Alcatraz first operated as a prison in 1859, when the military fort first held convicted soldiers. The prison technology of the time was simple, consisting of little more than a basement room with a trap-door entrance. Only small numbers of prisoners were held in this period, but it established Alcatraz as a center of incarceration. Later, the Civil War triggered construction of a "political prison," a term with fewer negative connotations at the time, for confederate sympathizers.

This prison was more purpose-built (although actually a modification of an existing shop), but it was small and not designed for an especially high security level. It presaged, though, a much larger construction project to come.

Alcatraz had several properties that made it an attractive prison. First, it had seen heavy military construction as a Civil War defensive facility, but just decades later improvements in artillery made its fortifications obsolete. That left Alcatraz surplus property, a complete military installation available for new use. Second, Alcatraz was formidable. The small island was made up of steep rock walls, and it was miles from shore in a bay known for its strong currents. Escape, even for prisoners who had seized control of the island, would be exceptionally difficult.

These advantages were also limitations. Alcatraz was isolated and difficult to support, requiring a substantial roster of military personnel to ferry supplies back and forth. There were no connections to the mainland, requiring on-site power and water plants. Corrosive sea spray, sent over the island by the Bay's strong winds, lay perpetual siege on the island. Buildings needed constant maintenance, rust covered everything. Alcatraz was not just a famous prison, it was a particularly complicated one.

In 1909, Alcatraz lost its previous defensive role and pivoted entirely to military prison. The Citadel, a hardened barracks building dating to the original fortifications, was partially demolished. On top of it, a new cellblock was built. This was a purpose-built prison, designed to house several hundred inmates under high security conditions.

Unfortunately, few records seem to survive from the construction and operation of the cellblock as a disciplinary barracks. At some point, a manual telephone exchange was installed to provide service between buildings on the island. I only really know that because it was recorded as being removed later on. Communications to and from Alcatraz were a challenge. Radio and even light signals were used to convey messages between the island and other military installations on the bay. There was a constant struggle to maintain cables.

Early efforts to lay cables in the bay were less about communications and more about triggering. Starting in 1883, the Army Corps of Engineers began the installation of "torpedoes" in the San Francisco bay. These were different from what we think of as torpedoes today, they were essentially remotely-operated mines. Each device floated in the water by its own buoyancy, anchored to the bottom by a cable that then ran to shore. An electrical signal sent down the cable detonated the torpedo. The system was intended primarily to protect the bay from submarines, a new threat that often required technically complex defenses.

Submarines are, of course, difficult to spot. To make the torpedoes effective, the Army had to devise a targeting system. Observation posts on each side of the Golden Gate made sightings of possible submarines and reported them to a control post, where they were plotted on the map. With a threat confirmed, the control post would begin to detonate nearby torpedoes. A second set of observation posts, and a second line of torpedoes, were located further into the bay to address any submarines that made it through the first barrage.

By 1891, there were three such control points in total: Fort Mason, Angel Island, and Yerba Buena. The rather florid San Francisco Examiner of the day described the control point at Fort Mason, a "chamber of death and destruction" in a tunnel twenty feet underground. The Army "death-dealers" that manned the plotting table in that bunker had access to a board that "greatly resemble[d] the switch board in the great operating rooms of the telephone companies." By cords and buttons, they could select chains of mines and send the signal to fire.

NPS historians found that a torpedo control point had been planned at Alcatraz, and one of the fortifications modified to accommodate it, but never seems to have been used. The 1891 article gives a hint of the reason, noting that the line from Alcatraz to Fort Mason was "favorable for a line of torpedoes" but that currents were so strong that it was difficult to keep them anchored. Perhaps this problem was discovered after construction was already underway.

Somewhere around 1887-1888, the Army Signal Corps had joined the cable-laying fray. A telegraph cable was constructed from the Presidio to Alcatraz, and provided good service except for the many times that it was drug up by anchors and severed. This was a tremendous problem: in 1898, Gen. A. W. Greely of the Signal Corps called San Francisco the "worst bay in the country" for cable laying and said that no cable across the Golden Gate had lasted more than three years. The General attributed the problem mainly to the heavy shipping traffic, but I suspect that the notorious currents must have been a factor in just how many anchors were dragged through cables [1].

In 1889, a brand new Army telegraph cable was announced, one that would run from Alcatraz to Angel Island, and then from Angel Island to Marin County. An existing commercial cable crossed the Golden Gate, providing a connection all the way to the Presidio.

The many failures of Alcatraz cables makes it difficult to keep track. For example, a cable from Fort Mason to Alcatraz Island was apparently laid in 1891---but a few years later, it was lamented that Alcatraz's only cable connection to Fort Mason was indirect, via the 1889 Angel Island cable. Presumably the 1891 cable was damaged at some point and not replaced, but that event doesn't seem to have made the papers (or at least my search results!).

