Considering the story of David Sharp, those who passed him on route to the summit seemed to have “placed their own personal glory and conquest of the mountain before the needs of another human being” (Savage & Torgler, 2013).
(Savage & Torgler, 2013) show “a weakening of the prosocial behavior in the more traditional climbers” since the introduction of commercial expeditions, “created by a crowding out effect, which may have led to the break down in prosocial behavior and the rise of antisocial behavior.” To explain their results, they emit several hypotheses and statements about the relationships between climbers and the communities they compose.
For example:
From the expeditions’ data of the Himalayan Database, we attempt at recreating the relationships between expedition members and use network analysis to evaluate these hypotheses and statements which could explain the weakening of prosocial behavior since the introduction of commercial expeditions.
The Himalayan Database as a Neo4j Graph Database
Neo4j Graph Database Data Model
There are many ways to model all the expeditions, peaks and members data. Following some Neo4j best practices (Neo4j Graph Academy - Graph Data Modeling Fundamentals) we modeled the data, representing some members or expedition properties as nodes (e.g. Route, Country), relationships (CLIMBED, ATTEMPTED) or as node multi-labels (e.g. Sherpa, Tibetan, NonSherpaNonTibetan for Members, not shown in the image).
Note: the tool to import of the Himalayan Database into a Neo4j graph database has been released here.
Within the graph model there are still multiple ways expedition member relationships could be represented. Making the assumption that all members of an expedition have partnered together in the expedition, we create a full mesh network between all the members of the expedition. To represent the strength of the relationship between two members we count the number of expeditions these two members joined together.
The Himalayan Database poses another challenge when it comes to modeling it as a graph. People do not have a unique identifier and we have limited information to differentiate them: first and last names, gender, year of birth and residence. But Sherpas all have the same last name, are 99.98% male and many are missing their year of birth, it is thus impossible to uniquely differentiate them. We generated a unique identifier for all people based on their first and last names, gender and year of birth. For Sherpas, in an attempt to improve their unique identification, we made the assumption that they rarely change residence and used it to generate their unique identifier. Despite the effort, counting those who appeared on overlapping expeditions, we estimated that about 9.7% of Sherpas are in reality non-unique. This analysis does not however account for non-unique Sherpas appearing in different seasons or years. The proportion of non-unique Sherpas is thus probably higher. We acknowledge that it will reduce the robustness of our analysis but this is the best we could reasonably assume.
If commercialization had such an impact as suggested by (Savage & Torgler, 2013) we should see some variations in the trends of some network analysis metrics through the various periods. Specifically:
A visual observation of the network giant component confirms the first expected trend. The community grows larger with each period, with the prevalence of Sherpas becoming more pronounced from the commercial period. We focus then on 5 global metrics of these networks:
The evolution of network metrics across the different expeditions periods.
Until commercialization the metrics show a network becoming mainly constituted of Non-Sherpa/Tibetans with less and less connections to Sherpas and Tibetans who are disappearing. The trends reverse from the commercial period.
Here, we disagree with the first point made by (Savage & Torgler, 2013). The proportion of Sherpas and Tibetans, along with the conductance and proportion of Sherpa/Tibetan to Non-Sherpa/Tibetan relationships declined until the commercial period. This suggests that until the commercial period, the community was becoming mostly Non-Sherpa/Tibetan, self-centered and less reliant on the local populations. This contradicts the hypothesis that traditional climbers are more and more strongly connected to Sherpas and the person they hire. Furthermore, from the introduction of commercialization the trends reverse and interactions between the two communities increase back. This can be explained by the hypothesis that, since the introduction of commercial expeditions, the network is composed of more climbers with less experience who rely on more support from Sherpas and Tibetans than in previous periods. The introduction of commercial expeditions seems thus to have a positive effect on the interactions between the two communities.
We also look at some expedition member metrics. For these metrics we look at the average but also at the variance within the 10th and 90th percentiles. The metrics are:
The evolution of relationships' strength per climbing period (mean, 10th, 50th and 90th percentiles).
We see that since the commercial expedition, most relationships have become very weak. The 90th percentile of relationship strength drops to 1, meaning that at least 90% of the relationships are just a one-time interaction during an expedition. The members' average relationship strength also dropped since the transitional period.
