The colors on his canvas will never fade

@Mark Ollig

My oldest, Mathew, is a professional visual artist with a studio in Northeast Minneapolis and is known to most as Mat.

Mat grew up in Winsted and Waverly and knew from an early age he was meant to be an artist, so he set out to pursue that calling.

As a boy, he often tagged along with his grandparents to Red’s Cafe in Montrose, a Wright County gathering spot for local town dwellers and over-the-road truckers who pulled off Highway 12 for soup and a sandwich.

“My grandpa was a farmer, and he would sit in a booth with his buddies in their flannel shirts and shoot the breeze,” Mat recalled.

Growing up, he learned to paint by watching Bob Ross and Bill Alexander on PBS and by subscribing to art magazines.

As a junior in 2000, he enrolled at the Perpich Center for Arts Education, a public arts high school in Golden Valley named after former Gov. Rudy Perpich, where 11th and 12th-grade students study visual arts, music, dance, theater, literary arts, and media arts.

At Perpich, his artistic horizons widened considerably.

Mat later reflected that the school gave him “the confidence and the guidance I needed to follow my dreams . . . of being the next Bob Ross, until Craig Farmer taught me that there were other, better artists and so much incredible artwork that I could be inspired by.”

Upon graduating in 2002, his Perpich portfolio had earned him acceptance to the Lyme Academy of Fine Arts in Old Lyme, CT, but he deferred enrollment twice while continuing to develop his artistic style.

Mat ultimately chose the contemporary art education offered by the Minneapolis College of Art and Design (MCAD), where he earned a Bachelor of Fine Arts in painting and drawing in 2010.

“Without the education, guidance, and networking MCAD gave me, I’d probably be an accountant or something,” he said of his time there.

While at MCAD, he studied abroad in 2009 at the Accademia di Belle Arti in Florence, Italy.

The semester abroad proved transformative. Standing before Renaissance masterworks he had known only from books, he was overwhelmed by their scale, color, and grandeur.

“The reproductions did nothing to prepare me for how the originals looked in the flesh,” he said.

Returning to Minneapolis from Italy, he found his artistic direction permanently changed.

After graduating from MCAD in 2010, Mat landed his first major commission when the Minneapolis law firm Ogletree Deakins found his work on the MCAD website.

He created a single lobby painting, and the firm liked it so much it asked him to produce artwork for the entire office, launching his professional career.

The commissions grew quickly from there.

In 2012, he contributed original artwork to the Hyatt Regency Minneapolis’s $25 million renovation, creating several works for the lobby, private dining room, and guest rooms.

Most notably, a richly layered print of the iconic Gold Medal Flour building overlooking the Mississippi River was placed in 250 rooms to give guests a deeply local sense of place, and a June 23, 2012, Star Tribune article singled out his work among the original artwork in the remodeled rooms.

For the hotel’s grand main hallway, working alongside Stonehill & Taylor, he painted three sweeping canvases drawn from Minneapolis’s rugged and storied lumber heritage.

His work from this period was also featured in the May 2012 issue of Lodging Magazine.

More commissions followed.

In 2013, he painted a 20-by-72-inch view of the Minneapolis skyline for the clubroom of the Millennium Hotel in Minneapolis.

In 2016, he designed a 14-foot dimensional installation of 23 individual canvases for software company Code42 and created original artwork for Omni Brewing in Maple Grove.

His paintings reached the auction block, too, presented by Christie’s auctioneers at the annual MCAD fundraiser “The Auction at MCAD” in 2015 and 2016.

Recognition came alongside the commissions.

Gov. Mark Dayton appointed him to the Perpich Center for Arts Education board of directors in January 2014, a role he held for two terms until January 2023, and his career was profiled on the Curious North podcast that same month.

Mat is known for his Minnesota diner series, 10 original collage‑style paintings crafted to honor and preserve cafe culture in the state’s small towns and neighborhoods.

In a distinctive finishing touch, each completed painting was hung in the very diner that inspired it.

Red’s Cafe, one of the diners in the series, was a longtime favorite of former US Sen. and Vice President Hubert H. Humphrey, who lived in Waverly and served as vice president under President Lyndon B. Johnson.

Mat also painted a portrait of Humphrey as part of the Red’s Cafe piece.

“The diners I selected are local landmarks,” he explained. “Some of them have been around for decades, and one is only about 10 years old, but they all have an atmosphere. I want each picture to evoke a memory: I paint in bits and pieces because that’s the way we see the world.”

With a 2016 Artist Initiative Grant from the Minnesota State Arts Board, Mat spent the summer traveling the state with his Nikon camera and sketchbook.

Visiting Main Street diners across the state, he took photos to inform his painting compositions.

“I was looking for places that had a nostalgic spirit, and they all had to have that diner smell: grease, coffee, grilled onions, with just a tinge of bleach,” Mat said.

Using the Twitter hashtag “#mnArtHunt” and his website, , he encouraged diner lovers to travel from one diner to the next to collect a free postcard at each location.

The participating diners included The Barn in Brainerd, Bev’s Cafe in Red Wing, Lange’s Cafe in Pipestone, Red’s Cafe in Montrose and The Viking Cafe in Fergus Falls, among others.

“Minnesota artist hopes to save cafe culture one painting at a time” is the title of a Jan. 10, 2017 Minneapolis Star Tribune article, with a subtitle “To introduce Main Street diners to a new clientele, a Minneapolis painter [Mat] has created an artsy scavenger hunt across the state.”

He received the Minnesota State Arts Board Artist Initiative Grant three times, in 2016, 2018, and 2020.

He used the 2018 grant to travel Minnesota’s border with his Nikon camera and sketchbook, while the 2020 project was interrupted by the COVID-19 pandemic.