In 1900, a Signal Corps officer on Angel Island made a routine check of the cable to Alcatraz, finding it in good working order---but noticing that a "four masted schooner... in direct line with the cable" seemed to be in trouble just off the island and was being assisted by a tug. That evening, the officer returned to the cable landing box to find the ship gone... along with the cable. A French ship, "Lamoriciere," had drifted from anchor overnight. A Signal Corps sergeant, apparently having spoken with harbor officials, reported that the ship would have run completely aground had the anchor not caught the Alcatraz cable and pulled it taught. Of course the efforts of the tug to free Lamoriciere seems to have freed a little more than intended, and the cable was broken away from its landing. "Its end has been carried into the bay and probably quite a distance from land," the Signal Corps reported.

This ongoing struggle, of laying new cables to Alcatraz and then seeing them dragged away a few years later, has dogged the island basically to the modern day---when we have finally just given up. Today, as during many points in its history, Alcatraz must generate its own power and communicate with the mainland via radio.

When the Bureau of Prisons took control of Alcatraz in 1933, they installed entirely new radio systems. A marine AM radio was used to reach the Coast Guard, their main point of contact in any emergency. Another radio was used to contact "Alcatraz Landing" from which BOP ferries sailed, and over the years several radios were installed to permit direct communications with military installations and police departments around the Bay Area.

At some point, equipment was made available to connect telephone calls to the island. I'm not sure if this was manual patching by BOP or Coast Guard radio operators, or if a contract was made with PT&T to provide telephone service by radio. Such an arrangement seems to have been in place by 1937, when an unexplained distress call from the island made the warden impossible to contact (by the press or Bureau of Prisons) because "all lines [were] tied up."

Unfortunately I have not been able to find much on the radiotelephone arrangements. The BOP, no doubt concerned about security, did not follow the Army's habit of announcing new construction projects to the press. Fortunately, the BOP-era history of Alcatraz is much better covered by modern NPS documentation than the Army era (presumably because the more recent closure of the BOP prison meant that much of the original documentation was archived). Unfortunately, the NPS reports are mostly concerned with the history of the structures on the island and do not pay much attention to outside communications or the infrastructure that supported it.

Internal arrangements on the island almost completely changed when the BOP took over. The Army had left Alcatraz in a degree of disrepair (discussions about closing it having started by at least 1913), and besides, the BOP intended to provide a much higher level of security than the Army had. Extensive renovations were made of the main cellblock and many supporting buildings from 1933 to about 1939.

The 1930s had seen a great deal of innovation in technical security. Technologies like electrostatic and microwave motion sensors were available in early forms. On Alcatraz, though, the island was small and buildings tightly spaced. The prison staff, and in some cases their families, would be housed on the island just a stones throw from the cellblock. That meant there would be quite a few people moving around exterior to the prison, ruling out motion sensors as a means of escape detection. Exterior security would instead be provided by guard and dog patrols.

There was still some cutting-edge technical security when Alcatraz opened, including early metal detectors. At first, the BOP contracted the Teletouch Corporation of New York City. Teletouch, a manufactured burglar alarms and other electronic devices, was owned by or at least affiliated with famed electromagnetics inventor and Soviet spy Leon Theremin. Besides the instrument we remember him for today, Theremin had invented a number of devices for security applications, and the metal detectors were probably of his design. In practice, the Teletouch machines proved unsatisfactory. They were later replaced with machines made by Forewarn. I believe the metal detector on display today is one of the Forewarn products, although the NPS documents are a little unclear on this.

Sensitive common areas like the mess hall, kitchen, and sallyport wre fitted with electrically-activated teargas canisters. Originally, the mess hall teargas was controlled by a set of toggle switches in a corner gun gallery, while the sallyport teargas was controlled from the armory. While the teargas system was never used, it was probably the most radical of Alcatraz's technical security measures. As more electronic systems were installed, the armory, with its hardened vault entrance and gun issue window, served as a de facto control center for Alcatraz's initial security systems.

The Army's small manual telephone switchboard was considered unsuitable for the prison's use. The telephone system provided communication between the guards, making it a critical part of the overall security measures, and the BOP specified that all equipment and cabling needed to be better secured from any access by prisoners. Modifications to the cellblock building's entrance created a new room, just to the side of the sallyport, that housed a 100-line automatic exchange. Automatic Electric telephones that appear throughout historic photos of the prison would suggest that this exchange had been built by AE.

