Error correction mechanisms in social networks can reduce accuracy and encourage innovation

Abstract

a b s t r a c tHumans make mistakes but diffusion through social networks is typically modeled as though they do not. We find in an experiment that high entropy message formats (text messaging pidgin) are more prone to error than lower entropy formats (standard English). We also find that efforts to correct mistakes are effective, but generate more mutant forms of the contagion than would result from a lack of correction. This indicates that the ability of messages to cross "small-world" human social networks may be overestimated and that failed error corrections create new versions of a contagion that diffuse in competition with the original.

Cultural Sentiments and Schema-Consistency Bias in Information Transmission

“…First, the predictions and results of this study speak only to the transmission of narratives or story-like information. Prior research indicates that a different process likely governs the transmission of discrete, context-free statements, which instead tends to favor schema-inconsistency (for an overview of this work, see Kashima 2000:595) and encourage message diversification (Brashears and Gladstone 2016). This tendency likely occurs because isolated statements are not subject to two of the primary forces encouraging schema-consistency bias in narrative transmission: filling in or making sense of gaps in story detail lost in previous transmission or in one’s own recollection due to memory limitations (Kashima 2000; Mesoudi and Whiten 2008).…”

“…To test for this alternative (less theoretically interesting) possibility, I use the Levenshtein distance between participants’ narratives and the narratives received as an indicator of random, meaning-independent error in participants’ retellings (Brashears and Gladstone 2016). Levenshtein distance is a natural language processing measure of the character-edit distance between two strings, calculated as the fewest number of edits—that is, character insertions (form → for u m = 1), deletions (ho u se → hose = 1), or substitutions ( m ug → p ug = 1)—needed to transform one string into the other.…”

The Influence of Technological Links, Social Ties, and Incumbent Firm Opportunistic Propensity on the Formation of Corporate Venture Capital Deals

“…We do not find evidence, however, for technological links’ strongest negative effect on CVC deal formation when there are indirect social ties and the incumbent firm has opportunistic tendencies. In comparing our results associated with direct ties to those of indirect ties, we speculate that negative information may be somewhat diluted when transmitted through multiple steps (Brashears & Gladstone, 2016). In other words, the fidelity of negative information on an incumbent firm’s opportunistic tendencies received indirectly through third parties (i.e., IVC investors) is lower than that attained by founders through their firsthand experience.…”

Culture, Cognition, and Internalization

“…Accordingly, people reinvent (or more accurately reconstruct) new token personal culture (e.g., a belief) when they interact with artifacts in the world, whether these are spoken, written, or conveyed via other semiotic processes. These last may even introduce opportunities for errors, modifications, and "misunderstandings" during the objectification and reconstruction process (Brashears and Gladstone 2016;Hunzaker 2016). The standard "conduit" version of internalization is misleading because it invites the inference of the property-preserving (and identity-preserving) transfer of some kind of non-material entity from the world to people or from one person to another.…”