Cultural, economic, and social forms of capital on Steemit [ORIGINAL RESEARCH]

Note: All data for this experiment are available in an appendix located at the end of the post.  

Abstract

I recently carried out a study on associations between user reputation, Steem Power, and lifetime author rewards using data available on the blockchain. Results show a causal cycle among these three factors, where 

• Steem Power predicts total lifetime author rewards and reputation score
• Reputation score predicts lifetime rewards
• Lifetime author rewards predicts level of Steem Power.

This cycle is visualized below. 

I argue these findings are in line with Bourdieu’s (1986) theory of conversions between social, cultural, and economic forms of capital, and that they confirm the central role of post quality in amount of Steem Power held.   

1. Introduction

The social status of every individual on Steemit is public information. Unlike offline society,  every user’s reputation, total economic capital, and lifetime author rewards for work are viewable by anyone as unambiguous numeric scores.  

Despite this abundance of easily-accessible data, sociocultural research on samples of Steemit users is so far nonexistent. To begin filling this gap, I will offer results from an exploratory study on associations between reputation, Steem Power, and lifetime author rewards, carried out in late August, 2016. Multiple regression analysis on a stratified sample of 52 Steemit users suggests a causal cycle exists among these three factors, which I argue resembles social theory by Bourdieu (1986) on the conversion of economic capital into social and cultural capital. Results also might help confirm the causal role of post quality in amount of Steem Power held by the user. 

2. Forms of capital on Steemit 

French sociologist Pierre Bourdieu (1986) suggested the existence of three forms of capital:    

• Social Capital: this is your standing within your social network and the benefits you derive from it. 
• Cultural Capital: This is things like your eduction/knowledge, literacy, eloquence, and social abilities.
• Economic Capital: This is your money and other objectively valuable holdings.

Bourdieu (pictured below) is famous for systematically explaining what many of us already think: that society is intrinsically rigged to benefit some people and not others. He suggests that economic capital underlies and reproduces the other forms of capital any given person or group possesses: capital is passed on through the family and through social networks, and this reproduces unequal social structures across generations because different families and social networks pass on different amounts of capital. This is a painfully obvious point, but it is one Bourdieu explains in tremendous detail: anyone can say this is obvious, it took an intellectual powerhouse to elegantly parse out its details.   

Image credit

For instance, your parent or your government's money buys your education, which determines your cultural literacy, or capital, and in turn who you socialize with. Your social capital emerges from reciprocal connections within your network, while how you are connected within the network determines what jobs you get and thus your amount of economic capital, which you pass on within your family and network.

Bourdieu suggests conversions of capital like this explain how unequal capital reproduces social stratification across generations: by effectively disguising it as the other forms of capital. According to Bourdieu, all social inequality is simply concealed financial inequality.  

Research focus and key terms
Since joining Steemit, I’ve wondered if comparable structural constraints on the success of individual users exists on Steemit. Does the structure of this new economic system reproduce the financial inequalities of its participants? I decided to test if Bourdieu’s constructs, i.e. his ideas about conversions of capital and the reproduction of inequality, will map onto aspects of the Steemit economy. 

I designed a study where Steemit reputation stands in for Bourdieu’s social capital; lifetime author rewards are taken as cultural capital, and Steem Power is taken as economic capital.

Justifying comparisons of Steemit to Bourdieu’s forms of capital 

Are the comparisons in the table above valid? I think so, but on the level of the individual, rather than on the intergenerational level noted in Bourdieu (there has not been time enough for Steem to become an intergenerational affair). That is: 

Your reputation ranking is literally your social capital Steemit — it is an incontestable measure of how well connected and respected you are within the social sphere.  

Your lifetime rewards are a literal measure of your cultural capital, in that they display your success in internalizing and reproducing the tastes and standards of Steemit society (see Bourdieu, 1986: 47-53).  

Steem Power is economic capital, not only in the sense that it can be turned into standard currency, but in its effect on voting power: in the physical world, the amount of money you have often determines how much your decisions matter and influence other factors within the system. Steemit is an almost ideal example of this because while some buying decisions in the physical world are minimal in their influence upon the economy, Steemit is set up to make all the things whales like explode in popularity and influence.  

Given these comparisons: I hypothesized user's reputation, lifetime authors rewards, and amount of Steem Power will predict one another in the conversional fashion outlined in Bourdieu (1986), wherein holding economic capital precludes other forms of capital. Other interactions between these factors are expected, but not specifically hypothesized. 

3. Methods

Data collection

I began by taking a stratified sample of users who currently had trending posts (8/24/2016). I copied and pasted all the top trending posts valued over $1 off of the Steemit front page and into a spreadsheet, of which there were 475. I decided on a sample size of 50 (about 10 and a half percent of the group), allowing for a 13% margin of error at a 95% confidence level in my results (way too high for academics and professionals, totally fine for exploratory studies like this one).  