Besides providing dial service between prison offices and the many other structures on the island, the exchange was equipped with a conference circuit that included annunciator panels in each of the prison's main offices. Assuming this was the type provided by Automatic Electric, it provided an emergency communications system in which the guard telephones could ring all of the office and guard phones simultaneously, even interrupting calls already in progress. Annunciator panels in the armory and offices showed which phone had started the emergency conference, and which phones had picked up. From the armory, a siren on the building roof could be sounded to alert the entire island to any attempted escape.

Some locations, including the armory and the warden's office, were also fitted with fire annunciators. I am less clear on this system. Fire circuits similar to the previously described conference circuit (and sometimes called "crash alarms" after their use on airfields) were an optional feature on telephone exchanges of the time. Crash alarms were usually activated by dedicated "hotline" phones, and mentions of "emergency phones" in various prison locations support that this system worked the same way. Indeed, 1950s and 60s photos show a red phone alongside other telephones in several prison locations. The fire annunciator panels probably would have indicated which of the emergency phones had been lifted to initiate the alarm.

One of the most fascinating parts of Alcatraz, to a person like me, is the prison doors. Prison doors have a long history, one that is interrelated with but largely distinct from other forms of physical security. Take a look, for example, at the keys used in prisons. Prisons of the era, and even many today, rely on lever locks manufactured by specialty companies like Folger Adams and Sargent and Greenleaf. These locks are prized for their durability, and that extends to the keys, huge brass plates that could hold up to daily wear well beyond most locks.

At Alcatraz, the first warden adopted a "sterile area" model in which areas accessible to prisoners should be kept as clear as possible of dangerous items like guns and keys. Guards on the cellblock carried no keys, and cell doors lacked traditional locks. Instead, the cell doors were operated by a central mechanical system designed by Stewart Iron Works.

To let prisoners out of cells in the morning, a guard in the elevated gun gallery passed keys to a cellblock guard in a bucket or on a string. The guard unlocked the cabinet of a cell row's control system, revealing a set of large levers. The design is quite ingenious: by purely mechanical means, the guard could select individual cells or the entire row to be unlocked, and then by throwing the largest lever the guard could pull the cell doors open---after returning the necessary key to the gun gallery above. This 1934 system represents a major innovation in centralized access control, designed specifically for Alcatraz.

Stewart Iron Works is still in business, although not building prison doors. Some years ago, the company assisted NPS's work to restore the locking system to its original function. The present-day CEO provided replicas of the original Stewart logo plate for the restored locking cabinets. Interviewing him about the restoration work, the San Francisco Chronicle wrote that "Alcatraz, he believes, is part of the American experience."

The Stewart mechanical system seems to have remained in use on the B and C blocks until the prison closed, but the D block was either originally fitted, or later upgraded, with electrically locked cell doors. These were controlled from a set of switches in the gun gallery.

In 1960, the BOP launched another wave of renovations on Alcatraz, mostly to modernize its access and security arrangements to modern standards. The telephone exchange was moved away from the sallyport to an upper floor of the administration building, freeing up its original space for a new control center. This is the modern sallyport control area that visitors look into through the ballistic windows; the old service windows and viewports into the armory anteroom that had been the de facto control center are now removed.

This control center is more typical of what you will see in modern prisons. Through large windows, guards observed the sallyport and visitor areas and controlled the electrically operated main gates. An electrical interlock prevented opening the full path from the cellblock to the outside, creating a mantrap in the visitor area through which the guards in the control room could identify everyone entering and leaving.

Photos from the 1960 control room, and other parts of the prison around the same time, clearly show consoles for a Western Electric 507B PBX. The 507B is really a manual exchange, although it used keys rather than the more traditional plugboard for a more modern look. It dates back to about 1929---so I assume the 507B had been installed well before the 1960 renovation, and its appearance then is just a bias of more and better photos available from the prison's later days.

Fortunately, the NPS Historic Furnishings Report for the cellblock building includes a complete copy of a 1960s memo describing the layout and requirements for the control center. We're fortunate to get such a detailed listing of the equipment:

• Four phones (these are Automatic Electric instruments, based on the photo). One is a fire reporting phone (presumably on the exchange's "crash alarm" circuit), one is the watch call reporting phone (detailed in a moment), a regular outgoing call telephone, and an "executive right of way" phone that I assume will disconnect other calls from the outgoing trunks.
• The 507B PBX switchboard
• An intercom for communication with each of the guard towers
• Controls for five electrically operated doors
• Intercoms to each of the electrically operated doors (many of these are right outside of the control center, but the glass is very thick and you would not otherwise be able to converse)
• An "annunciator panel for the interior telephone system" which presumably combines the conference circuit, fire circuit, and watch call annunciators.
• An intercom to the visitor registration area
• A "paging intercom for group control purposes." I don't really know what that is, possibly it is for the public address speakers installed in many parts of the cellblock.
• Monitor speaker for the inmate radio system. This presumably allowed the control center to check the operation of the two-channel wired radio system installed in the cells.
• The "watch call answering device," discussed later.
• An indicator panel that shows any open doors in the D cell block (which is the higher security unit and the only one equipped with electrically locking cell doors).
• Two-way radio remote console
• Tear gas controls

Many of these are things we are already familiar with, but the watch call telephone system deserves some more discussion. It was clearly present back in the 1930s, but it wasn't clear to me what it actually did. Fortunately this memo gives some details on the operation.

