Large sources of digital trace data (i.e. “Big Data”) have become increasingly important in the study of material culture. However, akin to the offline material culture traditionally studied by archaeologists, digital trace data is rarely a passive reflection of human behavior – it is a complex palimpsest produced through a variety of erasure and accretion formation processes. To better understand how digital trace palimpsests are formed and how digital formation processes influence and inform our ability to interpret the offline material processes they index, we introduce a computational method – high-frequency archaeological survey – which allows us to observe digital formation processes at a high temporal resolution, as well as a large spatial scale. Using this method every hour for one month, we surveyed posts from across the United States in Craigslist's “Free Stuff” category (popularly called “Curb Alert”), a user-generated source of big digital trace data, indexing material things that have been placed on users' curbs for removal by scavengers or trash collectors. For each post, we observed its time-to-erasure and any edits that were made during the study period – finding that the posts that survive represent a biased sample of those that were posted over the course of the month, conditioned by how recently and on what day the post is posted, the material characteristics of things that are posted about, as well as regional variation. Far from only being evidence of biased end-of-month data, however, we show that further analysis of identified digital formation processes can be an important object of study in its own right – in this case, shedding new light on social scientific questions linking the exchange of “free stuff” with the process of social stratification and urban inequality in the United States. Overall, our findings suggest the importance of accounting for and explicitly analyzing digital formation processes in studies that utilize digital trace data.

大量数字追踪数据（即“大数据”）来源在物质文化研究中变得越来越重要。然而，与考古学家传统上研究的线下物质文化类似，数字痕迹数据很少是人类行为的被动反映——它是通过各种擦除和增生形成过程产生的复杂重写本。为了更好地理解数字痕迹重写是如何形成的，以及数字形成过程如何影响和告知我们解释它们所索引的离线材料过程的能力，我们引入了一种计算方法——高频考古调查——它使我们能够以时间分辨率高，空间尺度大。我们在一个月内每小时使用这种方法，调查了美国各地 Craigslist 的“免费内容”类别（通常称为“遏制警报”）中的帖子，该类别是用户生成的大数字跟踪数据源，对已发布的重要内容进行索引。放置在用户的路边，供拾荒者或垃圾收集者清除。对于每个帖子，我们观察了其删除时间以及研究期间所做的任何编辑 - 发现幸存的帖子代表了该月内发布的帖子的有偏见的样本，其条件是最近和帖子发布的日期、发布内容的材料特征以及区域差异。然而，我们不仅是月末数据存在偏见的证据，而且还表明，对已识别的数字形成过程的进一步分析本身就可以成为一个重要的研究对象——在这种情况下，为解决与美国社会分层和城市不平等过程中的“免费物品”交换。总体而言，我们的研究结果表明，在利用数字跟踪数据的研究中，解释和明确分析数字形成过程的重要性。