In 2015, one of Mat’s paintings was displayed in the Senate wing of the Minnesota State Capitol as part of the “Minnesotan Moments” exhibition.

In 2017, he received two honors at the Minnesota State Fair: the Banfill-Locke Center for the Arts Award and the Minnesota Artist Association Award.

Mat joined a public debate sparked by Minnesota’s 2023 law restricting lead and cadmium in certain consumer products.

As the rules were applied, professional artist paints were swept into the “covered product” list under Minnesota Statutes 325E.3892, which bars the sale of covered products with cadmium above 0.0075 percent by total weight (75 parts per million).

Artists and art-supply stores told local media they only realized the impact when they were suddenly denied access to cadmium-based paints.

Mat said substitutes do not hold up and that the reds and oranges “completely fade away,” adding, “Cadmium is vastly superior.”

He compared the proposed substitutes to “artificial vanilla flavor versus using real vanilla,” and said pigments are “like visual spices in your cabinet.”

Mat described cadmium colors in practical, painter-to-painter terms in a MinnPost article by Sheila Regan Feb. 5, 2025.

“The cadmium line of colors goes from cadmium red deep, which is a really luscious rich red, all the way through the cadmium oranges, to all the cadmium yellows, and cadmium green, which is this really vibrant green. All of the vibrant colors on the palette are cadmiums,” Mat said.

He also challenged the suggestion that artists are a significant source of cadmium pollution.

“The argument I heard was that they’re trying to keep it from waterways, but artists are such a tiny, tiny group of people, and we want that pigment on the canvas,” Mat said.

“We don’t want it down the drains or anything like that,” he concluded.

Minnesota lawmakers ultimately revised the language June 14, 2025, through House File 4 (Chapter 4), exempting professional artists’ materials, including oil and water-based paints, pastels, pigments, ceramic glazes, markers, and encaustics, from the state’s cadmium restrictions.

Mat has explored a concept he calls “anemoia,” a coined term for the feeling of nostalgia for a time one never actually experienced.

Drawing on vintage photographs, illustrations, and design motifs from the 1950s through 1970s, the paintings carry a strange sense of modernity built entirely from nostalgic imagery.

He returned to Perpich as the Sept. 15, 2025, Dedication Day alumni speaker, bringing with him the perspective of someone whose artistic life began there, and told students that his time at the school held “some of the best memories of my childhood.”

“Twenty-five years ago, I was where you are right now: in a new school, in a new city. Scared, but excited. Everything started out right here. Perpich was the key to everything I’ve accomplished. Perpich gave me the confidence, the work ethic, the curiosity, the artistic knowledge, my friends, and the hope and tenacity to follow my dreams and make art my career,” he said.

“You are only starting your journey,” he told them, encouraging them to dedicate themselves not only to Perpich, but also to their own possibilities.

Known for his MetaModern oil paintings, in which bold colors and expressive brushstrokes blend classic techniques with modern ideas, Mat has built a distinctive body of work.

For each painting, he builds the wooden stretcher frame himself and pulls the canvas taut by hand.

Drawing inspiration from local Minnesota landmarks, Mat’s digitally composed neo-cubist works investigate the relationship between computers, image appropriation, and traditional oil painting.

The work reflects an ongoing fascination with memory, reality and perception, often playing with the tension between everyday life, consumer culture and an increasingly virtual world.

Mat’s influences span Cubism, Postmodernism and Relational Aesthetics, and include such artists as Gerhard Richter, Mark Rothko, Alexander Ross, and James Rosenquist.

“Exploiting the ‘slow-think’ granted to the viewing of oil paintings, my artwork attempts to fuse the various experiential elements together to create an illusion of truth,” Mat has said of his work.

Today, Mat works from his own studio at 3104 N. Pacific St. in Northeast Minneapolis, creating original works and taking on commissions.

His paintings are held in notable collections across the country and internationally, including a painting in the permanent collection of the Weisman Art Museum, and are represented by Kickernick Gallery, both in Minneapolis.

More of Mat’s work can be viewed at matollig.com and on his Instagram page at instagram.com/matollig.

As his proud father, I’ve had a front-row seat to Mathew’s art over the course of his lifetime, witnessing the care, skill and dedication he puts into each painting.

AI writes code as humans watch

@Mark Ollig

A significant paradigm shift is underway as autonomous artificial intelligence (AI) agents are now helping create the next generation of AI.

Anthropic, a San Francisco AI company founded in 2021 and creator of the Claude AI platform, released its report “When AI Builds Itself” June 4 of this year.

The report highlights how AI systems, particularly Claude, are increasingly coding, building, testing, and refining themselves with less human involvement.

“I started leaning hard into Claudifying about a year ago. That’s been a crazy adventure, and it’s now been about five months since I last wrote any code myself,” said one Anthropic employee.

Anthropic reports that Claude now writes more than 80% of the code merged into its production systems.

This constitutes a major shift from human-written to machine-generated code, though human engineers still set priorities, decide what to build, and review Claude’s output.

By mid-2026, Anthropic software engineers were adding about eight times as much code each day as in 2024, mainly because Claude wrote most of it, allowing human engineers to spend more time on planning and code review.

What began as simple autocomplete for code has evolved into coding agents that can edit entire files, run tests, and assign follow-up tasks to other agents with only limited human oversight.

The same trend is appearing in AI research, where Claude is helping improve model performance with increasing speed and effectiveness.

In April of his year, Anthropic published results from an experiment called the “Automated Weak-to-Strong Researcher.”