To make sure my sample was representative of all the trending posts, I used a technique called stratified random sampling. Once all posts had paid out (8/26/2016), I broke the group into value-strata, calculating what percent of the 475 were valued over $1000; what percent were valued between $100 and $999, what percent were valued between $10 and $99, and what percent were valued between $1 and $9 dollars. I made sure that my sample had roughly the same percentages per group as the original sample of 475 by generating random numbers between one and the number of posts per value-group. Due to rounding, the final sample size in was 52. 

Before posts had payed out, I logged each of the 52 user’s reputations, level of Steem Power, and lifetime author rewards in a spreadsheet (see data in appendix).   

Analysis
I plotted these three factors into a bubble chart, shown below, which revealed a three-way relationship where low scores on one factor associate with low scores on the others, and high scores also cluster together. Note that the vertical axis is on a logarithmic scale, meaning each tick up or down the Y axis indicates values at x100 those of the previous tick. 

I used stepwise multiple regression to determine each factor’s influence upon the other. Basically, this means I asked the computer “do any two of the factors predict the third?”, adding that if one of the two factors analyzed as predictive of the third accounted for all the variance of the other upon the factor of interest, the one who's variance was explained by the other predictor was excluded from the analysis. This procedure allows us to parse out what factors do and do not cause the others.  

I charted results as a path diagram indicating causal relationships among factors, shown in the results section below.  

4. Results 

The arrows in the path diagram below indicate causal direction determined in the regression analyses, and the numbers (.xx) near each arrow indicate how strong the causal connection is (.00 means no cause of one upon the other, .99 means an unrealistically strong estimate of causal influence).  For every one unit increase in the factor being caused, we would expect .xx increase in the other. For the stats nerds out there, these are standardized beta weights. 

Notice the two-way causal path between lifetime rewards (cultural capital) and Steem Power (economic capital). The present data show that for any given user, holding more Steem Power predict both higher lifetime author rewards and greater reputation scores, as hypothesized -- but that an even stronger causal relationship exists from rewards into Steem Power, i.e. that what people earn from their posts better predicts how much Steem Power they will posses than visa-versa.  

This finding supports the official Steemit claim that better posts by the user lead to them holding more Steem Power. However, The stronger relationships between Steem Power and reputation, and reputation with lifetime rewards also suggest structural constraints on user success exist, i.e. that without high capital holdings  success on steemit will be more challenging. 

5. Discussion

If future research (that is more robust, thorough, and stringently conducted studies with MUCH larger samples) in this arena yields similar results, we can say two things: 

1) Bourdieu's ideas map onto the Steemit economy to some degree: past economic holdings are a central root of much forthcoming personal successes within society (“the rich get richer”). However, outcomes for users on Steemit are not completely constrained by how much Steem Power they hold, as quality posting is doubtlessly  a major influence in acquiring Steem Power. 

2) While good content is absolutely key to your success on Steemit, structural inequalities like those in the real world do determine the social positions of users to some extent, creating what amounts to a self-reproducing social caste system that is, by nature, incredibly hard to escape from for those at the bottom.   

6. Social research on Steemit

Much has been posted on Steemit about the unequal influence of Whales versus minnows (etc.). I argue that discussion of inequality in influence and power demands not just polemics, slogans, and the occasional chart by authors on Steemit: analytical methods from the social sciences offer us a way to infer the underlying currents of Steemit that are at first invisible, or perhaps only hinted at by surface numbers.  

With enhanced data collection methods and collaborative analyses, Steemit may be an arena where sociological and anthropological research methods serve to filter our sea of data. This is, after all, what we social scientists do: we offer both theoretical and practical analyses of the essential, hidden forces, currents, and themes of society which normally both come and go without saying within our society (Bourdieu: 1977: 167).  

7. References

Bourdieu, P. (1977) Outline of a Theory of Practice, New York: Cambridge University Press   
Bourdieu, P. (2011). The forms of capital.(1986). In Cultural theory: An anthology, 81-93. United Kingdom: Wiley-Blackwell    

8. Appendix: Study Data

reputation	power	lifetime author payout
67	12674	40364
55	3448	3.59
63	5091	16857
61	2077	21456
65	4654	16132
65	5962	20543
61	3814	21456
51	108	2
61	1333	5446
67	16091	29968
55	251	628
65	4654	16132
55	836	1302
51	3	0
54	354	1937
50	4	201
56	1101	397
63	3063	7551
58	2174	5807
51	215	156
47	3	81
57	1431	564
50	20	67
61	3450	2453
47	3	0
65	4269	9097
55	441	1672
53	62	140
64	1439	7063
61	3814	21456
58	1141	4759
55	253	905
61	681	2945
54	23046	220
59	1615	4579
68	23145	31372
52	69	256
58	293	643
55	454	291
58	407	239
63	14660	6276
54	611	643
61	4863	3462
47	21	11
52	1368	571
53	76	326
37	3	0
55	570	395
62	5119	20311
42	3	0
48	40999	0
45	59	0