Reflection #2 – [1/23] – [Pratik Anand]

The paper poses an interesting research question that how much success of a Kickstarter campaign depends on the campaign’s presentations, pitch and other factors which have no relation to the product itself. It is interesting because unlike other kind of media like advertisements, direct impact of such influences can be measured in terms of donation to the Kickstarter projects.

Tanushree Mitra et al. list out a number of factors which influence the viewers, positively or negatively. These factors, or control variables, are : project goal, its duration, video or animation used for the pitch, category of the product, Facebook connectivity etc. Impact of a video or an animation is well understood as they provide the information in a short amount of time and keep the viewers engaged compared to a large block of text. Project duration also plays a key factor. I can understand why longer project duration is seen negatively and such projects are less likely to reach their funding goal. The viewers have little interest or trust over paying for a product whose result they may see after a long duration. Products which take longer to develop are tell-tale signs of complexity and can lead to disastrous failures. Such trust deficit can only be offset by strong brands which usually Kickstarters don’t have.

Tanushree Mitra et al. built a logistic regression model for prediciting the success of kickstarter campaigns with these control variables. It resulted in 17.03 % error rate in 10-cross validation.
The authors factors in the phrases of language used in the Kickstarter campaign and the error rate reduces to 2.24 % which shows a strong correlation between language of the pitch and the success of the product. They try to explain the phrases as a trigger for one of these phenomena – Reciprocity, Scarcity, Social Proof, Social Identity, Liking and Authority.
Many of these phenomenon like scarcity, social proof and identity as well as authority are well studied psychological phenomenon, especially in the retail and entertainment industries which employ all kind of techniques – from loyalty bonuses, exclusive cards to ad campaigns which instill a FOMO (Fear of Missing Out) among the users [1]. Every other advertisement has an “expert” who claims that the given product/service is the best. Tanushree Mitra et al. reference these as part of Theory of Persuasion. Since, these are older tricks in the classroom marketing and advertisement books, it is debatable that how much effective they are in the kickstarter campaigns. Correlation does not imply causation.
Reciprocity, on the other hand, stands out as an effective technique. Kickstarter campaign, by their nature, do not give anything in return to the backers except for the promise that the product will come out for the consumers. If the Kickstarter campaign gives back something tangible to the backers, it is a very visible addon for them.
The paper shows that adding phrases and control variables to their model, they achieve high degree of accuracy in predicting success of a campaign. If platforms emerge for Kickstarters to tune their pitches based on these suggestions, will their effect subside from overuse ?
This study was performed in 2014, more than 3 years ago. Kickstarter now is a very different and diverse platform with newer options, long list of high profile success and failures (Pebble has been acquired after a string of losses, Ubuntu phone was a failed campaign, Oculus is a major player in VR etc). Product discovery portals like Product Hunt are also influencing popularity of the campaigns. Do these conclusions hold up for Kickstarter of 2017 ?

Reference :
1) https://www.salesforce.com/blog/2016/10/customer-loyalty-program-examples-tips.html

Read More

Reflection #1 – [1/18] – [Pratik Anand]

First paper : CRITICAL QUESTIONS FOR BIG DATA

Second paper : Data ex Machina: Introduction to Big Data

Summary

Both papers start with the introduction to big data. One of main points raised is the changing definition of big data – earlier it was about size of data, now it is about search, aggregate and processing the data. First paper directly jumps to the critical analysis of big data phenomenon. Second paper, on the other hand, takes a more comprehensive and structured approach to the analysis of big data. Both paper discuss the source of data acquisition, emergence of an entire new set of digital data, data privacy concerns, its analysis and conclusion. They show the changing approach to research due to big data and how abundance of data gathered might not lead to the true representation of the whole problem sample. Even conclusions provided from balanced representative data might also be subjective and biased. The mention of “Big Data Hubris” shows that having large volume of data does not correlate to better results. While the first paper continues critical analysis of big data, second paper provides future trends where volume of data will grow in size and diversity of platforms and more generic data models will take over.

Reflection

The first paper does a really good job of raising important questions related to big data, starting with its definition. The second paper is more of an introduction to big data and only provides a high level view of issues related to big data. For the initial reading, second paper can be used to provide overview of the big data industry, followed by the first paper for discussions related to its most important issues. Issues like privacy are unethical data collection are central to this debate as individuals, corporations or even governments can easily misuse the data.The contrast between the nature of the two papers is quite evident. The second paper mentions the problems faced by big data field today and its future trends whereas the first paper discusses over the problems caused due to big data and critically analyses its aspects. Aspects like uneven access to digital data and poorly understood analysis results can have ramifications on large sections of society and the first paper raises the right kind of questions for civil discussions.

Questions

1) Since the volume of data generated will continue to grow, how the government can ensure its protection and ethical treatment ?
2) Who should actually own the digital footprint data – individual or the respective companies collecting it ?
3) Is the data-driven approach is the right approach for all kind of social problems ? Will it lead to less focus towards areas where it is inherently difficult to generate large volume of data ?

Read More