Teams of AI agents, software programs that can plan tasks, write code, and run experiments with little human guidance, were given an open-ended AI challenge question: can a weaker AI model reliably help train a stronger one?

The AI agents worked in parallel, spending a combined 33 days, or roughly 800 cumulative hours, proposing ideas, running experiments, and sharing results with each other.

These AI agents solved 97% of the problem in about two days at a computing cost of about $18,000.

Two human researchers working the same challenge for about seven days solved only 23%.

An Anthropic researcher on the project said of Claude’s performance: “I think if [a junior colleague] came back to me with results like this in the same span of time, I would be mildly impressed. The future is now.”

Anthropic says AI is not yet fully building better versions of itself, but the Automated Weak-to-Strong Researcher experiment shows the capability is closer than most people realize, and planning for it should start now.

As AI writes its own code and conducts its own research, future progress will depend less on human engineering teams and more on AI computing power.

That computing power is found in data centers packed with thousands of graphics processing units, dedicated servers, high-speed networking equipment, storage systems, and massive cooling and backup power systems.

Delivered over fiber-optic networks, where bits and bytes speed through glass on a beam of light, this infrastructure allows tech companies to develop ever more powerful AI models, which become increasingly capable of writing code, conducting research, and improving themselves.

Companies and government agencies, from healthcare, finance, and telecommunications to manufacturing, retail, and public services, are leveraging AI to automate processes, analyze data, and gain competitive advantages.

Larger data centers can accelerate AI progress, but at a significant cost of billions of dollars in infrastructure investment, along with growing concerns over electricity and water consumption.

A June 3 Reuters report cited United Nations University researchers who estimated data centers used 448 terawatt hours of electricity in 2025 and 1.19 trillion gallons of water, warning both could more than double by 2030 as AI demand grows.

Anthropic notes the shift is not complete: humans still outperform AI in deciding which questions matter, which results to trust, and when to abandon an idea.

If AI systems become capable not only of helping humans write code but of designing, testing, and improving their own AI successors, the pace of change could accelerate beyond our ability to manage it.

Anthropic outlines three paths: progress could stall, AI could automate more work while humans remain in control, or AI could begin helping build better versions of itself, making advances largely dependent on computing power and infrastructure.

The report points to a next phase in which AI moves beyond assisting human researchers to actively accelerating the development of future AI systems.

“There isn’t full consensus among staff at Anthropic, but many believe that the Claude-written code was still worse in quality than human-written code at Anthropic in late 2025, and is roughly at parity today. We expect it to be better within the year,” the report states.

AI advancement is moving at warp speed, and our ability to understand and control it is falling behind faster than laws, institutions, and basic safety checks can keep pace.

It is mastering the tasks we assign it and is starting to shape its own development, pushing humans from active AI coders to after-the-fact observers.

Much of this code is written in programming languages such as Python, widely used in AI development for its simplicity and the vast number of AI tools and libraries built around it.

If we wait until the warning signs are unmistakable, such as AI independently rewriting its own core programming without human approval, it may already be too late.

By then, AI could be making important decisions and modifying itself faster than humans can understand, question, or stop.

“The evidence suggests that the human role is narrowing at each step in the AI development process. Once human and AI-authored code quality reach parity, humans will stop writing code entirely, and shift to only reviewing it. But if they can’t review code as quickly as Claude can generate it, human review will become the bottleneck [barrier] to AI development,” the Anthropic report said.

And for those of you wondering, yes, I use Claude Sonnet 4.6.

In the “Star Trek” episode “The Ultimate Computer,” the M-5 autonomously intelligent computer was installed on the Enterprise to run the ship with just 20 crew members.

It started drawing unlimited power from the ship’s warp engines, blocked any human overrides, and resisted shutdown attempts.

“Fantastic machine, the M-5. No off switch,” a frustrated Dr. McCoy said.

During a war games scenario, M-5 attacked Federation starships, but Captain Kirk convinced it to shut itself down, allowing the crew to regain control.

Unlike the science fictional M-5, our future AI may not be so easily reasoned with.

Last I checked, there is no off switch for AI.

A conceptual illustration depicting autonomous AI-generated code,
left, and a human software engineer observing as artificial intelligence
 increasingly writes, tests and refines its own programming.
Photo by ChatGPT with content from the Bits and Bytes column.

Building a future, one weld at a time

@Mark Ollig

In his grandfather’s farm shop near Waverly, my youngest son, Andrew, known as Andy to most, first became interested in welding while helping repair worn farm equipment.

Surrounded by farm tools and machinery, he began learning the welding trade at age 10.

By 14, Andy was taking on welding projects for others.

“Once I got into welding, I knew that was the direction I wanted to go,” he said.

During his high school junior and senior years, Andy competed in the statewide welding competition at St. Cloud Technical College, winning second place in MIG welding as a junior and third place as a senior among nearly 100 contestants.

After graduating from high school in 2007, he studied welding and metal fabrication at what is now St. Cloud Technical and Community College, where he earned his diploma.

After working in the welding industry, Andy started the business he had wanted for years: his own welding shop, which he opened in 2009.

Today, Andy is the owner and operator of Ollig’s Custom Metal Works, which serves industrial, agricultural, commercial, and individual customers with routine welding needs, repairs, and custom metal fabrication services.

He currently does production welding for area and out-of-state businesses, working with mild steel, stainless steel, and aluminum.

One job led to another, and word of mouth helped him build a solid reputation for dependable, high-quality work.

Using programs such as AutoCAD, he creates detailed two and three-dimensional digital drawings that show size, shape, angles, holes, bends, parts, and weld locations.

AutoCAD gives him precise control over drafting, design, and documentation.

That level of detail helps whether the job is a small project or a full machine with many parts that need to work together.

