To Download Credit Application: Fill & Download for Free

Kodak has become a byword for failure — generally believed to be failure to innovate, however the real story is a lot more complex and nuanced than that. While Kodak dominated the film industry, in fact Kodak also invented digital imaging, and pretty much led the field for innovation for 30 years.So what went wrong?Credit for the invention of the original imaging chip — the Charge Coupled Device (CCD) — actually belongs to Bell Laboratories researchers Willard Boyle and George E. Smith. The pair had conceived of a simple shift register device that simply transferred electrical charge from one cell to another like a bucket brigade, and which would become the basis of the CCD image sensor.A part of AT&T’s 1956 consent decree with the Justice Department required Bell Laboratories to openly share it’s research and license it’s patents. As a division of AT&T, Bell Labs itself produced nothing but research — AT&T products were traditionally manufactured by Western Electrical, which was another division of AT&T. Every year, Bell Labs would hold a presentation to all of it’s technology licensees at an auditorium at Bell Labs.It was a young Bell Labs scientist called Gil Amelio, responsible for some of the fundamental work behind the CCD patents, that became responsible for sharing the CCD research. In the audience for his 1971 presentation on CCDs was John Atala of Fairchild Semiconductor. Amelio also presented CCDs at an International conference in Scotland, and later to Toshiba and Sony in Japan, amongst others.Following an invitation from John Atala, Amelio followed the opportunity to pursue his research all the way through to a product at Fairchild. There, he led the team that would product the World’s first imaging sensors — a linear 500 ‘pixel’ device, and a 100 x 100 pixel CCD array, first announced by Fairchild Semiconductor on November 22, 1973. Collectively, Gil Amelio and the Fairchild team ended up taking numerous patents on CCD design and manufacture, which would ultimately be almost the only money Fairchild ever made on the technology.The first commercial application for CCD technology was a 1,728 element linear sensor developed by Fairchild for Xerox fax machines.Originally developed at the request of the U.S. Navy, it was Fairchild’s first crude 100 x 100 CCD sensor array that seems to have piqued the most interest, however.In December 1974, a Kodak executive named Gareth A. Lloyd was interested enough to ask a freshly recruited 23-year old with a Masters Degree in Electrical Engineering, “Why don't you see if you can do some imaging with this device?”Steven Sasson was starting from scratch in uncharted territory. He ordered two of Fairchild’s Type 201 100 x 100 pixel CCD sensors. These chips were so early in their development cycle that the instructions included in the box were hand-written. It would be another six months before academic papers started to appear on the use of these devices. (It wasn’t until June 1975, for example, that the Jet Propulsion Laboratory (JPL) published one of the first academic papers that explored the use of a CCD for imaging).Steven Sasson later recounted in an interview with journalist Stewart Wolpin, “I started to look around how these things worked and read whatever I could about them. Then I thought about how to capture images, then maybe building a camera. It became really came clear to me that if I could digitise an image, freeze it and hold it and analyse it and store it and look at it, that was sort of the goal.”There was no blueprint, no prior art that Sasson could follow. Taking image capture as his starting point, he used a 4-bit Motorola analogue-to-digital converter, and a dozen 4-kilobit dynamic ram (DRAM) chips to capture the meagre output of the CCD chip. He worked to coordinate the clocking of the CCD chip wit the A/D converter and memory chips. The CCD sensor needed a 20V supply, which dictated 16 AA batteries, while the logic circuitry needed only 5V. All of the development was done using just an oscilloscope to check that the CCD circuitry was functioning correctly, but having captured impulses from the CCD, he still had no persistent storage. Sasson added the lens from a Kodak XL55 Super-8mm movie camera, a Memodyne Model 300 data cassette recorder, and effectively created the World’s first portable digital camera.By December 1975, Steven Sasson had captured, stored and displayed the first digital camera image — of lab technician Joy Marshall, working down the hall at Kodak’s apparatus laboratories.Unlike Kodak’s consumer film cameras, this was no polished product. It looked like something made from Meccano, weighed 4 Kg, and was about the size of a toaster. “It's an odd-looking beast," Sasson himself noted. “Odd today and really odd in 1975.” A lot of the internal connections were wire-wrap connections on bread-board circuits, and as often as not, these connections failed, rendering the camera useless.Like Kodak’s film cameras however, this crude device could capture and store 30 images, although each image was a 4-bit, grayscale image, just 100 x 100 pixels, and took 23 seconds to load to tape. Kodak was granted the first of it’s patents for an “electronic still camera” in 1978, but outside Kodak’s apparatus lab, where Sasson worked, the project was still top-secret.Kodak was not the only company interested in the CCD.By the time Sasson had displayed his first images, there had been an IEEE conference on CCDs