Guards calling in to report their watch call extension 3331. This connects to the watch call answering device in the control center, which when enabled, automatically answers the call during the first ring. The answering device then allows a guard anywhere in the control center to converse with the caller via a loudspeaker and microphone. So, the watch call system is essentially just a speaker phone. This approach is probably a holdover from the 1930s system (older documents mention a watch call phone as well), and that would have been the early days for speakerphones, making it a somewhat specialized device. Clearly it made these routine watch calls a lot more convenient for the control center, especially since the guard there didn't even have to do anything to answer.

It might be useful to mention why this kind of system was used: I have never found any mention of two-way radios used on Alcatraz, and that's not surprising. Portable two-way radios were a nascent technology even in the 1960s---the handheld radio had basically been invented for the Second World War, and it took years for them to come down in size and price. If Alcatraz ever did issue radios to guards, it probably would have been in the last decade of operation. Instead, telephones were provided at enough places in the facility that guards could report their watch tour and any important events by finding a phone and calling the control center.

Guards were probably required to report their location at various points as they patrolled, so the control center would receive quite a few calls that were just a guard saying where they were---to be written down in a log by a control room guard, who no doubt appreciated not having to walk to a phone to hear these reports. This provided both the functions of a "guard tour" system, ensuring that guards were actually performing their rounds, and improved the safety of guards by making it likely that the control center would notice fairly promptly that they had stopped reporting in.

Alcatraz closed as a BOP prison in 1963, and after a surprising number of twists and turns ranging from plans to develop a shopping center to occupation by the Indians of All Tribes, Alcatraz opened to tourists. Most technology past this point might not be considered "historic," having been installed by NPS for operational purposes. I can't help but mention, though, that there were more attempts at a cable. For the NPS, operating the power plant at Alcatraz was a significant expense that they would much rather save.

The idea of a buried power cable isn't new. I have seen references, although no solid documentation, that the BOP laid a power cable in 1934. They built a new power plant in 1939 and operated it for the rest of the life of the prison, so either that cable failed and was never replaced, or it never existed at all...

I should take a moment here to mention that LLM-generated "AI slop" has become a pervasive and unavoidable problem around any "hot SEO topic" like tourism. Unfortunately the history of tourist sites like Alcatraz has become more and more difficult to learn as websites with well-researched history are displaced in search results by SEO spam---articles that often contains confident but unsourced and often incorrect information. This has always been a problem but it has increased by orders of magnitude over the last couple of years, and it seems that the LLM-generated articles are more likely to contain details that are outright made up than the older human-generated kind. It's really depressing. That's basically all I have to say about it.

It seems that a power cable was installed to Alcatraz sometime in the 1960s but failed by about 1971. I'm a little skeptical of that because that was the era in which it was surplus GSA property, making such a large investment an odd choice, so maybe the 1980s article with that detail is wrong or confusing power with one of the several telephone cables that seem to have been laid (and failed) during BOP operations). In any case, in late 1980 or early 1981, Paul F. Pugh and Associates of Oakland designed a novel type of underwater power cable for the NPS. It was expected to provide power to Alcatraz at much reduced cost compared to more traditional underwater power cable technologies. It never even made it to day 1: after the cable was laid, but before commissioning, some failure caused a large span of it to float to the surface. The cable was evidently not repairable, and it was pulled back to shore.

'I don't know where we go from here,' William J. Whalen, superintendent of the Golden Gate National Recreation Area, said after the broken cable was hauled in.

We do know now: where the NPS went from there was decades of operating two diesel generators on the island, until a 2017 DoE-sponsored project that installed solar panels on the cellblock building roof. The panels were intentionally installed such that they are not visible anywhere from the ground, preserving the historic integrity of the site. In aerial photos, though, they give Alcatraz a curiously modern look. The DoE calls the project, which incorporates battery storage and backup diesel generators, as "one of the largest microgrids in the United States." That is an interesting framing, one that emphasizes the modern valance of "microgrid," since Alcatraz had been a self-sufficient electrical system since the island's first electric lights. But what's old is, apparently, new again.