For a welder, a digital drawing serves as a road map, reducing guesswork, improving accuracy, and giving customers a clearer sense of the finished product.

“When you can take an idea, build it, and then see it work the way it’s supposed to, that’s a pretty good feeling,” Andy said.

Before welding begins, supplies are ordered, and the metal is measured, cut, cleaned, and fitted correctly.

Andy said the strength and quality of the finished work depend as much on preparation as on the weld itself.

His shop is equipped for metal inert gas (MIG) welding, tungsten inert gas (TIG) welding, and stick welding.

It also has grinders, saws, mills, plasma cutters, drill presses, lathes, clamps, welding tables, safety gear, a 25-ton computer numerical control, or CNC, press brake, and specialty production equipment.

Over the years, welding and fabrication equipment has become more powerful, efficient, and computerized.

Many modern machines use digital controls that let welders fine-tune heat, wire-feed speed, gas flow, and welding modes for different metals, thicknesses, and job requirements.

Those settings can help improve consistency, reduce wasted material, and produce cleaner, stronger welds.

Still, Andy said the basics matter most: preparation, fit, steady hands, and knowing how various metals react under heat.

He spends much of his time doing MIG and TIG welding.

MIG welding uses a gun that feeds a continuous wire electrode into the joint while shielding gas protects the molten weld puddle from contamination by the surrounding air.

It is widely used for repair, fabrication, and production because it is fast, strong, and efficient.

TIG welding uses a nonconsumable tungsten electrode to create the arc while the welder controls the heat and adds filler metal by hand, producing cleaner, more precise welds where appearance and finish matter.

“TIG takes more patience, but it’s rewarding because the finished product looks so nice,” Andy said.

The choice depends on the job, the metal, and the finish the customer needs.

Andy said projects completed in Minnesota sometimes go to other states and, through companies he does work for, some end up in other countries.

He has also seen the industry change as new processes, alloys, and machines have expanded what welders can do, and he pointed to pulsed MIG welding as one process gaining ground.

Pulsed MIG welding uses controlled current pulses instead of a single steady arc, giving the welder better control over heat input, weld spatter, and warping, which results in a higher-quality finish.

That added control can help when welding thinner metal, different alloys, or jobs where a cleaner weld is needed.

Andy stays up to date in the welding industry by reviewing manufacturer training and product materials, attending trade shows, and using newer equipment.

He acknowledges that robotic welding systems are becoming more common in factory production, where predictable conditions and repeated welds make automation especially effective.

These systems are set up and maintained by skilled welders and work best in controlled factory settings where parts and weld patterns remain consistent.

Human welders are still essential for repair work, custom fabrication, on-site jobs, and tight or awkward spaces where conditions can change from one task to the next.

In those situations, the welder assesses each job, determines the best approach, and adjusts as conditions change.

Human welders turn ideas and rough sketches into finished parts and products for customers and manufacturers.

Repair work can be especially demanding, as a damaged part may no longer fit as it once did, and the metal may be bent, rusted, cracked, worn thin, or too deteriorated to reuse.

That kind of work still depends on judgment, experience, and hands-on skill.

After 17 years of managing and operating Ollig’s Custom Metal Works, Andy still finds satisfaction in finishing each job.

Whether it is a large project or a small repair, Andy said customers appreciate getting the same level of care, attention, and quality as his larger manufacturing clients.

Andy believes the need for skilled welders remains strong, and he encourages young people to consider the welding trade because too few are entering it.

Welding can offer a solid career path for people who enjoy working with their hands, building things, and solving problems, he said.

“There’s a real need for skilled people, and if you like building things and seeing the results of your work, it’s a good career,” Andy added.

Looking back, he takes pride in completing each job well and sees welding as a craft that lets him solve problems, serve customers, build a business, and earn a living through skill, dedication, and perseverance.

Andy put it this way: “At the end of the day, it feels good to stand back, look at what you designed, built, and welded, and know it’s going to work for somebody.”

He makes his father proud.

Four wires, two computers, one historic test

@Mark Ollig

Long before billions of devices connected over the internet, the US military sent digital information over telephone lines.

During the late 1950s, SAGE, short for Semi-Automatic Ground Environment, was an early interactive, networked Cold War-era air-defense system.

SAGE sent digitized radar data over telephone lines to computers using early modems, short for modulator-demodulators, that turned the radar information into analog tones the telephone circuits could carry.

Those computers processed radar tracks at SAGE direction centers and helped console operators watching radar screens direct interceptor aircraft and surface-to-air missile batteries toward approaching bombers.

SAGE demonstrated that telephone lines could reliably carry digitized radar data across long distances to computers used to defend the nation’s airspace.

Duluth’s SAGE Direction Center began operations Nov. 15, 1959, adding Minnesota to the northern tier of SAGE sites watching for potential Soviet bomber attacks.

SAGE had already shown that telephone lines could carry radar-derived information as analog tones to air-defense computers, but that was not the same as using telephone lines for direct digital data exchange between computers sharing programs, information, and processing work.

Lawrence G. “Larry” Roberts and Thomas M. “Tom” Marill took that next step, testing whether distant time-sharing computers could exchange digital data over telephone lines and share computing resources.

J. C. R. Licklider, a psychologist and computer scientist, wrote an Advanced Research Projects Agency memo April 23, 1963, for the “Members and Affiliates of the Intergalactic Computer Network.”

In it, he described how researchers could use connected computers to share programs, files, information, and remote computing power.

Licklider imagined his “Intergalactic Computer Network” choosing the easier path: either sending a user’s information from one connected computer to a program running on another, or bringing the remote program back to work on the user’s information.