in San Diego, CA. that October, that had attracted no less than 48 presentations. Fairchild, Bell Northern, Intel, Sony, Texas Instruments, NEC, Matsushita Electrical (Panasonic) and RCA were all actively pursuing CCD development.By June 1975 RCA had unveiled an experimental ‘tubeless’ TV camera, that featured a CCD sensor (SID 5000) with 512 x 380 pixels. The camera utilised 3 of these sensors to deliver NTSC-standard colour output.In December 1976 a satellite named Kennan was launched containing the first the first real-time imaging device in space, based on linear CCD sensor technology. Because the satellite is always in motion, a linear array enabled the satellite to essentially “scan” the surface of the Earth. The Lawrence Livermore Laboratory had also applied RCA’s CCDs to X-Ray detection.After seeing Gil Amelio make his presentation in Japan, Sony invested years developing their own CCD sensor capability. In 1981 Sony launched the Mavica — a still camera with 570 x 490 pixel resolution. The Mavica recorded images onto a 2” floppy disk, and could replay them on a TV, but it was still not a digital camera. A CCD sensor is an analogue device, with electrical output proportional to the light falling on the sensor, and the Mavica recorded analogue signal on it’s disks. Sony would not have a digital Mavica (FD5 model) in market until 1997.Canon introduced their RC701 and Nikon launched the QV 1000C in 1988, but both of these cameras, like the Sony Mavica, were analogue cameras writing to special 2” floppy disks.Steve Sasson’s crude digital camera prototype was literally still years ahead of the competition.By 1987, Kodak had perfected a 1-megapixel (1,320 x 1,025) sensor, called the M-1, and by 1988 Kodak’s Federal Systems Division had adapted it to a 35mm Canon F1 film camera.Adapting a standard 35mm camera to a digital sensor was relatively easy. The focal plane (the point in the camera on which the image is focused), is easily exposed by opening the back of the camera, and the standard flash contacts signal the point at which the shutter is wide-open. Kodak replaced the standard camera back with a CCD panel, sandwiched to a thermo-electric cooling panel, to improve the signal-to-noise ratio. The camera was then connected by a ribbon cable to the digital mechanism. That mechanism was a breadbox-sized device that contained, among other things, a lead-acid battery, 100Mb hard drive and controller — hardly a practical field camera.Attaching a CCD sensor to a mature camera like a Canon F1, with it’s complement of lenses and attachments, made perfect sense. It remained a one-off proof-of-concept experiment however.Kodak tried again with a more compact version, once again based on the Canon F1, in 1989. Using solid-state memory, and avoiding the thermoelectric cooler, made for a much more compact and rugged system, although it was still a two-device tethered system, and once again remained a proof-of-concept project. Just two devices were made.An all-in-one device called the ‘Hawkeye’ followed, based on the Nikon F3 camera. It was Kodak’s first non-tethered device, based on solid state memory, and it appeared no bigger or bulkier on the F3 than Nikon’s optional MD-4 motor drive.The ‘Hawkeye’ at least appeared field-ready, although there was another version, called the ‘Hawkeye II’ that featured the modular, tethered design of the first FSD prototypes. The final Hawkeye prototype featured Kodak’s new M3 sensor, providing an RGB colour image for the first time. Only a handful of any of these cameras were ever produced, however one of these machines was carried aboard the space shuttle Atlantis, in November 1991.In an attempt to consolidate all of the work being done by the Federal Systems and Professional divisions of the company, by 1989 Kodak had established an Electronic Photography Division. The intent was to coordinate research and development and create products that Federal and Professional divisions could take to their respective markets.The first fruit of the combined effort was a product called the D-5000, or ECAM.Developed by a team that included Steve Sasson and Robert Hills, the D-5000 was actually the first digital SLR camera to look and function pretty much like a modern DSLR. The D-5000 used Kodak’s own 1,280 x 1,024 pixel colour CCD sensor, and offered storage with removable flash memory cards. The D-5000 had access to the full range of Pentax K-mount lenses and accessories, including through-the-lens flash photography, but appears to have been another prototype and was never officially marketed.Kodak’s last modular, tethered camera hit the market as the DCS 100, in 1991. Based on Nikon’s F3 camera body, which could still be used with Nikon’s MD4 motor drive in tethered configuration, the DCS 100 was based on Kodak’s latest M3, 1,280 x 1,024 CCD sensor. The DCS 100 could be purchased in any of six model variants, in colour and monochrome versions, with or without a JPEG compression board included.The physical bulk of the system was dictated by components such as sealed lead-acid batteries, and a 200 Mb hard disk drive, good for 1,609 uncompressed or 600 compressed images. Connectivity was via a 50-pin SCSI port on the unit, which also supported NTSC video output.Kodak boasted that professionals could switch to digital without switching cameras, but the tethered systems alone were a lot more equipment than most journalists were accustomed to carrying around — and that was just the camera! To make this system a palatable