Home

Warmth. As David’s body began to shut down from exposure he was likely hallucinating of a familiar place feeling a sense of warmth and comfort, or so we believe. Instead, he was fading under a rock ledge beneath the top of the world. It was estimated that nearly 40 climbers passed by David Sharp on May 14th, 2006 as they made their way to the summit of Mount Everest. Many tried to help him, but his condition was dire. At that height everyone knew the risks and had to carry on unless they wanted to end up in David’s position. Later, some climbers expressed confusion when asked about David. Death is so prevalent in this world that they mistook him for a famous body known as “green boots” , the corpse of an Indian climber who has been there since 1996 and which is now a grim milestone on the way to the peak. Those that consciously left David, did so to ensure the survival of their team. His condition was so bad that he could not possibly be saved. Teams that came upon him had to consider their own survival over the aid of this man, who in their mind was already dead. David’s death sparked immense controversy, but little has changed on the frozen peaks of the Himalayas. Death, injury, are all too real for those who attempt to summit (David Sharp and Everest Controversy on Mountainzone.Com, n.d.).

Community Structure: Deterioration and Growth Social Media and Climbing Topic Trends, Route Memos, and What They Tell Us About Climbers The Measured Effects of Commercialism on Climbing Communities

The first known map of the Himalayas, to any degree of accuracy, was created in 1590 by Antonio Monserrate, a Spanish missionary. Later in 1733 Jean-Baptiste Bourguignon d’Arville compiled a map of Tibet and the Himalayan range based on his exploration attempts. As time progressed, so did attempts at exploring this mountain range along with those willing to catalog locations and routes. Peaks did not initially have names; they were labeled with letters and then roman numerals, given the seemingly infinite number of peaks (Himalayas - Study and Exploration | Britannica, n.d.). Mount Everest started its lineage, humbly, as “H”. Later “XV”, until it was finally renamed to honor Sir George Everest, an accomplished surveyor (Sir George Everest | British Geodesist | Britannica, n.d.). When computations became more advanced in the 1860s, it was realized that Mount Everest was taller than any known peak. Mountaineering in the region can be traced back to the 1880s, but it wasn’t until the 1920 that attempts at Everest were recorded, and in 1953 Edmund Hillary and Tenzing Norgay were the first to conquer the mountain (Himalayas - Study and Exploration | Britannica, n.d.).

An Expedition for Knowledge

The motivations of “inner quest” or the “conquest for glory” of early expeditions came at a visible cost. Despite the high risk of injury and death, interest in the Himalayas has only increased over time. Elizabeth Hawley, while never actually climbing the Himalayas, was known as the “Sherlock Holmes of the mountaineering world” (Bhandari & Schultz, 2018). Her efforts to catalog and organize Himalayan expedition information has led to many publications and the Himalayan Database is considered a significant resource for planning (Himalayan Database, n.d.). To begin our journey, we examine Hawley’s work, the only comprehensive historical data resource of expeditions in the Nepal Himalayas.

The dataset consists of three primary data structures: the peaks, expeditions and members datasets. We explore this data to frame our questions. The work of this exploratory analysis can be viewed on our render dashboard. The dashboard includes an in-depth EDA (exploratory data analysis) which goes well beyond the insights noted here.

Note: due to a potential "cold start" of the container hosting the application when accessing the app, it can take time to load.

The popularity of mountaineering and expeditions in the Himalayan mountain ranges have skyrocketed in the past few decades (the spatial analysis graph is a strong indicator of this). There are several guided expedition organizers that support people’s interest in scaling these peaks. We can see with commercial expeditions a decrease in fatalities when oxygen usage becomes more prevalent. On Everest, we see that for non-commercial routes there was a peak in oxygen usage in the 90’s followed by a dip until 2022. Member deaths interestingly fell during these years as well and drastically reduced in the 2020s. Similarly for commercial routes, the oxygen usage steadily rose while member deaths crept lower.