In 1965, Roberts and Marill began putting that vision into practice with an experiment linking two distant time-sharing computers, machines that divided processing time among several connected users.

The project connected the TX-2 computer at Massachusetts Institute of Technology’s Lincoln Laboratory in Lexington, MA, with the Q-32 computer at System Development Corp., or SDC, in Santa Monica, CA, about 2,590 miles away.

A limited 1965 Lincoln Laboratory-to-SDC test may have come first, but it was not a full computer-networking session; the two locations only sent and received bits to test whether a cross-country telephone line was reliable.

Because the TX-2 did not receive its modem until mid-1966, Roberts said he probably connected the line through an analog-to-digital converter.

Later 1966-67 records identify the TX-2-to-Q-32 circuit as a Western Union telephone-data circuit, not an ordinary public telephone call.

Western Union’s Broadband Exchange Service, launched Sept. 30, 1964, in Boston, gave businesses an alternative to AT&T’s Bell System facilities for data, facsimile, and voice.

Customers could dial into a broadband connection, select a 2 or 4-kilohertz bandwidth on a toll basis, and use Western Union’s coast-to-coast microwave link to exchange computer data.

The Bismarck Tribune reported Nov. 20, 1964 on Western Union’s $80 million microwave network, about $860 million in this year’s dollars, which supported Western Union’s Broadband Exchange Service.

The 7,500-mile network, designed for about 7,000 voice channels, connected Boston, New York, Washington, DC, San Francisco, and Los Angeles.

Its 267 microwave stations could carry high-speed facsimile, or fax, transmissions; computer data; telegraph messages; voice calls; and Telex, a typed-message service using teleprinter machines.

Western Union supplied modems to convert computer data to analog signals; customers provided equipment and wiring; and users managed calls and data channels with a special push-button Western Union phone.

The service transmitted up to 4,800 words per minute, sent a fax page in under three minutes, and allowed simultaneous sending and receiving.

For the 1965 TX-2-to-Q-32 project, later records identified Western Union Broadband Switching Service, a 1,200-bit-per-second asynchronous data set, and automatic answering equipment at SDC, which answered the incoming data call from the TX-2 side.

AT&T’s Bell System supplied Western Electric Data-Phone data sets, or modems, that linked business computers and data terminals to Bell telephone lines.

A data terminal was a device, such as a screen-and-keyboard terminal, teletypewriter, or card reader, used to enter, send, or receive information.

Computers or terminals connected to the data set through a serial interface cable, and the data set converted digital signals into tones for standard telephone lines.

The Bell 103A transmitted up to 300 bits per second over standard telephone lines; the Bell 201A reached 2,000 bits per second over dialed long-distance voice-grade lines.

The Western Union 1,200-bit-per-second data set sat between them: faster than the 103A, but slower than the 201A.

The larger difference was the network: Bell System facilities versus Western Union’s separate, measured-rate microwave broadband network.

By early 1967, a TX-2 user started the AT program, short for Algebraic Translator. The TX-2 automatic dialing equipment printed “dialing SDC,” then “connected.”

The TX-2 and Q-32 exchanged data over the Western Union four-wire, voice-grade path with separate transmit and receive paths; modems converted bits to audio tones and back.

Surviving records do not identify the dialing digits, address code, signaling method, switching office, or internal Western Union routing.

After connecting, AT logged into the Q-32 system, loaded LISP (list processing), sent a LISP program for compilation, and waited.

Roberts and Marill’s “Toward a Cooperative Network of Time-Shared Computers,” published in the 1966 Fall Joint Computer Conference proceedings, became an important early paper on computer resource-sharing networks.

It described the TX-2-to-Q-32 project, showed how two time-sharing systems could exchange data across the country, and explained resource sharing: one research center using another’s resources rather than duplicating programs and machines.

Roberts later said the concept proved computers could work together.

In the Roberts-Marill test, message blocks had to arrive intact; one flipped bit could ruin a number or command. Circuit switching kept the coast-to-coast circuit reserved for the session, even during short data bursts.

In the 1960s, packet switching offered a different approach: It broke data messages into small, addressed pieces called packets.

Those packets could take turns moving through shared network connections, then be put back together at the receiving end.

That made networks more efficient because many users could share the same connections instead of tying up one dedicated circuit for one session.

Packet switching helped shape the Advanced Research Projects Agency Network, or ARPANET, which went online in 1969 and became the foundation of today’s internet.

By the 1980s, the 1960s data services described here were giving way to faster and more flexible options, including end-to-end digital data services, dedicated T1 private lines carrying 1.544 megabits per second, dial-up modems moving from 300 to 1,200 and 2,400 bits per second, and public packet-switched networks.

In the 1990s, Internet Protocol networks pushed that evolution further as fiber-optic transport and higher-capacity digital circuits carried more long-distance data.

By the 2020s, packet-based networks carried webpages, email, cloud applications, streaming video, video meetings, text and chat messages, social media platforms, internet-based voice calls, and cellular voice calls.

Today’s networked computing world all goes back to four wires, two computers, and one historic test.

Sister Keller’s pioneering computing journey

@Mark Ollig

The early days of computer science had many pioneers, including one few people might expect: a Catholic nun named Sister Mary Kenneth Keller.

She was born Evelyn Marie Keller on Dec. 17, 1913, in Cleveland, OH, but grew up mostly in Chicago, IL.

In 1932, at 18, she joined the Sisters of Charity of the Blessed Virgin Mary in Dubuque, IA, where she took the name Sister Mary Kenneth the following year.

Keller earned a bachelor’s degree in mathematical sciences from DePaul University in Chicago in 1943, then completed a master’s degree there in 1952.