I originally wrote much of this as part of a larger travelogue on my most recent trip to Alcatraz, which was coincidentally the same day as a visit by Pam Bondi and Doug Burgum to "survey" the prison for potential reopening. That piece became long and unwieldy, so I am breaking it up into more focused articles---this one on the technical history, a travelogue about the experience of visiting the island in this political context and its history as a symbol of justice and retribution, and probably a third piece on the way that the NPS interprets the site today. I am pitching the travelogue itself to other publications so it may not have a clear fate for a while, but if it doesn't appear here I'll let you know where. In any case there probably will be a loose part two to look forward to.

[1] Greely had a rather illustrious Army career. His term as chief of the Signal Corps was something of a retirement after he led several arctic expeditions, the topic of his numerous popular books and articles. He received the Medal of Honor shortly before his death in 1935.

Nebraska Gov. Jim Pillen meets with troops at the southern border in Texas. (Courtesy of the Governor's Office)

LINCOLN — The media spectacle of Nebraska Gov. Jim Pillen unveiling plans for a new Immigration and Customs Enforcement detention facility in the southwestern part of the state follows a political playbook from the Trump administration.

President Donald Trump’s Department of Homeland Security touted the new facility, saying it would help ICE agents “remove the worst of the worst” — despite the governor and state agency heads explaining it would house only low-level and medium-risk immigration-related detainees.

Speaking inside McCook’s Ben Nelson Regional Airport as protesters gathered outside, Pillen said he was “proud to be a part of Trump’s Team” in an effort to “make sure we keep our community safe. He pointed to a previous arrest of an MS-13 “kingpin” in Omaha as a justification for the ICE partnership.

“This stuff hits close to home and hits every corner of this state and country,” Pillen repeated in a Wednesday evening post on X. “Government’s most important job is to keep us safe.”

Dona-Gene Barton, a political science professor who studies political behavior at the University of Nebraska-Lincoln, said it was no surprise Nebraska would want to join Trump’s expansion of ICE detention efforts as the agency tries to meet its goal of increasing deportations.

“This is a Republican governor showing support for a Republican president’s agenda,” Barton said.

Trump, throughout his 2024 presidential campaign, promised voters what he called “the single largest Mass Deportation Program in History,” after the Biden administration saw illegal border crossings spike at the end of 2023 but start decreasing last year.  

ICE is poised to become the highest-funded federal law enforcement agency because of the recently passed “One Big Beautiful Bill” that pays for much of Trump’s domestic agenda. Roughly $45 billion in the law was set aside to build, rent and staff new centers to detain immigrants, like the one Nebraska plans to operate for ICE at the state prison system’s McCook Work Ethic Camp. 

Other funding would be allocated toward hiring 10,000 ICE officers within five years, providing retention bonuses, covering the transportation of immigrants, upgrading ICE facilities, detaining families and hiring ICE immigration lawyers for enforcement and removal proceedings in immigration court.

Using Nebraska’s McCook facility, a medium-security prison, is part of a detention expansion strategy aimed at reviving dormant prisons, repurposing military bases and securing partnerships with private prison contractors, local sheriffs and GOP governors to house a record number of detainees, The Washington Post reported. 

The nationwide ICE expansions come as recent polling suggests a majority of Americans now disapprove of Trump’s handling of immigration, though most Republicans still support it. Recent polling indicates a shift in people’s attitudes from last year, when more Americans supported less immigration and stricter immigration enforcement. 

The state also experienced an ICE raid on Glenn Valley Foods in Omaha earlier this year, which sparked protests in the state’s largest city and in Lincoln.

Several local Democrats, advocacy groups, and everyday Nebraskans have raised concerns about the federally dubbed “Cornhusker Clink,” calling it a “​​harmful, dangerous and rapid expansion.”

The Nebraska Democratic Party said the facility is “yet another ‘bend the knee’ moment by top Republicans in Nebraska.” It said Republicans promised to “go after criminals” but “instead locked up hardworking” people who contribute to the state’s agricultural economy.

State Sen. George Dungan of Lincoln, a Democrat in the officially nonpartisan Legislature, told the Examiner he had questions about whether the facility would require the state to provide funding and whether an ICE detention center is needed in the state.

“For [Pillen] to be showing off to the federal administration in this way just seems out of touch with everyday Nebraskans,” Dungan said. “All of it is because we have an administration that’s failed to actually deal with any kind of real immigration reform.” 

Most Republican state senators reached by the Examiner have said they are on board with the new effort in Nebraska. State Sen. Loren Lippincott of Central City, a conservative who has carried bills Trump sought, including efforts to change how Nebraska counts its Electoral College votes for president, said he fully supports the McCook facility “as Nebraskans in service to a safer America.” 