In these examples we see that team composition can vary greatly, and that commercialism has correlated with some climbing behavior (e.g., the use of oxygen). An expedition team has many factors which contribute to its function and development. As communities form and change over time, the social interactions of these teams, from a network perspective, may provide insight on how behaviors may change. As climbing technology increases we see that teams making use of supplemental oxygen face fewer fatalities. For teams who run commercial ascents, the survival of team members is an important service.

As we consider changes over time, we cannot ignore the decrease in the average days to summit. For Everest, it went down from an average of 33 days to 21 days. The number of commercial expeditions has been steadily increasing on the other hand, especially in the 21st century. For Everest about 54% of the expeditions in 1990 were commercial while in the year 2000 it was 95%.

While trends indicate safer, faster climbs. The harsh reality of the peaks remind us of the real dangers that come with climbing. In 2014, both Everest and Lhotse saw a spike in the percentage of hired expeditioner deaths. An ice avalanche caused immense loss in Sherpa (hired expeditioners) lives (“2014 Mount Everest Ice Avalanche”, 2023). While the dangers are very real, this has not deterred many, to "conquer the useless”.

Commercial Expeditions on Everest in 1990

Commercial Expeditions on Everest in 2000

Our questions

What really drives people to expose themselves to the vagaries of a potentially perilous expedition? How have the expeditions and the community changed over time? Has the introduction of commercialism impacted people’s motivations and community dynamics?

To try to answer our questions we based our analysis on the Hawley and Salisbury Himalayan Database. Historians and scientists have studied this dataset using classic statistical tools. We propose 2 novel approaches to deepen our analysis:

First by enhancing the database with social media data and using Natural Language Processing techniques like short text topic modeling (hereafter referred to as STTM) on Twitter and Reddit posts. This will reveal topic trends and facilitate unsupervised intent discovery Reddit posts. Additionally, clustering, topic modeling and feature extraction on the Himalayan Database route memos is performed for each period of exploration as defined below.
Second by using network analysis techniques on the graph representing all the interactions between mountaineers.

We define periods of climbing expeditions based on the original interpretations from (Salisbury et al, 2021) and suggest a new period based on the connection with social media. The 5 periods are defined as follows:

Compared to other studies, we end the commercial period in 2004 as we consider the introduction of social-media from about 2005 as a new period changing how members of Himalayan expeditions share their experiences with their audience but also share information with each other.

Through these novel approaches we gain insights about the evolution of the interactions between the different communities within the expedition members population.

Ethical Considerations, Limitations and Next Steps

The Himalayan Database, we base this project on, has a limited amount of data on commercial and non-commercial expeditions. We are unaware of a more complete dataset regarding these and so the narrative is based on this limited data from this dataset’s perspective. The public version we are using has already been stripped of Personal Identification Information, mitigating ethical issues related with the handling of such data.

We will be incorporating social media data along with this dataset, which comes with limitations, since it was noted that not all expeditioners and organizations post on the social media sites that allow for access to their information. As such this will bias our analysis to be based on the media sites we have access to, what is publicly available on them and who actively posts in these sites. We mention this clearly in our analysis of topics and trends.

A challenge we faced with route memos in the Himalayan Database was that many of them were an amalgam of factual data that had been processed and placed in a structured format in the database tables, and stories that we are interested in which summarize experiences for those who follow in the tracks of pioneers. In the next iteration, we would split the route memos into individual sentences and eliminate those with just factual information. We could then perform intent discovery on the remaining longer sentences, each being a document in an expedition, using BERTopic. Since BERTopic assumes one topic per document, we should be able to extract multiple topics for each expedition's route memo. We should be able to get a better sense of similarity of expeditions (in each period) by assessing the similarity of main topics in each expedition. It would also help us visualize topic trends over time from a multi-topic perspective of each memo.

Finally we acknowledge that the results of the network analysis are highly dependent on the graph modeling and community detection algorithms used, and that different approaches might yield different results. Further work would explore different modeling of relationships, measure of their strength and use different network cohesiveness metrics to evaluate the consistency of the results found here. Also, instead of detecting communities using different algorithms, other approaches could focus on the Sherpa/Tibetan and Non-Sherpa-Tibetan communities' relationships over time.