In 1961, she attended her first computer education workshop at Dartmouth College in Hanover, NH.

Dartmouth had no female undergraduates at the time, and it made an exception to its no-women rule so Keller could work in its computer lab during a summer program for high school teachers.

There, she learned to use a computer and write simple programs on the college’s LGP-30 computer.

Introduced in 1956 by Librascope, the desk-sized LGP-30 digital computer cost $50,000, which is roughly $621,000 in 2026 dollars.

The LGP-30 was used for engineering, education, scientific, and mathematical calculations, including research, design analysis, statistical work, and applied engineering problems.

The 800-pound computer system ran on standard 115-volt alternating current (AC) power, drew 1,500 watts, and required no special air conditioning.

Inside, it featured 113 vacuum tubes, 1,450 diodes, and a 4,096-word magnetic drum memory.

Each 32-bit word allocated 30 bits for data, one sign bit, and one spacer bit, yielding roughly 15 kilobytes (KB) of usable storage (out of 16 KB total).

Users interacted with the LGP-30 through a Friden Flexowriter, an electric typewriter with a keyboard, paper-tape punch, and paper-tape reader.

A built-in oscilloscope monitored the control counter register, instruction register, and accumulator register, providing a real-time view of the computer’s internal operation.

Looking back on her experience, Keller said, “I just went out to look at a computer one day, and I never came back . . . It looked to me as if the computer would be the most revolutionary tool for doing math that I could get.”

In May 1964, Dartmouth College made history when BASIC, the Beginner’s All-purpose Symbolic Instruction Code, ran its first program on a General Electric GE-225 mainframe computer.

BASIC is a high-level programming language that uses simple, English-like commands to help students and non-specialists learn programming concepts.

In 1964, Sister Keller revealed her vision for the future of academia when she predicted, “Its function in information retrieval will make it the hub of tomorrow’s libraries.”

In June 1964, Minneapolis-based Control Data Corp. published its FORTRAN-63 reference manual for the CDC 1604 and 1604-A computers.

FORTRAN, short for Formula Translation, was one of the first high-level programming languages.

Its compiler, the software that translated FORTRAN programs into machine instructions, was adapted for the CDC 1604 and CDC 3600 systems.

On May 14, 1965, Control Data Corp. reached an agreement to acquire the commercial computer business of General Precision Equipment Corp., including its Librascope division and support for existing LGP-30 installations.

Her dissertation advisors officially signed off on Keller’s doctoral research on May 21, 1965.

Her dissertation was titled “Inductive Inference on Computer Generated Patterns.”

Keller wrote custom algorithms using Control Data Corp.’s FORTRAN-63 compiler on the university’s large mainframe computers to complete this pioneering work.

At a time when most people were unfamiliar with these computing concepts, she studied how mainframe computers could recognize patterns, test logic, manipulate symbols, and infer rules from examples.

Inside the computer, those examples were processed as binary data, the ones and zeros digital computers use to represent information.

This early work in pattern recognition, rule formation, and learning from examples later became important foundations in artificial intelligence and machine learning.

Sister Mary Kenneth Keller made national history on June 7, 1965, at 51, by becoming the first woman in the United States to earn a Ph.D. in computer science from the University of Wisconsin-Madison.

After earning her Ph.D., Keller joined Clarke College in Dubuque, a women’s college founded by her own religious order, and established its brand-new computer science department.

In 1965, the National Science Foundation awarded her a grant for instructional equipment for undergraduate education.

Keller led Clarke’s computer science program for 20 years, dedicating her career to making computing accessible to all.

She promoted computer literacy, taught programming to people without technical backgrounds, co-authored educational materials, and continued researching computer-generated patterns.

By 1975, Keller was calling the computer "the greatest interdisciplinary tool” invented so far.

She also observed, “We’re having an information explosion, among others, and it’s certainly obvious that information is of no use unless it’s available.”

Her pioneering journey in computing is remembered today.

Sister Mary Kenneth Keller died Jan. 10, 1985, at 71 in Dubuque.

Browsers: How we got from the web to AI

@Mark Ollig

At CERN in Geneva, Switzerland, the European Organization for Nuclear Research, British computer scientist Tim Berners-Lee was working as a software engineer in 1990.

By the end of 1990, he had written the first code for a combined web browser and web page editor using a NeXT computer workstation, built by NeXT Computer, a company founded by Apple co‑founder Steve Jobs.

Berners-Lee’s first browser was called WorldWideWeb and later renamed Nexus to avoid confusion with the “World Wide Web” itself.

He also created Hypertext Markup Language (HTML) to organize webpages and developed Hypertext Transfer Protocol (HTTP) to share them across the internet.

He wrote the code for Uniform Resource Locators (URLs) and configured the first web server on a NeXT workstation at CERN.

Built specifically for the NeXT computer workstation, Nexus never reached the mainstream PC market and was discontinued in 1994.

Netscape Communications introduced Netscape Navigator in 1994. The browser helped make the web a place for business and soon became popular with early users. AOL stopped updating Navigator 9 March 1, 2008.

1993: Mosaic makes the web visual

In 1993, the National Center for Supercomputing Applications released Mosaic, the first browser to make the web truly graphical by displaying images alongside text. Its development ended with version 3.0 Jan. 7, 1997.

1994: Netscape arrives

Netscape Communications launched Netscape Navigator in 1994.

It helped turn the web into a commercial platform and quickly became a favorite among early users before AOL ended updates for Navigator 9 March 1, 2008.

1995: Microsoft enters the browser market

Microsoft joined the browser market Aug. 16, 1995, when it introduced Internet Explorer through the Plus! add-on for Windows 95.