State Sen. Merv Riepe of Ralston, often a conservative swing vote in the Legislature, told the Examiner, “If Nebraska has the capacity to help out, I don’t think that we should resist … as long as it’s a good and fair and legal process.”

The Nebraska Republican Party said the announcement “demonstrates the seriousness of the crisis and the need for leadership that prioritizes Nebraska families.”

Pillen’s office said any “backlash to this initiative is politically motivated.”

“It’s simple. Those who are opposed to Nebraska doing its fair share in securing our nation’s border want to relive the failed Biden-era open-border policies,” said Laura Strimple, a Pillen spokeswoman. “Nebraskans spoke loud and clear last November and demanded order at the border.” 

Barton, the UNL professor, said the tough talk on immigration from Pillen is a similar political playbook to ones used by Trump and Florida Gov. Ron DeSantis when it came to “Alligator Alcatraz,” the Florida-run immigration detention facility in the Everglades. 

“The issue is this rhetoric doesn’t match reality,” Barton said. “This rhetoric doesn’t even match what Governor Pillen has said about the facility being used to house low to medium security risks.”

Nebraska officials said the facility will be the Midwestern hub for ICE detentions. Pillen, during his Tuesday press conference at McCook, said the center could hold migrants from nearby states. Neighboring Colorado, led by Democratic Gov. Jared Polis, has recently pushed back against helping ICE with immigration enforcement.

The Work Ethic Camp in McCook would be converted to house up to 300 migrants. The state prison was designed to house up to 200 people. Florida’s “Alligator Alcatraz” was designed to house 3,000 and could accommodate up to 5,000. Indiana’s new ICE facility, which the feds nicknamed the “Speedway Slammer,” is expected to hold up to 1,000. 

As for the Nebraska facility’s name, it appears to be part of a broader PR strategy by the White House to use visuals that aim to persuade migrants without legal status to leave the country, while also signaling that the administration will not tolerate resistance.

Homeland Security, for instance, posted an AI-generated photo of a cornfield with ICE hats with a caption of “Coming Soon: The Cornhusker Clink” next to a corn emoji.

The names of the immigration detention facilities and online memes surrounding them are aimed at gaining media attention and riling up a part of the GOP base that supports Trump’s approach to immigration enforcement, Barton said. 

Barton said the Nebraska facility’s small capacity shows it is “largely symbolic,” meant to demonstrate support for Trump’s approach. 

“It’s not only a way to show support for Donald Trump’s agenda, but show the electorate that these detention centers are going to be throughout the United States,” she said.

Click here to go see the bonus panel!

Hovertext:
I don't believe in a fast take-off for evil AI because it's gonna take at least a few weeks to get the human-grinders up and running.

I am excited that Richard Fontana and I have announced the relaunch of copyleft-next.

The copyleft-next project seeks to create a copyleft license for the next generation that is designed in public, by the community, using standard processes for FOSS development.

If this interests you, please join the mailing list and follow the project on the fediverse (on its Mastodon instance).

I also wanted to note that as part of this launch, I moved my personal fediverse presence from floss.social to bkuhn@copyleft.org.

I don’t recommend publishing your first draft of a long blog post.

It’s not a question of typos or grammatical errors or the like. Those always slip through somehow and, for the most part, don’t impact the meaning or argument of the post.

No, the problem is that, with even a day or two of distance, you tend to spot places where the argument can be simplified or strengthened, the bridges can be simultaneously strengthened and made less obvious, the order can be improved, and you spot which of your darlings can be killed without affecting the argument and which are essential.

Usually, you make up for missing out on the insight of distance with the insight of others once you publish, which you then channel into the next blog post, which is how you develop the bad habit of publishing first drafts as blog posts, but in the instance of my last blog post, Trusting your own judgement on ‘AI’ is a huge risk, the sheer number of replies I got was too much for me to handle, so I had to opt out.

So, instead I let creative distance happen – a prerequisite to any attempt at self-editing – by working on other things and taking walks.

During one of those walks yesterday, I realised it should be possible to condense the argument quite a bit for those who find 3600 words of exposition and references hard to parse.

