Welcome to our wiki! Read the instructions and get going!

Hospex data research


This article describes scientific data insights on internet-based hospitality exchange communities (hospex). The supposed "communities" are listed in chronological order -- hospitalityclub.org (HC), couchsurfing.org (CS) and bewelcome.org (BW). More results are derived from our own BW data than from the data of other communities. It is important to mention that the three communities are interconnected; presenting the data together provides additional information. Bewelcome charts is a continuation of this article.

Data available

CS data has been made unavailable for external research since its incorporation in 2011. Available data on other communities includes the google search data for HC and CS and the mirrored statistics of CS in 2011, which are no longer available concerning the CS website. For what it's worth, the current official statistics page of CS presents figures heavily reduced and clearly different from their statistics published in 2011.

Related work

We can identify three categories of existing scientific publications claiming hospex as their subject matter -- non-data scientific articles, analysis of survey data, and analysis of CS data. Survey data gives insights into mindsets, but not into real behavior processes on hospex.

The following list of publications used pre-incorporation CS data:

Bilattice-Based Aggregation Operators for Gradual Trust and Distrust by P. Victor, C. Cornelis, M. De Cock and E. Herrera-Viedma

Analysis & Generative Model for Trust Networks by Pranav Dandekar

Trust Propagation with Mixed-Effects Models by Jan Overgoor, Ellery Wulczyn and Christopher Potts

Surfing a web of trust: Reputation and Reciprocity on CouchSurfing.com by Debra Lauterbach, Hung Truong, Tanuj Shah and Lada Adamic

Google Search vs. Signup

Monthly google search volume (green lines) and the number ot people signing up monthly (blue lines) are highly correlated. The correlations are 0.971 for CS and 0.913 for BW. Unfortunately, we have no signup data for HC. However HC signup data can be estimated using the google search data. It should be mentioned that the google search volume of Bewelcome is about 1-2% of that of Couchsurfing. Therefore, we could plot both on the same scale.

Seasonality of hospex activities

Hospex activity is seasonal with peaks in summer and lows in winter. All travel related google searches, for example "Booking.com", follow the same pattern with a peak in July or August. Growth-independent seasonal variations for the sum of HC and CS in the year 2004-2012 are depicted on the plot below.

CS caused HC's dieback from the year 2006 on

HC existed before CS. It predominated google searches until 2007, despite the statement in the redrafted CS statistics describing CS as "leading social travel since 2004". Beginning in 2007, CS has surpassed HC in terms of google search as you can see on the graph below. The green line depicts the time point of the CS database crash. What you can also see with the naked eye is that HC google searches declined immediately following this same point in time, and they remained seasonal only until 2009. The obviously lack of seasonality in HC google searches after 2009 suggests that search for HC was no longer primarily related to travel.

Members migrated from HC to CS in the years from 2006 to 2009, causing rapid growth by CS. In fact, if total existing CS data (blue line) is extended in time, it would predict 19 millions members in 2017 (red line). But the rapid growth curve of CS, which was powered by the dieback of HC, stopped in 2009. Today, the green line, which uses post-2009 data, predicts only 15 millions members in 2017.

This break in growth of the signup rate can be seen better, after some transformation, on the plot below. Here deviations from the power function fit and seasonal oscilations are depicted. We also observe the negative peak in summer 2006 due to CS database crash.

CS's "a better world - one couch at time" failed

The 75% of the world population which lives in poor countries with per capita GDP under $10,000 represented only 10% of CS members in 2011.

CS and Airbnb

There is a difference between hospitality for money (airbnb) and hospitality for free (hospex). Basically, paid money frees from social obligations. Listen to Money Changes Everything by Dan Ariely to better understand. The graph below shows that the rapid growth of interest in airbnb has not yet affected interest in CS. Airbnb is not a "better CS" in the way that CS was a "better HC".

Results on Bewelcome

Bewelcome charts is a continuation of this article.