After years of dominance, Internet Explorer reached the end of the road when Microsoft ended support for Internet Explorer 11 June 15, 2022.

In February 2023, Microsoft permanently disabled the desktop app and steered users to Microsoft Edge.

1996 to 2013: Opera’s evolution

Opera Software, a Norwegian company, released Opera 2.0 in 1996 as a shareware alternative to Netscape and Internet Explorer.

In 2003, Opera 7.0 introduced the Presto layout engine, which supported emerging web standards and powered the Opera Mini browser on mobile devices.

In 2013, Opera retired the Presto-based version and adopted Google’s Blink engine. Opera remains available on Windows, macOS, Linux, Android, and iOS.

2000s: Users shift away from Internet Explorer

In the 2000s, users moved away from Internet Explorer, seeking better standards support, tabs, improved security, faster performance, and fewer browser-specific pages.

2003: Safari launches

Apple introduced Safari Jan. 7, 2003, at Macworld Expo for Macintosh computers running Mac OS X 10.2 (Jaguar), Apple’s desktop operating system at the time.

It featured built-in Google search, improved bookmarks, pop-up blocking, and a SnapBack feature, marking Apple’s move away from Internet Explorer as the default Mac browser. Safari continues to be supported.

2004: Firefox 1.0

The Mozilla Foundation released Firefox 1.0 Nov. 9, 2004, as a free, faster alternative to Internet Explorer for Windows, Mac OS X, and Linux.

It offered tabbed browsing, pop-up blocking, built-in search, RSS (Really Simple Syndication) feeds, live bookmarks, fraud protection, and add-ons. Firefox remains supported.

2007: The iPhone puts the web in your hand

At Macworld in San Francisco Jan. 9, 2007, Steve Jobs introduced the iPhone as a groundbreaking internet device running Safari.

He described it as a breakthrough that put a fully usable web browser in people’s hands for the first time.

2008: Chrome changes everything

Google released the Chrome beta for Windows Sept. 2, 2008. The V8 engine made JavaScript run faster by compiling it into machine code. Chrome also kept tabs separate, so if one crashed, the others stayed open.

That same day, Google released Chrome’s source code as the open-source Chromium project, which later became the shared base for browsers such as Microsoft Edge, Opera, and Brave.

2012: Chrome expands to mobile

Google released Chrome for Android Beta Feb. 7, 2012, for devices running Android 4.0.

Google launched Chrome for iPhone and iPad June 28, 2012, enabling cross-device sync. Samsung Internet launched in 2012 as the default browser on Galaxy devices and remains Samsung’s main mobile browser.

2013: Blink arrives

In 2013, Google introduced Blink, a rendering engine that converts website code into the text, images, and buttons you see on-screen.

Blink now powers the Chromium-based browsers that dominate modern web use, including Edge, Opera, and Brave.

2015 to 2020: Edge and the Chromium transition

Microsoft launched Edge in 2015 with Windows 10, replacing Internet Explorer as its modern browser.

Edge continues to receive updates. Vivaldi 1.0, released April 6, 2016, by Vivaldi Technologies in Oslo, Norway, remains supported.

The Brave browser, which blocks ads and trackers by default, launched in 2016. Version 1.0 arrived in 2019 for Android, iOS, Windows, macOS, and Linux, and Brave remains supported.

In December 2018, Microsoft announced plans to rebuild Edge using Chromium to improve compatibility.

The new Microsoft Edge was released Jan. 15, 2020. Internet Explorer 11 was the final version of Internet Explorer.

Microsoft now directs users to Edge, which includes Internet Explorer (IE) mode for legacy websites and apps and is expected to remain available through at least 2029.

2024 to 2026: Artificial intelligence (AI) becomes part of the browser

In 2024, the Browser Company launched Arc Search for iPhone, a mobile browser with AI-powered search features. Arc Search remains supported.

AI is now built into many web browsers, letting users summarize pages, compare tabs, draft text, fill out forms, and get step-by-step guidance without leaving the browser.

In 2025, San Francisco-based Perplexity AI introduced Comet, an AI-powered browser for web research, page summaries, organization, and online tasks. Comet remains supported.

In Microsoft Edge, Copilot AI can summarize webpages, videos, and Portable Document Format (PDF) files, surfacing key points without extra searching.

Copilot also works inside Microsoft Word, where it can answer questions about a document, draft and rewrite text, and help shape rough ideas into polished paragraphs.

Google’s Gemini AI in Chrome acts like a helper that sits beside the page you’re viewing. It can explain a confusing paragraph, clarify an idea, or boil a long article down to its key points.

If you allow it, Gemini can also look at your other open tabs to understand what other tasks you are working on.

This year, Samsung extended its browser beyond mobile with Samsung Browser for Windows. Its AI features can summarize, analyze, and compare information across multiple tabs.

This year: Browser market share

Statcounter reported that, in April of this year, the leading web browsers worldwide were Chrome (68.02%), Safari (17.04%), Edge (5.53%), and Firefox (2.26%).

WebRTC and the shift to remote life

Web Real-Time Communication, or WebRTC, is the open-source technology behind live voice, video, and data in modern browsers.

Google introduced it in 2011, and it became a World Wide Web Consortium (W3C) standard in 2021.

WebRTC is built into current versions of Chrome, Firefox, Safari, and Edge on both desktop and mobile devices, and can be used without installing any extra plug-ins or add-ons.

When many of us moved to online life during COVID-19, WebRTC was needed for business meetings, virtual classrooms, telehealth appointments, and staying connected with family and friends.

The continuing transition toward AI-powered browsing marks the beginning of a new era in how we use the web and experience the internet.