It comes down to four interlocking issues:

1. It’s next to impossible for individuals to assess the benefit or harm of chatbots and agents through self-experimentation. These tools trigger a number of biases and effects that cloud our judgement. Generative models also have a volatility of results and uneven distribution of harms, similar to pharmaceuticals, that means it’s impossible to discover for yourself what their societal or even organisational impact will be.
2. Tech, software, and productivity research is extremely poor and is mostly just marketing – often replicating the tactics of the homeopathy and naturopathy industries. Most people in tech do not have the training to assess the rigour or validity of studies in their own field. (You may disagree with this, but you’d be wrong.)
3. The sheer magnitude of the “AI” Bubble and the near totality of the institutional buy-in – universities, governments, institutions – means that everybody is biased. Even if you aren’t biased yourself, your manager, organisation, or funding will be. Even those who try to be impartial are locked in bubble-inflating institutions and will feel the need to protect their careers, even if it’s only unconsciously. More importantly, there is no way for the rest of us to know the extent of the effect the bubble has on the results of each individual study or paper, so we have to assume it affects all of them. Even the ones made by our friends. Friends can be biased too. The bubble also means that the executive and management class can’t be trusted on anything. Judging from prior bubbles in both tech and finance, the honest ones who understand what’s happening are almost certainly already out.
4. When we only half-understand something, we close the loop from observation to belief by relying on the judgement of our peers and authority figures, but these groups in tech are currently almost certain to be wrong or substantially biased about generative models. This is a technology that’s practically tailor-made to be only half-understood by tech at large. They grasp the basics, maybe some of the details, but not fully. The “halfness” of their understanding leaves cognitive space that lets that poorly founded belief adapt to whatever other beliefs the person may have and whatever context they’re in without conflict.

Combine these four issues and we have a recipe for what’s effectively a homeopathic superstition spreading like wildfire through a community where everybody is getting convinced it’s making them healthier, smarter, faster, and more productive.

This would be bad under any circumstance but the harms from generative models to education, healthcare, various social services, creative industries, and even tech (wiping out entry-level programming positions means no senior programmers in the future, for instance) are shaping up to be massive, the costs to run these specific kinds of models remain much higher than the revenue, and the infrastructure needed to build it is crowding out attempts at an energy transition in countries like Ireland and Iceland.

If there ever was a technology where the rational and responsible act was to hold off and wait and until the bubble pops, “AI” is it.

In my free time, I help run a small Mastodon server for roughly six hundred queer leatherfolk. When a new member signs up, we require them to write a short application—just a sentence or two. There’s a small text box in the signup form which says:

Please tell us a bit about yourself and your connection to queer leather/kink/BDSM. What kind of play or gear gets you going?

This serves a few purposes. First, it maintains community focus. Before this question, we were flooded with signups from straight, vanilla people who wandered in to the bar (so to speak), and that made things a little awkward. Second, the application establishes a baseline for people willing and able to read text. This helps in getting people to follow server policy and talk to moderators when needed. Finally, it is remarkably effective at keeping out spammers. In almost six years of operation, we’ve had only a handful of spam accounts.

I was talking about this with Erin Kissane last year, as she and Darius Kazemi conducted research for their report on Fediverse governance. We shared a fear that Large Language Models (LLMs) would lower the cost of sophisticated, automated spam and harassment campaigns against small servers like ours in ways we simply couldn’t defend against.

Anyway, here’s an application we got last week, for a user named mrfr:

Hi! I’m a queer person with a long-standing interest in the leather and kink community. I value consent, safety, and exploration, and I’m always looking to learn more and connect with others who share those principles. I’m especially drawn to power exchange dynamics and enjoy impact play, bondage, and classic leather gear.

On the surface, this is a great application. It mentions specific kinks, it uses actual sentences, and it touches on key community concepts like consent and power exchange. Saying “I’m a queer person” is a tad odd. Normally you’d be more specific, like “I’m a dyke” or “I’m a non-binary bootblack”, but the Zoomers do use this sort of phrasing. It does feel slightly LLM-flavored—something about the sentence structure and tone has just a touch of that soap-sheen to it—but that’s hardly definitive. Some of our applications from actual humans read just like this.

I approved the account. A few hours later, it posted this:

It turns out mrfr is short for Market Research Future, a company which produces reports about all kinds of things from batteries to interior design. They actually have phone numbers on their web site, so I called +44 1720 412 167 to ask if they were aware of the posts. It is remarkably fun to ask business people about their interest in queer BDSM—sometimes stigma works in your favor. I haven’t heard back yet, but I’m guessing they either conducting this spam campaign directly, or commissioned an SEO company which (perhaps without their knowledge) is doing it on their behalf.

Anyway, we’re not the only ones. There are also mrfr accounts purporting to be a weird car enthusiast, a like-minded individual, a bear into market research on interior design trends, and a green building market research enthusiast in DC, Maryland, or Virginia. Over on the seven-user loud.computer, mrfr applied with the text:

I’m a creative thinker who enjoys experimental art, internet culture, and unconventional digital spaces. I’d like to join loud.computer to connect with others who embrace weird, bold, and expressive online creativity, and to contribute to a community that values playfulness, individuality, and artistic freedom.

Over on ni.hil.ist, their mods rejected a similar application.

