Research programme
What we are trying to find out
Our programme is deliberately narrow. Rather than survey content moderation in general, we ask a small number of answerable questions about state-originated removal demands concerning Indonesia, and we answer them from primary records. Narrow questions produce checkable answers; broad ones produce essays.
Research questions
-
How many removal demands concerning Indonesia originate with the state?
Counting government requests, law-enforcement requests, and court orders in the public record, distinguished from the far larger volume of private copyright complaints, and reported as a documented floor rather than a total.
-
Which bodies are the recurring senders?
Whether demands concentrate in a small number of ministries, agencies, and courts, or are distributed widely across the state apparatus.
-
Which legal provisions are invoked?
Which statutes and articles appear as the stated basis for removal, how often UU ITE features among them, and which of its provisions recur.
-
What kinds of material are targeted?
The subject matter described in the requests themselves — with particular attention to demands touching journalism, political criticism, and civil society speech.
-
How do recipients respond?
Whether platforms record full compliance, partial compliance, or no action, and whether the pattern differs by sending body, legal basis, or period.
Studies
A baseline of state removal requests concerning Indonesia
Before anyone can argue about trends, someone has to establish what the public record contains. This study assembles a systematic count of state-originated removal requests concerning Indonesia held in public notice archives, broken down by requesting body, legal basis, and recorded outcome, with the coverage gaps documented as carefully as the counts.
- Answers
- Questions 1, 2, and 5.
- Evidence
- Notices of type
GovernmentRequest,LawEnforcementRequest, andCourtOrderin the Lumen Database, cross-checked against platform transparency reports covering Indonesia. - Method
- Date-windowed extraction, entity-name normalisation, and manual review of a sample to estimate classification error. Full detail on the methodology page.
- Period covered
- To be fixed once we have surveyed how far back the record offers usable coverage. The exact window will be stated with the output rather than chosen in advance to suit it.
- Intended output
- Public analysis article with parameters, normalisation tables, and code published alongside it. The publication date will be announced on the publications page as soon as it is set.
Which UU ITE provisions appear most often in removal demands
Requests that cite domestic law reveal which statutory provisions are used as instruments against online material. This study reads the legal citations recorded in court orders and law-enforcement requests to establish which articles recur, and against which categories of content.
- Answers
- Questions 3 and 4.
- Evidence
- The
laws_referencedfield on court orders and theregulationsfield on law-enforcement requests, read together with the recorded subject of each demand. - Method
- Citation extraction with a published coding scheme, double-coded by two members with disagreements recorded rather than silently resolved.
- Intended output
- Public analysis article, sequenced after the baseline study.
Demands touching journalism and civil society speech
A closer reading of the subset of demands whose recorded subject concerns reporting, political criticism, or the work of civil society organisations. This is the subset where the public interest in documentation is highest and where the ethical care required is greatest, which is why it is sequenced last rather than first.
- Answers
- Question 4, in depth.
- Evidence
- A qualitatively coded subset of the baseline corpus, with individuals never named unless they have already self-identified publicly and the public interest is clear.
- Intended output
- Public analysis article. Scope and ethical review to be settled before work begins.
Why this order
The baseline comes first because every later claim depends on it. Without a documented denominator, statements about which provisions dominate or which speech is targeted have nothing to be a proportion of. The most sensitive study comes last, once the corpus is understood and our handling procedures have been tested on less delicate material.
We would rather publish a narrow, checkable count on schedule than a sweeping account late. Where a study proves unanswerable with the available records, we intend to publish that finding too, because a documented dead end saves the next researcher the same months.