It took seven months for the New Zealand police to set up their first team for scanning the internet after the mosque attacks – but it was almost immediately in danger of being shut down.

An internal report released under the Official Information Act (OIA) said this was despite the team already proving its worth “many times over” in countering violent extremists.

The unit still does not have dedicated funding, despite a warning last July it risked being “turned off”.

This is revealed in 170 pages of OIA documents charting police intelligence shortcomings over the last decade, from pre-2011 extending through to mid-2020, and their attempts to overhaul the national system since 2018.

These show police had no dedicated team before 2019 to scan the internet for threats – what is called an OSINT team, for “Open Source Intelligence”.

“The OSINT team was stood up quickly last year with seconded staff to ensure… [an] appropriate emphasis on this new capability,” an internal report from July 2020 said.

In fact, police began the planning at the end of 2018, then “accelerated” it after the attacks, but it took till late October for the team to start, and training began in November 2019, a police statement to RNZ last week said.

This was all well after a January 2018 official assessment of the domestic terrorism threatscap said: “Open source reporting indicates the popularity of far right ideology has risen in the West since the early 2000s”.

When the police OSINT unit was finally set up, there was no guarantee it would last.

“This team is not permanent,” the July 2020 report said.

“This has meant uncertainty for staff and our intelligence customers.”

‘Seriously compromises’
The team had no dedicated budget, and lacked trained staff.

It also was still looking for tools to “quickly capture and categorise online intelligence elements”.

“The lack of a strong OSINT capability seriously compromises our intelligence collection posture, especially in major events,” said the report last July.

This is the sort of scanning that can pick up threats on 4chan or other extremist sites.

Despite the shortcomings, the internet team’s worth had already been proven “many times over in recent months, particularly in the counterterrorism and Countering Violent Extremism space”, the report said.

Three people have faced extremist charges in the last year or so.

‘Turned off’
An April 2019 report said police would begin recruiting for OSINT analytics and other specialists in April-May 2019.

Police had lacked a tool to search the dark web – where the truly egregious chat and trades take place on the internet – so bought one.

But last July’s report said “currently we run the risk” of OSINT “being turned off unless there is a dedicated budget”.

In a statement on Friday, police told RNZ: “The OSINT team has been funded as part of the overall allocation for intelligence since it was established.

“Maintaining this capability is a NZ Police priority, and dedicated funding is being sought as part of next year’s internal funding allocation process (note, this is funding from within Police’s existing baseline).

“Additional supplementary funding was also received in the last financial year to support the work of OSINT.”

They had known they needed the team, they said.

“Prior to March 15, New Zealand Police used some OSINT tools to support open source research of publicly available information and had identified the requirement to develop a dedicated capability.

“The development of this capability was accelerated by the events of March 15.”

‘9/11 moment’
The OIA documents show the OSINT intelligence weakness was not an isolated example.

These warned police needed to avoid “a ‘9/11’ moment” – a situation where police obtain information about a threat but do not understand it due to a failure to analyse how the dots join up, as happened to CIA and FBI before the terror attacks on New York in 2001.

The solution was to have “a complete intelligence picture”.

But the July 2020 report then laid out very clearly how police did not have this:

“Recent operational examples conclude there is no current ability to access all information in a timely and accurate manner,” it said.

“Currently there is no tool that can search across police holdings [databases] when undertaking analysis of investigations.

“We are still depending on manual searches.”

‘Locked down or invisible’
“Sources are either locked down or invisible to analysts. Our intelligence picture is consequently incomplete.”

The 31-page, July 2020 report detailed the police’s ‘Transforming Intelligence’ programme, dubbed TI21, that was begun in December 2018 and meant to be complete by this December.

It indicated the right technology would not be in place – or in some cases even identified – for 6-18 months.

As things stood, “there are many single points of failure in our intelligence system”, the report said.

Threat information was broken up into silos, without a centralised document management system or powerful enough analytic and geospatial software to connect the threats.

A section of the 2020 report detailing problems within the police’s High-Risk Targeting Teams has been mostly blanked out.

The OIA documents describe what is and is not working, especially when it comes to national security and counterterrorism, but also around intelligence on gang and drug crime, family violence, combating child sex offending, and the like, at a point many months after both the mosque attacks and the beginning of the system overhaul.

The Royal Commission of Inquiry into the mosque attacks in late 2020 called police national security intelligence capabilities “degraded” – not just once but six times.

It showed weaknesses elsewhere when it came to OSINT: The Security Intelligence Service had just one fulltime officer doing Open Source Internet searching, and the Government Communications Security Bureau had few resources for this, too. It was not till June 2019 that the Government’s Counter-Terrorism Coordination Committee suggested “leveraging open-source intelligence capability”.

Police, unlike SIS, did not do an internal review of how they had performed in the lead-up to March 15.

They did get a review done of how they did 48 hours after the attacks, which praised their efforts.

Tools missing

Among the key systems police have been lacking are:

• A national security portal “to search across police holdings”
• A national security person-of-interest tool
• A child sex offender management tool
• Cybercrime reporting systems – a “strategic demand” that “police intelligence is unable to effectively report on it”

Police in a statement said they had now “achieved a number of milestones”.