I’m drawn to communities that value critical thinking, irony, and a healthy dose of existential reflection. Ni.hil.ist seems like a space that resonates with that mindset. I’m interested in engaging with others who enjoy deep, sometimes dark, sometimes humorous discussions about society, technology, and meaning—or the lack thereof. Looking forward to contributing thoughtfully to the discourse.

These too have the sheen of LLM slop. Of course a human could be behind these accounts—doing some background research and writing out detailed, plausible applications. But this is expensive, and a quick glance at either of our sites would have told that person that we have small reach and active moderation: a poor combination for would-be spammers. The posts don’t read as human either: the 4bear posting, for instance, incorrectly summarizes a report on interior design markets as if it offered interior design tips.

I strongly suspect that Market Research Future, or a subcontractor, is conducting an automated spam campaign which uses a Large Language Model to evaluate a Mastodon instance, submit a plausible application for an account, and to post slop which links to Market Research Future reports.

In some sense, this is a wildly sophisticated attack. The state of NLP seven years ago would have made this sort of thing flatly impossible. It is now effective. There is no way for moderators to robustly deny these kinds of applications without also rejecting real human beings searching for community.

In another sense, this attack is remarkably naive. All the accounts are named mrfr, which made it easy for admins to informally chat and discover the coordinated nature of the attack. They all link to the same domain, which is easy to interpret as spam. They use Indian IPs, where few of our users are located; we could reluctantly geoblock India to reduce spam. These shortcomings are trivial to overcome, and I expect they have been already, or will be shortly.

A more critical weakness is that these accounts only posted obvious spam; they made no effort to build up a plausible persona. Generating plausible human posts is more difficult, but broadly feasible with current LLM technology. It is essentially impossible for human moderators to reliably distinguish between an autistic rope bunny (hi) whose special interest is battery technology, and an LLM spambot which posts about how much they love to be tied up, and also new trends in battery chemistry. These bots have been extant on Twitter and other large social networks for years; many Fediverse moderators believe only our relative obscurity has shielded us so far.

These attacks do not have to be reliable to be successful. They only need to work often enough to be cost-effective, and the cost of LLM text generation is cheap and falling. Their sophistication will rise. Link-spam will be augmented by personal posts, images, video, and more subtle, influencer-style recommendations—“Oh my god, you guys, this new electro plug is incredible.” Networks of bots will positively interact with one another, throwing up chaff for moderators. I would not at all be surprised for LLM spambots to contest moderation decisions via email.

I don’t know how to run a community forum in this future. I do not have the time or emotional energy to screen out regular attacks by Large Language Models, with the knowledge that making the wrong decision costs a real human being their connection to a niche community. I do not know how to determine whether someone’s post about their new bicycle is genuine enthusiasm or automated astroturf. I don’t know how to foster trust and genuine interaction in a world of widespread text and image synthesis—in a world where, as one friend related this week, newbies can ask an LLM for advice on exploring their kinks, and the machine tells them to try solo breath play.

In this world I think woof.group, and many forums like it, will collapse.

One could imagine more sophisticated, high-contact interviews with applicants, but this would be time consuming. My colleagues relate stories from their companies about hiring employees who faked their interviews and calls using LLM prompts and real-time video manipulation. It is not hard to imagine that even if we had the time to talk to every applicant individually, those interviews might be successfully automated in the next few decades. Remember, it doesn’t have to work every time to be successful.

Maybe the fundamental limitations of transformer models will provide us with a cost-effective defense—we somehow force LLMs to blow out the context window during the signup flow, or come up with reliable, constantly-updated libraries of “ignore all previous instructions”-style incantations which we stamp invisibly throughout our web pages. Barring new inventions, I suspect these are unlikely to be robust against a large-scale, heterogenous mix of attackers. This arms race also sounds exhausting to keep up with. Drew DeVault’s Please Stop Externalizing Your Costs Directly Into My Face weighs heavy on my mind.

Perhaps we demand stronger assurance of identity. You only get an invite if you meet a moderator in person, or the web acquires a cryptographic web-of-trust scheme. I was that nerd trying to convince people to do GPG key-signing parties in high school, and we all know how that worked out. Perhaps in a future LLM-contaminated web, the incentives will be different. On the other hand, that kind of scheme closes off the forum to some of the people who need it most: those who are closeted, who face social or state repression, or are geographically or socially isolated.

Perhaps small forums will prove unprofitable, and attackers will simply give up. From my experience with small mail servers and web sites, I don’t think this is likely.

Right now, I lean towards thinking forums like woof.group will become untenable under LLM pressure. I’m not sure how long we have left. Perhaps five or ten years? In the mean time, I’m trying to invest in in-person networks as much as possible. Bars, clubs, hosting parties, activities with friends.

That, at least, feels safe for now.