Recent Development of Mosaic Theory in Taiwan’s Privacy Law

Regular readers of this column may agree that, generally speaking, Taiwan’s judges are open to new legal theories bred in more advanced jurisdictions, especially when it comes to tech laws.  A recent example is the transplant of the “mosaic theory.”

First arising in a 2010 decision by the US Court of Appeals for the D.C. Circuit (and later reviewed by the US Supreme Court under the name United States v. Jones), the mosaic theory renders a novel interpretation (but not without criticism) of the Fourth Amendment of the US Constitution, as it “considers whether a set of nonsearches aggregated together amount to a search because their collection and subsequent analysis creates a revealing mosaic.”Orin S. Kerr, The Mosaic Theory of the Fourth Amendment, 111 MICH. L. REV. 311 (2012) (available at: 

Introduced into Taiwan soon afterwards, the term “mosaic theory” made its first presence in Taiwan’s court decision in June 2015. Taipei District Prosecutors Office v. Chang, 104 Shang-Yi 352, Taiwan High Court (June 2015).  So far, the theory has been discussed (and mostly affirmed) in no less than 12 decisions, or 9 cases, all of them arising from misuse of GPS tracking devices as Jones did.  However, from the very start of the transplant, local mutation has occurred.  Of the 9 cases, only one involved disputed government search.  In all the other cases, including the first one, the defendants were private individuals, some of them being detectives or their clients.  More, the focus of all 9 cases was whether the victim’s privacy was infringed to give rise to criminal or tortious liability, rather than whether a due process rule on search was violated by a law enforcement officer.  Simply put, in Taiwan, the mosaic theory is basically adduced to support criminal charges of privacy violation triggered by GPS stalking. 

In fact, the mosaic theory plays as a critical piece in criminal cases against GPS stalkers.  That is because, in the absence of an anti-stalking act, the plaintiff or prosecutor resorts to Taiwan’s Criminal Code, in particular its Article 315-1, but many GPS stalking cases cannot fit in the wording of this Article without the aid of the mosaic theory.  Under Article 315-1 of the Criminal Code, one who “uses instruments or equipment without a justifiable cause to peep at or eavesdrop on other’s non-public activity, speeches, talks, or the private part of the body…” shall be punishable with monetary fines or imprisonment of up to three years (italics added).  However, without the “revealing mosaic” effect, locational data obtained from GPS surveillance usually do not instantly link to or reveal a non-public activity of the person under surveillance. 

Despite the years of unperturbed evolution in the local soil, the Taiwanese version of the mosaic theory recently encountered its first setback in court, and it is Taiwan’s Supreme Court.  Taipei District Prosecutors Office v. Lin, 108 Tai Shang 1750, Taiwan’s Supreme Court (January 2020).  By tightening the requirement of “non-public activity,” the Supreme Court’s holding in Lin curtailed the space in which the mosaic theory used to thrive in Article 315-1 cases. 

In its essence, Lin arose out of a domestic discord.  A GPS tracking device was installed on the wife’s car to send locational data of the car to a private investigator allegedly hired by the husband.  Together with this was a series of paparazzi style stalking and covert photographing at restaurants, hotels and other venues.  Before any truly revealing matters had been collected (or presented to the wife), the wife discovered the plot.  She filed a criminal action against the husband and his team based on an Article 315-1 charge, and won all the way up to the second instance. 

The second instance decision rendered by the Taiwan High Court provides a good illustration of how the mosaic theory is adapted to apply to a privacy dispute in Taiwan.  Its holdings in this aspect are worth quoting in full length: 

As the text of this Article does not include a restriction on the basis of “venue,” we cannot preclude its application to privacy intrusion/interference conduct that occurs in a public venue.  Generally speaking, we should consider that vehicle drivers and passengers expect to keep confidential their locational information linked with the vehicle traces, unless they have explicitly abandoned this sort of privacy, such as in the case where a bus or taxi bears on its outside a driver’s name plate, or a security officer drives a company-provided car equipped with a GPS tracker to perform his duty.  If a vehicle moves along a public road not in a manner that attracts notice, and if there is no public figure or public interest involved which brings the public the need to be informed, we should hold that the driver and passengers do not want their whereabouts to be disclosed.  More, as identity information of a vehicle user can be concealed in a car or by a helmet [in the case of motorbikes], … there is no need for the users to take other measures to safeguard the secrecy of their activities. 

However, by using a GPS tracker, one can monitor a vehicle’s travel paths and parking locations 24 hours a day and 7 days a week, and consequently grasp the user’s whereabouts in the utmost details and on a long-term basis.  As trivial and insignificant as such information may seem to be, when a “trawl net surveillance” is applied to collect and analyze locational data in great amount, it will, in a wider perspective, reveal the internal connections of separate activities, and the whereabouts data will unwantedly yield an unknown image of a person’s private life.  [That is to say,] through long-time and intensive analysis of data on vehicle traces, one can obtain a panoptic view of the driver’s usual route, driving speed, parking spots, and stay time, hence detecting the vehicle user’s daily routines, life details and behavioral patterns. … Even though a target vehicle user’s privacy is not physically interfered, her psychological privacy has been destroyed.  As such, this constitutes the criminal offense at issue. 

The above holding goes in line with the “mosaic theory” established by US courts in cases of similar nature.  Just like seemingly trivial and insignificant tiles when arranged together in a pattern can present a vast image, great amount of information, being an aggregated whole, can bring forth severe harm to privacy, even though the victim does not instantly perceive the harm.  For this reason, vehicle users should have a reasonable expectation of privacy to be free from long-time and persistent collection and recording of their vehicle traces.  

However, for all the second-instance court’s emphases on the harm to privacy caused by long-term GPS vehicle surveillance, the Supreme Court revoked the lower court’s decision and remanded the case for further review.  The Supreme Court reminded the parties that:

A “non-public activity” means an activity for which the actor has an expectation or intent to conduct in secrecy and free from public disclosure (hence the prerequisite “subject expectation of privacy”) and has made use of the surroundings or appropriate equipment to adequately ensure the private nature of the activity (hence the prerequisite “object private environment”).  Such activities include, for example, dressing/undressing in a private dwelling, using of a public toilet, singing in a rented karaoke box, negotiating in a hotel room, or sleeping in a camping tent, etc., each with an expectation of privacy… 

Nevertheless, in the box “content of the recorded activities” of the table appended to the lower court’s judgment, there is scarcely any meaningful information. Other than a mere indication that the private investigator took certain photos of the victim’s “non-public activities” in a room, a company’s office, or a taxi, the lower court’s decision did not set out any details of the specific content of the alleged “non-public activities” … (italics added)

It seems to us that [the lower court’s holdings] merely addressed the scope of privacy protection accorded by a private law [rather than a public law or a criminal law].  Yet it failed to elucidate … on what ground it found that the recorded content had reflected the victim’s “non-public activity” that met the above-defined prerequisites. 

While the mosaic theory was not rejected by the Supreme Court, a long shadow has been cast over its prospect in a Criminal Code case.  After all, how could locational data of a vehicle add up to a disclosure of privacy in a case where photos taken at places unveiled by the locational data have been questioned about their capability of revealing non-public activities?  The same goes with the Supreme Court’s emphasis on the necessity of specifying the details of the non-public activities.  Just how specific and detailed does a “revealing mosaic” need to be — will a depiction of behavioral pattern or a way of life be good enough?  It might take quite some time for a clear answer to emerge.