diff --git a/chi_rebuttals/2017-airbnb_couchsurfing/README.txt b/chi_rebuttals/2017-airbnb_couchsurfing/README.txt
new file mode 100644
index 0000000..bb81a0d
--- /dev/null
+++ b/chi_rebuttals/2017-airbnb_couchsurfing/README.txt
@@ -0,0 +1,6 @@
+Rebuttal for:
+
+Klein, Maximilian, Jinhao Zhao, Jiajun Ni, Isaac Johnson, Benjamin Mako Hill,
+and Haiyi Zhu. 2017. “Quality Standards, Service Orientation, and Power in
+Airbnb and Couchsurfing.” Proceedings of the ACM on Human-Computer Interaction
+1 (CSCW): 58:1–58:21. https://doi.org/10.1145/3134693.
diff --git a/chi_rebuttals/2017-airbnb_couchsurfing/airbnb_and_counchsurfing-2017-rebuttal.txt b/chi_rebuttals/2017-airbnb_couchsurfing/airbnb_and_counchsurfing-2017-rebuttal.txt
new file mode 100644
index 0000000..94de3f0
--- /dev/null
+++ b/chi_rebuttals/2017-airbnb_couchsurfing/airbnb_and_counchsurfing-2017-rebuttal.txt
@@ -0,0 +1,29 @@
+Thank you very much to the reviewers for their excellent critique and affirming our faith in peer review. In this rebuttal we respond to the most serious issues: the “too compacted” qualitative method, and  “arbitrariness of operationalization.” We are confident in making changes in a camera-ready timeframe.
+
+1. Qualitative analysis was “too condensed” and lost “variance in opinions” (1AC, R1).
+We did condense the qualitative part to give room to the quantitative analysis. We are going to put more details of the qualitative analysis back in the revision. First, we will include the statistics of the themes leading to our propositions. For instance, themes that lead to our first proposition were “Airbnb is a service with minimum quality standards” (mentioned by 12 participants) and “Couchsurfing has no quality reputation” (8). Our use of the RQDA tool will make this easy work for all propositions.
+
+Second, we will include more quotes and detailed discussions for each proposition. For example, complicating the “People vs. Places” theme, we heard an Airbnb relationship can grow beyond business-only. Participant 3 shared “[Once] I was supposed to stay there for a month but I ended up staying there for four to five months just to help out [my host] after her mother had died.” Here we find a very intimate blooming of the guest-host relationship around a death. Indeed it shows that the theme is not "People *or* Places", but that every interaction exists on a fluid spectrum which money does not exclusively determine. We had these quotes in a previous version of the paper and can easily include them in the revision.
+
+In order to make more room for the expansion of the qualitative work we are going to shorten the “Popularity Shift” section which was considered “unnecessary” by R1.
+
+2. Arbitrariness of Operationalizations (1AC, R1).
+We appreciate the reviewers' concern that these analyses are not well-justified in the paper. We did have specific reasons for each, which we include here. We will also include brief discussions of rationale in the revised paper.
+
+2.1: Testing proposition 1: correlation of house price and Airbnb and Couchsurfing hosting rate.  
+We interpreted a regression in which Airbnb hosting rates increased with median house prices, but in which Couchsurfing rates remained unchanged, as showing Airbnb having minimum quality standards but not Couchsurfing. 1AC provided alternative interpretations: (a) selection bias, (b) cost of living, (c) spaces more amenable to hosting. We agree, our correlational study does not give causation and can not rule out an alternative explanation like (a) selection bias. Still, the value of this mixed-methods examination is to demonstrate how interviewees’ perceptions (“Airbnb standards”) manifest in the aggregate data, and that the two approaches triangulate and complement each other.  
+
+Because cost of living (b) may necessitate increased Airbnb hosting, we re-ran our regressions partially-joined with cost of living data [1] and found no major changes in significance or effect size in the 10% of data we were able to automatically match.
+
+Yet (c) suggests having more expensive, shareable space affects propensity to Airbnb-host. However that would not explain why Couchsurfing-hosts remain steady in similar markets.
+
+2.2. Testing proposition 2: Why use profile images and facial recognition?
+R2 liked the occurrence-of-faces in profile images test, but noted that we could have alternatively examined the number of images and length of home descriptions. This is a good point which we too considered. However the user interfaces are notably different and would introduce confounds. Whereas Airbnb requires text, images and amenity itemization about listings, Couchsurfing lacks the itemization feature and requires no content at all. Both sites do require profile images which allowed a fairer comparison, and so became the subject of our test.
+
+
+2.3 Testing proposition 3: guest-host power dynamics in Airbnb and Couchsurfing.
+Our test of whether hosts or guests rated each other more positively was not arbitrary at all, but a replication and extension of the previous study “Power Imbalance and Rating Systems” (2016). Reassuringly our test confirmed the previous results about Couchsurfing being host-dominant, and showed the opposite to be true in Airbnb. 1AC and R1 requested more justification for whether using a classifier trained on movie review data was “similar enough to provide good results”? We agree that cross-domain use is not ideal. However in a case study using *movie-reviews* to classify *book-reviews*, accuracy was degraded slightly, but not to the almost-random level when used on *non-review data*[2]. That is why we think it is sufficient for our purposes to use *movie-reviews* to classify *user-reviews*.
+
+[1] C2ER, ACCRA Cost of Living Index, Annual Average. (2010)
+[2] Customizing sentiment classifiers to new domains: A case study. (2005)
+