOMO/LUMO energy levels determined from cyclic voltammetry. Blends of PBFI-T with the thiazolothiazole−dithienylsilole donor polymer (PSEHTT) gave a 2.60% power conversion efficiency (PCE) with a 7.34 mA/cm 2 short-circuit current. In contrast, PBFI-S:PSEHTT blends had a 0.75% PCE with similarly reduced photocurrent and external quantum efficiency. Reduced free energy for charge transfer and reduced bulk electron mobility in PBFI-S:PSEHTT blends compared to PBFI-T:PSEHTT blends as well as significant differences in bulk film morphology are among the reasons for the large loss in efficiency in PBFI-S:PSEHTT blend solar cells.

Industrial recommender systems are frequently tasked with approximating probabilities for multiple, often closely related, user actions. For example, predicting if a user will click on an advertisement and if they will then purchase the advertised product. The conceptual similarity between these tasks has promoted the use of multi-task learning: a class of algorithms that aim to bring positive inductive transfer from related tasks. Here, we empirically evaluate multi-task learning approaches with neural networks for an online advertising task. Specifically, we consider approximating the probability of post-click conversion events (installs) (CVR) for mobile app advertising on a large-scale advertising platform, using the related click events (CTR) as an auxiliary task. We use an ablation approach to systematically study recent approaches that incorporate both multitask learning and "entire space modeling" which train the CVR on all logged examples rather than learning a conditional likelihood of conversion given clicked. Based on these results we show that several different approaches result in similar levels of positive transfer from the data-abundant CTR task to the CVR task and offer some insight into how the multi-task design choices address the two primary problems affecting the CVR task: data sparsity and data bias. Our findings add to the growing body of evidence suggesting that standard multi-task learning is a sensible approach to modelling related events in real-world large-scale applications and suggest the specific multitask approach can be guided by ease of implementation in an existing system. CCS Concepts: • Computing methodologies → Multi-task learning; Neural networks; Batch learning.

Type 1 diabetes is an autoimmune disease in which insulin-secreting β-cells are destroyed, leading to a life-long dependency on exogenous insulin. There are no approved disease-modifying therapies available, and future immunotherapies would need to avoid generalized immune suppression. We developed a novel plasmid expressing preproinsulin2 and a combination of immune-modulatory cytokines (transforming growth factor-beta-1, interleukin [IL] 10 and IL-2) capable of near-complete prevention of autoimmune diabetes in non-obese diabetic mice. Efficacy depended on preproinsulin2, suggesting antigen-specific tolerization, and on the cytokine combination encoded. Diabetes suppression was achieved following either intramuscular or subcutaneous injections. Intramuscular plasmid treatment promoted increased peripheral levels of endogenous IL-10 and modulated myeloid cell types without inducing global immunosuppression. To prepare for first-in-human studies, the plasmid was modified to allow for selection without the use of antibiotic resistance; this modification had no impact on efficacy. This pre-clinical study demonstrates that this multi-component, plasmid-based antigen-specific immunotherapy holds potential for inducing self-tolerance in persons at risk of developing type 1 diabetes. Importantly, the study also informs on relevant cytokine and immune cell biomarkers that may facilitate clinical trials. This therapy is currently being tested for safety and tolerability in a phase 1 trial (ClinicalTrials.gov Identifier: NCT04279613).

No abstract