Since the transitional period we also see that more than 50% of the members only join one expedition. We could thus argue that the weakening of the relationships within the community started before the emergence of commercial expeditions as the number of expeditions sharply increased in the transitional period. A hypothesis for this effect is that it is easier to be “a group of people who are well known to each other and interacted often” (Savage & Torgler, 2013) when there is less than 1,000 members than when there are more than fifteen thousand.
The evolution of member's average number of expedition joined per climbing period with the 10th, 50th and 90th percentiles.
Finally, when we look period by period, we see that Sherpas make up at least half of the top 10 influencers. This suggests that the Sherpas might have been as influential in the development of the mountaineering culture and the norms of the expeditions in the early days of Himalayan Expedition history than the British gentlemen.
| Exploratory | Expeditionary | Transitional | Commercial | Social Media |
|---|---|---|---|---|
| Crawford Colin Grant | Sherpa Da Norbu | Sherpa Ang Rita | Sherpa Ang Phuri | Mosedale Timothy John (Tim) |
| Wood-Johnson George W. | Sherpa Ang Dawa | Sherpa Ang Kami | Sherpa Danu | Sherpa Pasang Nuru |
| Sherpa Ila Kitar | Kato Yukihiko | Sherpa Nawang Thile (Pemba Dorje) | Rutkiewicz Wanda | Sherpa Nuru Wangchu |
| Shebbeare Edward Oswald | Sherpa Ang Dorje | Gurung Moti Lal | Novak Jiri | Madison Garrett Christian |
| Morris Charles John | Muraki Junjiro | Stremfelj Marko | Maret Gerald | Kim Mi-Gon |
| Sherpa Lobsang | Sherpa Ajiba (Ajeeba) | Vitale Ulises Sila | Kraus Andreas (Andy) | Serpa Ang Phurba |
| Greene Charles Raymond | Thurmayr Alois | Gaillard Michel Patrick | Mazzoleni Lorenzo | Egocheaga Rodriguez Jorge |
| Sherpa Ang Pasang | Sherpa Phu Dorje | Sherpa Ang Jangbu | Murray Michael | Sherpa Dorje Sonam Gyalzen |
| Sherpa Lhakpa Chedi | Sen Amulya | Tabei Junko | Sherpa Pemba Lama | Sherpa Pasang Namgyal |
| Sherpa Jigme | Sherpa Lobsang | Sherpa Mingma Norbu | Rackl Robert | Sherpa Dawa Wangchuk (Dawa Ongchu) |
Top 10 Influencers by period show many Sherpas in every period, even the early ones.
The results may vary depending on the Neo4j CELF algorithm hyperparameters, resulting in partly different people in the top 10 or a different order but the overall conclusion remain the same: Sherpa have a lot of influence on the network across periods together with foreign expedition leaders and expedition company owners.
As local Sherpa and Tibetan communities were central to these expeditions, it seems fair to assume that their culture and norms were also central to the development of the Himalayan expedition culture and norms. These observations thus support with some nuance the hypothesis that “it is the Sherpas themselves that are the maintainers of the prosocial behavior within the traditional non-commercial group”. Within these influencers we also see some Tibetans and, since the commercial period, many foreign expedition leaders, mountain guides and expedition owners. So, more than just the Sherpas, it seems that it is all the people working, guiding, leading expeditions, whether they are Sherpas or not, who are playing a big role in maintaining the prosocial behaviors.
In conclusion, a lot of the weakening in the community relationships seems to have started before the introduction of commercial expeditions, as the number of expeditions and members grew. It even seems, on the contrary, to have a positive effect on the interactions between Sherpa/Tibetan and Non-Sherpa/Tibetan climbers, probably because the less experienced non-commercial climbers rely more on Sherpas and Tibetans. We cannot conclude from these metrics about the quality of these relationships or the evoked “emotional bond”, yet these findings do not fully align with some of the statements in (Savage & Torgler, 2013). At least in terms of quantity. If commercialization weakened the pro-social behavior, the reasons seem to, at least partly, lie somewhere else than in the weakening of the relationships.
Finally, if the introduction of social-media has changed how climbers communicate about their expeditions to the public and how people share information within the expedition community, but also outside of it, we do not see any change in the trends started in earlier periods.
- Project scout and environment setup
- Exploratory Data Analysis
- Dash App (lead and creator)
- Docker container
- Scraping and API retrieval of social media data and analysis (Twitter/Reddit)
- Final write-up (lead)