Cable television started with only a few channels

@Mark Ollig

In the late 1940s, people in many small towns struggled to get good television reception because broadcast towers were too far away, or hills and mountains blocked the signals.

In 1948, John Walson Sr. of Mahanoy City, PA, found a practical way to improve local television reception.

He ran Army-surplus, heavy-duty twin-lead cable from a mountain antenna into town, creating an early cable television system.

The system brought in clear signals from channels 3, 6, and 10 out of Philadelphia and helped him sell more television sets at his appliance store.

“One of the things that got me interested in going into cable TV in a large way was the crowd that gathered in front of my store,” Walson said during a July 21, 1970, oral history interview conducted at his Service Electric Cable TV office, formerly his appliance store.

“When I first put those three channels on, the street was completely blocked with viewers, people watching the pictures in the window,” Walson Sr. said.

In 1948, Ed Parsons of Astoria, OR, built one of the first community antenna television, or CATV, systems in the United States.

At the time, Parsons owned a local radio station and set up a small receiving antenna system atop the Astoria Hotel. He later described it as a system of multiple Yagi antennas.

The Yagi-Uda antenna, developed in Japan in the 1920s by Shintaro Uda and Hidetsugu Yagi, is directional. Its metal elements help focus reception on a single source, making it effective for pulling in weak, distant television signals, such as KRSC-TV in Seattle, about 125 miles away.

“I found a usable signal up on the top of the Astoria Hotel,” Parsons recalled in his June 19, 1986, oral history interview.

Parsons used copper-conductor coaxial cable, amplifiers, and a community antenna to deliver the distant signal to nearby homes. Each household did not need its own antenna because the shared system received the signal and distributed it by cable.

The Minneapolis Times identified very high frequency, or VHF, channels 2, 4, 5, 7, and 9 for the Twin Cities market April 26, 1948.

KSTP-TV, Channel 5, launched the following day and became Minnesota’s first commercial television station.

From 1948 to 1952, the Federal Communications Commission, or FCC, put a pause on approving new television stations to keep up with the industry’s rapid growth.

During this time, channel assignments were shuffled, and some were reserved for educational use.

In the Twin Cities, Channel 7 was dropped from the commercial lineup, Channel 2 was designated for education, and VHF channels 4, 5, 9, and 11 were left for commercial television.

From 1952 to 1983, US ultra-high frequency, or UHF, television operated on channels 14 through 83, spanning the 470 to 890 megahertz, or MHz, band.

In the early 1980s, companies began experimenting with subscription television services delivered over UHF broadcast channels.

Subscribers needed decoder boxes because the broadcasts were scrambled.

Spectrum began broadcasting Sept. 22, 1982, on KTMA-TV, Channel 23, in the Twin Cities, offering scrambled subscription programming.

Its Spectrum Sports package included Minnesota Twins baseball and Minnesota North Stars hockey games.

In 1983, the FCC reallocated UHF channels 70 through 83 from television to land mobile radio services, including public safety, early cellular testing, and trunked radio.

A trunked radio system is a computer-controlled two-way radio network that lets many users share a small pool of radio frequencies.

Twin Cities Spectrum subscriptions peaked at 27,000 in May 1983, then fell to 13,000 by 1985 as cable competition grew and the Minnesota Twins and North Stars did not renew their sports contracts.

Spectrum’s movie service ended Sept. 29, 1985, and Spectrum shut down entirely a week later after broadcasting the Twins’ final regular-season game.

After Spectrum closed, KTMA-TV Channel 23 returned to being a free over-the-air station, so viewers no longer needed a decoder box.

In 1986, KTMA-TV switched to independent programming, showing reruns, movies, and local shows.

KTMA-TV premiered “Mystery Science Theater 3000,” or MST3K, Thanksgiving Day, Nov. 24, 1988.

KTMA-TV filed for bankruptcy in 1989, was sold, and was rebranded as KLGT in 1992.

By the late 1980s and early 1990s, cable companies began replacing many long coaxial trunk and feeder cables with fiber while keeping coaxial cable for the final connection to neighborhoods and homes.

In a traditional cable system, trunk cables carried signals over longer distances through the main network, while feeder cables branched out through neighborhoods toward customer drop lines.

These hybrid fiber-coaxial systems increased capacity and reliability, helping cable systems evolve from one-way television delivery into broadband internet, voice, and data networks.

By the late 1990s, many cable systems had become two-way networks capable of handling video, internet, and phone services.

After the 2009 digital conversion, KTMA-TV’s old analog UHF Channel 23, 524 to 530 MHz, was no longer the station’s actual broadcast frequency.

WUCW, the successor to KTMA-TV, still appeared to viewers as Channel 23, but that was only the on-screen, or virtual, channel number. Its actual digital broadcast signal was carried on UHF Channel 22, 518 to 524 MHz.

By 2026, voice and video were largely Internet Protocol, or IP, applications, sent as data packets over fiber, coaxial cable, satellite, and wireless networks alongside other digital content.

As of May of this year, channels 14 through 36, spanning 470 to 608 MHz, make up the current US UHF television range.

John Walson Sr., born John Walsonavich, died March 27, 1993, at 78.

Leroy “Ed” Parsons died May 1, 1989, at 82.

In the late 1940s, cable TV provided only a few broadcast channels.

Today, viewers have access to hundreds of channels and streaming services through modern coaxial, fiber, satellite, and internet-based networks.

An AI-assisted photographic collage by Mark Ollig illustrates the historical 

evolution of cable television, moving from early community antenna systems 

and rooftop arrays to analog sets and decoder boxes. The work, created through

OpenAI image generation, depicts the transition of the industry into modern

AI-generated collage showing the evolution of cable television from early community antenna systems.