Key among them was introducing a National Security Portal to manage persons of interest.

Also, they now had standardised ways of improving quality and a National Intelligence Operating Model to ensure a consistent approach.

“The OSINT team, a new case management tool and “refined intelligence support to major events… has increased the capability, capacity and resilience of Police Intelligence to reduce and respond to counter-terrorism risks”.

The “Transforming Intelligence” documents refer repeatedly to having three new Target Development Centres set up in Auckland, Wellington and Christchurch.

However, this was jettisoned last year, while the overhaul did stick with introducing Precision Targeting Teams in August 2018, police said.

These teams aim to target “our most prolific offenders” early on “to reduce crimes such as burglary, robbery and other violent and high-volume offending”.

Pressure on
Police are plugging the holes in national intelligence while under pressure.

The volume of leads coming in had increased “considerably” since March 2019, the July 2020 report said.

“This has put increased strain on our people to manage cases of concern.”

The intelligence weaknesses have persisted under four police commissioners since the national intelligence system was set up in 2008.

Intelligence staff have been quitting at three times the average rate in the public sector, and the documents laid out urgent plans to improve career pathways and value the likes of field officers and collections staff more.

The July 2020 report said demand on workers at the Integrated Targeting and Operations Centre was “unsustainable”.

Deep-seated cultural problems across the police were recently uncovered by RNZ’s Ben Strang, whose reporting triggered an official investigation that found 40 percent of officers had been bullied or harassed.

The Transforming Intelligence 2021 programme covers 10 areas: Intelligence Operating Model, National Security, Open Source, Child Protection Offender Register, Critical Command Information, Collections, Intelligence Systems, Performance, Training and Intelligence Support to major events.

There is a stark contrast between how the police leadership described their intelligence systems, and what other documents state.

Timeline

2003

– The Government Audit Office underscores the importance of national security planning

– Police attempt to develop a national security plan deferred due to other priorities

2006

– Police appoint first national manager of intelligence – before this it was led at district level

2008

– New national intelligence model introduced, that lasts till 2019

2011

– March: Police national security intelligence review finds many gaps and recommends a slew of fixes

2014

– Police assess rightwing extremist threat nationally, the last time this happens before the end of 2018

2015

– Sept: Police review finds 2011’s shortcomings remain, recommends changes

– Police liaison officers begin work with SIS and GCSB

2018

– August: Precision Targeting Teams begin

– Nov/Dec: Police launch Transforming Intelligence overhaul, while praising the old model

2019

– March: Mosque terrorism attacks

– April: A report ramping up the intelligence overhaul celebrates the old model’s effectiveness

– Sept: Police approve high-level operating model for intelligence

– Oct: Police set up dedicated internet scanning team for first time

– Internet scanning team identifies counterterrorism threats

– Dec: Aim to set up professional development structure to reduce Intelligence staff attrition by 15 percent

2020

– National Intelligence Centre leadership team appointed

– Feb: Intelligence training plan in place; national workshops

– July: Stocktake of Intelligence overhaul finds many gaps

– Dec 2020-Dec 2021: Aim to identify new intelligence gathering and analysing tech, including a police-wide system

This article is republished under a community partnership agreement with RNZ.

Article by AsiaPacificReport.nz

By Dr Amanda Watson

In Papua New Guinea, the Post-Courier featured a front-page story with the headline “PNG tops world in ‘porn’ search” on January 17. In previous years, there have also been similar stories asserting that PNG beats all other countries when it comes to internet searches for pornography.

For any nation, this accolade would be unwelcome. As PNG prides itself on being a Christian country with strong traditional cultures and values, coupled with tough laws banning importation of pornographic magazines and movies, the headline has produced consternation.

The ruling political party in PNG has released a statement and the competing newspaper has also published a response. Both reactions argue that the Post-Courier’s front page story is inaccurate.

The front-page article included the assertion that 100 percent of all internet searches in Western Highlands Province are for the term ‘porn’. Clearly, not every internet search in that province includes this term.

So, what is going on? My blog will examine the source of the newspaper story and assess its credibility. It will also discuss internet access trends in PNG.

The source of the media reports is Google Trends. This is an interactive website run by Google, probably the world’s most popular internet search engine, which presents information about the searches that are conducted through Google.

For instance, a user can type in the word “car” and see information about how popular the search term is over time and also where it is popular, comparing regions, countries and cities.

First glance
At first glance, the site appears to suggest that 100 percent of all searches conducted using Google in the United Kingdom feature the word “car”. But this is not possible. There’s no way that all of the people in Wales, Scotland and other parts of the United Kingdom only ever use Google when they want to find out information about different kinds of motor vehicles.

Instead, the way it works is that the figures represent rankings, not percentages. The Post-Courier’s story was misleading in that it included percentage symbols alongside bar graphs. As Google Trends explains: “100 is the location with the most popularity as a fraction of total searches in that location”.

In other words, the United Kingdom had more searches during the time period for the word “car” compared to other countries, as a percentage of the total number of searches, which would also have included many other words, including “weather”, “news”, “school”, “restaurant”, “bank” and more.

