By Sarah N. Lynch and Mark Hosenball

WASHINGTON/LONDON (Reuters) – The pattern is clear: Hate-filled manifestos posted on websites populated by white supremacists, followed by gun attacks against blacks, Jews, Muslims, or Latin American immigrants.

In some cases, the killers use their internet posts to praise previous attacks by other white nationalists. And after new assaults, the manifestos get passed around, feeding the cycle of propaganda and violence.

Following the racially-motivated attack that killed 22 people at a Walmart store in El Paso, Texas, President Donald Trump said he wants police to do more to stop extremists who are active online before they can turn to murder.

But identifying and stopping the extremists who plan to launch an attack is much easier said than done.

Law enforcement experts say that the constitutional right of free speech means police cannot arrest someone simply on the basis of extremist rants online, unless they make a specific threat.

“You couldn’t just open a case on the words,” said Dave Gomez, a retired FBI agent who has worked on cases of both international and domestic terrorism.

“Posting something like that on the internet doesn’t harm anybody,” he said, adding that police can only successfully investigate a white supremacist when you can “connect his words to an overt act.”

The White House will discuss violent extremism online with representatives from a number of internet and technology companies on Friday, according to a White House spokesman.

Social media companies are reluctant to spy on or censor their users, though increasingly they are responding to demands that they take down obvious incitements to violence. And civil rights groups warn that tighter monitoring can lead to unconstitutional abuses of power

Another former FBI agent, who asked not to be identified, said closer monitoring of extremists’ websites would anyway be unlikely to prevent new mass shootings.

“There is not enough manpower. There is not enough technology to properly monitor the internet,” he said. “This is the number one thing we always say in law enforcement: ‘You can’t stop crazy. You can’t even predict crazy.’”

Trump said after the mass shootings last weekend in El Paso and Dayton, Ohio, that he would ask the Justice Department to work with local, state and federal agencies as well as social media companies “to develop tools that can detect mass shooters before they strike.”

Even before those attacks, The FBI in early July requested bids for a contractor to help it detect national security threats by trawling through social media sites.

“The use of social media platforms by terrorist groups, domestic threats, foreign intelligence services, and criminal organizations to further their illegal activity creates a demonstrated need for tools to properly identify the activity and react appropriately,” the FBI said in its request.

PRESSURE

Top law enforcement and domestic security officials from the United States, the United Kingdom, Canada, Australia and New Zealand met with leading social media and internet companies in London last week, and pushed them to help authorities track suspicious users.

The government officials noted in an agenda paper for the meeting that some companies “deliberately design their systems in a way that precludes any form of access to content, even in cases of the most serious crimes.”

“Tech companies should include mechanisms in the design of their encrypted products and services whereby governments, acting with appropriate legal authority, can obtain access to data in a readable and usable format,” the agenda paper said.

A final statement from the meeting said little about encryption, however, and neither company nor government officials talked about what was discussed.

Facebook and Microsoft confirmed they attended but Google, which was invited, did not respond to a request for comment. Other attendees included Roblox, Snap and Twitter, the statement said.

FBI agents say that broad surveillance powers enacted by Congress in the wake of the Sept., 11, 2001 attacks helped them track international terrorist groups and stop people with links to foreign groups like al Qaeda and Islamic State before they could carry out crimes.

But they key law criminalizing “material support” for terrorism does not apply to investigations or prosecutions of domestic terrorists, such as violent white supremacists, that commit hate crimes.

This week, the FBI Agents Association called on Congress to make domestic terrorism a federal crime in order to give agents more tools.

The Electronic Frontier Foundation, which promotes internet civil liberties, said the sheer amount of users posting aggressive content online makes it almost impossible to identify and track the people who pose an actual threat.

“Even though it seems like there is another mass shooting every week, if you are looking at the number of mass shooters versus the total population, it’s still a tiny, tiny number which means this is still a very rare event,” said Jeremy Gillula, the group’s tech products director. “It’s like trying to predict where lightning is going to strike.”

(Reporting by Sarah N. Lynch and Mark Hosenball; Editing by Alistair Bell)