Reaction to New Security Threat Class

Yuval Elovici, Lior RokacharXiv 2014 · arXiv

Abstract

When a new cyber threat emerges, how long does it take for the defensive community to respond? This paper quantifies the lag between the discovery of a new threat class and the arrival of scientific research and commercial solutions. By analyzing a massive corpus of literature, the authors identify a dangerous "window of opportunity" for attackers that typically lasts two to three years before effective commercial products hit the market.

Introduction: Fighting the Last War

In military history, generals are often accused of "fighting the last war"—being perfectly prepared for yesterday's tactics while remaining blind to today’s innovations. The cybersecurity community faces a similar struggle. While new threats are identified and named constantly, there is a significant delay before the scientific community publishes research and an even longer delay before the industry files patents and releases products.

This research seeks to measure this "reaction time" across five major threat classes: Phishing, SQL Injection, BotNets, Distributed Denial of Service (DDoS), and Advanced Persistent Threats (APT). By quantifying this gap, the authors highlight a structural vulnerability in the development of cyber defenses.

Methodology

The researchers employed a bibliometric analysis—a data-driven examination of the history of scientific and professional literature. The process involved four key stages:

  1. Identification: Defining the "birth" of a threat class based on its first mention in newsgroups, media, or technical reports.
  2. Data Collection: Searching 11 major bibliographic databases for publications related to the five chosen threat classes between 1990 and 2012.
  3. Classification: Applying ensemble learning algorithms to automatically classify documents as "Academic," "Professional," or "Patent."
  4. Milestone Tracking: Comparing the dates of the first reported incident against the first scientific paper, professional article, and patent application.

Dataset Composition

The study analyzed a massive corpus of metadata. Starting with an initial extraction of 146,282 papers, the researchers arrived at a final dataset of 72,221 unique items after deduplication. The scale of the data across various bibliographic databases is detailed below.

Table showing the distribution of papers by type across the bibliographic databases used in the study.

The study used the first known mentions of specific threats (e.g., the 1987 Felix and Hauck paper for Phishing or the 1998 "Puppy" report for SQL Injection) as the baseline for "Day Zero."

Key Results: The Timeline of Reaction

The paper provides a stark timeline of collective reaction speeds. In almost every case, the scientific community takes about one year to produce its first formal publication after a threat is identified. Industry reaction, measured by patent filings, is significantly slower, taking between one and six years.

Timeline of first reports, professional mentions, scientific publications, and patent applications for major threat classes.

Specific findings include:

  • Phishing: First reported in 1987; the first patent wasn't filed until 2004—a staggering 17-year gap.
  • BotNets & DDoS: Both emerged around 1999. Scientific papers appeared within a year (2000), but patents lagged until 2001 (DDoS) and 2004 (BotNets).
  • APT: This class had the fastest turnaround. Mentioned in 2006, with professional, scientific, and patent activity all appearing by 2008.

Analysis and Insights

The 2-3 Year Vulnerability Window

Because commercial products usually follow patents by at least a year, there is a predictable "window of opportunity" for attackers. For the first 24 to 36 months of a new threat's life, users are essentially unprotected by effective commercial tools.

The Industry "Wait-and-See" Approach

The industry lag isn't necessarily due to technical incompetence, but rather market demand. Companies rarely invest in a product until they are sure the threat is widespread enough to be profitable. This creates a lag where the "good guys" wait for a market to form while the "bad guys" exploit the lack of defenses.

The 10:1 Ratio

On average, for every 10 scientific publications on a threat, only one patent is filed. This suggests that much of the defensive research discovered in academic labs never makes it into a commercial product, representing a significant gap in technology transfer.

Limitations and Conclusion

The study acknowledges that a threat might exist long before it receives a unique name (like "Phishing"), potentially underestimating the true age of a threat. Furthermore, the study tracks the volume of publications rather than correlating them with actual economic damage.

Despite these limitations, the research clearly demonstrates that our current defensive pipeline is reactive rather than proactive. Future work should investigate whether the scientific community is prioritizing the most damaging attacks or simply the most popular ones.

Summary updated

Abstract

Each new identified security threat class triggers new research and development efforts by the scientific and professional communities. In this study, we investigate the rate at which the scientific and professional communities react to new identified threat classes as it is reflected in the number of patents, scientific articles and professional publications over a long period of time. The following threat classes were studied: Phishing; SQL Injection; BotNet; Distributed Denial of Service; and Advanced Persistent Threat. Our findings suggest that in most cases it takes a year for the scientific community and more than two years for industry to react to a new threat class with patents. Since new products follow patents, it is reasonable to expect that there will be a window of approximately two to three years in which no effective product is available to cope with the new threat class.
BibTeX
@misc{elovici2014reaction,
  author = {Yuval Elovici and Lior Rokach},
  title = {{Reaction to New Security Threat Class}},
  year = {2014},
  eprint = {1406.3110},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1406.3110},
}
Figure