Final Thoughts and Conclusions

We began our expedition into the minds and hearts of climbers with an understanding of how climbing has progressed since the early 20th century. The data indicated a surge in climbers from the late 1990s, and we demonstrated that commercialism is different in expeditions (e.g., the use of oxygen, fatalities, etc). This helped us frame the questions:

We attempted to answer these questions with novel approaches that have not been published on this data set. Considering the similar trend in the rise of social media, we evaluated the trending topics and sentiments of the twitter climbing population. We learned that the most common tweets and topics relate to early expeditions. Tweets quote Sir Edmund Hillary, discuss motivations like personal glory, and to pursue a challenge that has no tangible reward. We also observe hints at the community dynamic, where concern for climber safety is noted. Along with a mixed population of experienced and novice climbers. The tweet topics also acknowledge the danger of climbing, but with strong topics like “bucket-list” (a list of things to accomplish before death), the desire to challenge oneself is a stronger motivation than personal safety. We see this clearly in the rise of novice climbers and their community structure in the network analysis.

Using a Neo4j database, a network was created to evaluate past and present communities. What we saw, was a change around the connections with Sherpas and Tibetans. Early, the connection was strong, but declined over time. With commercialism developing, the connections rise again which most likely is related to the increase of novice climbers, suggesting this is not a reason we observe the decline of prosocial behavior suggested in David Sharp’s story.

We suggest that social-media has changed how climbers interact with others and share information. Their discussions of technically challenging routes, and concern for other climbers show their awareness of the risks, yet many are willing to undergo these expeditions. We observe that leadership, courage, and personal achievement are motivating factors, more than personal safety. Discussions also overlap with the community analysis, where we see many inexperienced climbers looking for help, establishing connections, and attempting to at least achieve ascent to Everest base camp.

Analyzing the content of the route memos we demonstrate that at its core, climbing hasn’t changed. Those who document their ascents and explore new routes share the same stories and information as those that came before them. There is no specific content that is period specific, and while external changes continue to occur with technology and community shifts, motivations of elite, experienced climbers, appear unchanged and untouched by the commercial and social-media periods.

Finally, we measure what effect of commercialism has had on the climbing community and explore statements made by (Savage & Torgler, 2013). We fail to reject the notion that commercialization of the peaks had no effect on cohesiveness or relationships’ strength, showing that in this period the split between commercial and non-commercial climbers also demonstrates skill level. More experienced and elite climbers remain unchanged from earlier periods and rely less on Sherpas and Tibetans while novice climbers require their aid. Commercialism shows a positive effect on conductance, meaning these commercial communities are more connected to each other than their non-commercial counterparts.

What we see overall is that the methods, motivations, and communities of elite climbers have been impacted little over time. Since the early 1900’s as expeditions increased, there has been a steady select group of mountaineers who look to explore, document, and move in smaller circles within the climbing community. Newcomers to this scene in the late 1990s do not need connections in Nepal to establish a climbing team or find a guide. They use technology to fill their gaps in experience and have a communal online “experience” that is openly shared. These individuals rally behind the stories of others and we see a theme of inexperienced individuals extremely motivated to venture to the Himalayas later in the commercial period as social-media becomes more prevalent. We propose the emergence of this new period as unique from the changes we see within communities, the effect social-media has on motivation. Welcome to the communal era of climbing.

Statement of Work (April 2023)

Simi Talkar

- 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)

Brian Seko

- Data cleaning and structure
- Route Memo Clustering
- Route Memo Topic Modeling
- Climbing Period Feature Analysis (not included here)
- Final Write-Up (lead)

Matthieu Lienart

- Scraping of additional Himalayan peak data
- Data cleaning and structure
- Neo4j Database
- Network Analysis
- Poster Creation (lead)
- Website (lead)
- Final write-up

An Expedition for Knowledge

Commercial Expeditions on Everest in 1990

Commercial Expeditions on Everest in 2000

“Climbing is the conquering of the useless”

Our questions

Exploratory period

Expeditionary period

Transitional period

Commercial period

Social Media period

Analyses

Community Structure: Deterioration and Growth

Social Media and Climbing

Topic Trends, Route Memos, and What They Tell Us About Climbers