Another example is the term “Highlands”. When inserted into Google Trends, bar graphs appear showing 75 for PNG. Again, this does not mean that 75 percent of the Google searches conducted by people in PNG are for this word.

Instead, it means that compared to other countries – where, for example, the term “mountains” might be more commonly used – the term “Highlands” is searched for fairly frequently in PNG.

Now, turning to the term “porn”, when looking at trends over the past five years, PNG is not listed in the top 25 countries. In fact, when the author visited the Google Trends website shortly after the Post-Courier story was published, it proved difficult to replicate the Post-Courier’s results.

I changed the time period to the past 12 months and the results revealed that once again PNG did not feature in the top 25 nations. I generated similar results for other time periods, as is shown in Table 1.

Table 1: Country rankings: Google Trends enquiries on 26 January 2017 using the term ‘porn’

Difficult to check
It’s important to note that the Post-Courier’s findings were not easy to duplicate and that in fact PNG does not feature in the top 25 listing for most time periods. Google Trends results are constantly being updated in real time and therefore it is very difficult to check or verify the Post-Courier’s story.

In addition, the tool only presents the top 25 countries – therefore it is not possible to determine a country’s actual ranking if it does not appear in the top 25.

It’s also helpful to point out that the size of a country’s population does not impact upon the ranking, as the ranking refers to the frequency of use of a word, for instance “porn”, as compared to all other words inserted into Google in that place, including “school”, “highway”, “buai”, “election”, “Highlands”, “Australia”, etc.

In other words, the word “Highlands” is used in PNG more often as a percentage of all searches, compared to the word “mountains”. It’s also worth noting that some users may have blocked their location, meaning that Google cannot tell where they are based, and this would of course make any data regarding locations of searches somewhat inaccurate.

Western Highlanders might also be curious to know how their province rates. While the Post-Courier showed a graph suggesting that the Western Highlands is the province with the most searches for the term “porn” versus other words used, compared to other provinces of PNG, the results are inconsistent.

As is shown in Table 2, Western Highlands Province (WHP) moves around the rankings a great deal, depending on the time period in question. For instance, in the past 7 days, WHP didn’t feature at all in the top ten provinces, whereas it’s in the top position when looking at the last 5 years.

When focusing on other provinces, their positions also move around a great deal. In short, the author feels that the rankings vary so much when comparing provinces in PNG as to be meaningless.

Table 2: Western Highlands Province (WHP): Google Trends enquiries on 26 January 2017 using the term ‘porn’

Note: Google Trends results are only showing for the first four provinces in the “past 30 days” time period, for the first eight provinces in the ‘past 4 hours’ category and for the first five provinces in the “past hour” time period.

Significant improvement
In the last couple of years there has been a significant improvement in the accessibility of the internet in PNG, due to mobile network upgrades and expansions, as well as availability of cheap smartphone handsets.

While most people in PNG still do not have access to electricity, many do now live within mobile network coverage. The majority of this coverage is second generation (2G) which is suitable only for voice calls and text messaging.

But around urban centres, both Digicel and bmobile Vodafone now offer third generation (3G) service, which can be used to surf the internet, correspond through email and use social media platforms such as Facebook and WhatsApp.

In Port Moresby and Lae, Digicel offers 4G service. Telikom PNG is in the process of launching a new, digital mobile phone service which will aim to compete with the other players.

All these changes have meant that a growing number of people in PNG are accessing the internet for the first time. In particular, the number of Facebook users based in PNG continues to rise. Interest in and use of Facebook is fuelled by mobile phone companies offering special promotions through which Facebook use is either free or very cheap.

Nonetheless, many people in PNG still use basic handsets and rarely access the internet, if ever.

In short, this context means that many of the internet users in PNG have only had internet access for a year or two. As people in PNG are among the latest in the world to gain access to the internet, they may be unaware of the range of activities or kinds of searches that they could undertake through this medium.

Alarmist reports not helpful
Publication of alarmist, misleading reports suggesting that online porn consumption is sky-high in PNG is not going to help to strengthen understanding about the medium or how to use it.

Having examined the recent Post-Courier article and the Google Trends website, it’s now clear that the Post-Courier article was incorrect and that PNG does not necessarily rank highly for internet porn searches.

The assertion in the newspaper’s sub-heading that “almost all Papua New Guineans look up the word ‘porn’” is not supported by the evidence. It also seems plain that any comparison of provinces within PNG is unhelpful.

Even if patterns could be determined in the Google Trends material, given limited internet access and use by most people across PNG, it would be unwise to draw conclusions regarding how provinces compare to one another.

Further research will be required to unpack whether Google Trends does convey some useful data. Academic research would also be valuable in order to learn about the internet use of groups of people in PNG.

Amanda H A Watson is a lecturer in Public Policy at the University of Papua New Guinea (UPNG), based in Port Moresby under the UPNG-ANU partnership. She is also a visiting fellow with the State, Society and Governance in Melanesia Program at the Australian National University (ANU). This article was first published on the Development Policy Centre’s blog DevNet and is republished here with permission.