option for photojournalists, Kodak sold them with a nylon courier bag, and a hard travel case. By the time the system was deleted in 1994, just 987 units had been sold.For customers coming of age in a Photoshop-and-Pagemaker world, Kodak launched the DCS 200 at the annual MacWorld event in 1992. Based on Nikon’s N8008 camera, which was then the cheapest Nikon with a removable back, the maturity of Kodak’s capability in digital photography enabled them to take the DCS 200 to market in less than a year, surprising even Nikon, who were not aware that their N8008 had even been adapted in this way.Once again, the technology dictated the size of the package, and the new 2.5” hard drives supported 80 Mb of storage, or 50 images, on-board below the adapted camera. Rechargeable batteries gave way to AA batteries in both the camera body and in the camera back, and 3,240 DCS 200s in various configurations are thought to have been sold by 1994.The most important feature of the DCS 200 was actually it’s architecture. With only minor modifications, the DCS architecture became the basis for the NCS 2000 series, DCS 4XX and EOS DCS X systems. The built-in hard 2.5” drive was substituted for PCMCIA-format slots that supported the new 1” micro-drives as well as emerging flash memory options. The same basic architecture supported CCD sensors from 1.2 to 6 megapixels, with 12-bit A/D conversion and SCSI support.In another development, Nikon and Kodak finally formed a closer technical partnership, which would spawn numerous new models, based on Nikon camera bodies. The new Kodak DCS4XX series digital cameras were all based on Nikon’s new N90 body, as was the NC 2000 model, produced exclusively for Associated Press members.In total the DCS410/DCS460 range was responsible for about 5,000 unit sales, and was Kodak’s most successful foray into digital photography up to that time. If those don’t seem like massive volumes, it’s worth remembering the unit price of these cameras: the AP version sold for US$17,500 each. (At the time a standard Nikon N90 camera body retailed for just US$1,090).The DCS 460 debuted Kodak’s M6 CCD — a new 6-megapixel sensor that made the DCS460 “the World’s highest resolution portable digital camera.” The most famous DCS 460 is undoubtedly the one that astronaut John Glenn carried into space aboard the space shuttle Atlantis in March 1996. That hardly justified the US$35,600 list price though, and by 2000 when the model was closed out it retailed for US$2,500.Kodak’s Federal Systems Division adapted DCS 4XX cameras with GPS technology, and even repackaged the electronics for underwater configurations and ‘rugged’ applications.In another development of the DCS 200 architecture, a version was produced for Canon, who could hardly ignore the opportunity to offer their own customers digital technology. Canon adapted their EOS 1N bodies to take Kodak’s NC2000 camera back, and created the EOS DCS 1, 3 and 5 series cameras, ranging from 1.2 to 6 megapixels, in monochrome, colour and infrared versions. Canon also marketed the EOS DCS 1 and 3 versions, and all in all, about 1,000 of all variations of this model were sold.Better than anyone, Kodak understood that professional segment of the market relied on larger medium-format cameras, and introduced a digital back for medium format cameras like the Hasselblad. Really a modified DCS 460, the DCS 465 sold a modest 200 units.At Tokyo MacWorld in 1994, Apple launched the QuickTake 100 — ostensibly the first ‘consumer’ market digital camera. The QuickTake 100 offered 640 x 480 pixels (0.3 megapixels!) of 24-bit colour resolution, could store just 8 images in it’s 1Mb memory, and download to a Macintosh computer via an AppleTalk serial network at a maximum of 230 Kb/sec!The QuickTake 100 was actually manufactured for Apple by Kodak, and while Kodak might initially have been less interested in the consumer end of the market, the AppleTalk connection with Macintosh computers rounded out Apple’s desktop publishing vision. The SD-video 640 x 480 pixel resolution seemed light even then, but for US$749 it represented good value for money, and there was nothing else of that quality in the consumer market. Kodak later brought out it’s own almost identical DC40 model, with a 756 x 504 pixel sensor.‘Hybrid’ technology might have been a blind spot for Kodak. For all their innovation in digital technology, Kodak didn’t appear to have considered ways in which digital technology might have augmented the traditional film business. Digital technology was about to transform and improve the traditional photography product in a more subtle way. If digital photography threatened film sales, photo-processing was actually where the silver was hidden. Processing traditionally earned the industry more than twice as much as film sales.For years the byline of Kodak advertising had been George Eastman’s original promise, “You press the button, we do the rest.” Kodak had perfected a vertically integrated imaging system — cameras, film, developing and printing. The brand implied an assurance that this system would produce the best results, and there was indeed some truth to that. Independent tests proved that Kodak, Fuji and Agfa had all optimised their film and photographic paper to complement the subtle colour biases in each.It was Fuji that first introduced the digital minilab in 1996.The Fuji Frontier SP-1000 used a conventional film processing system and a revolutionary new digital printer. Fuji’s Frontier digital minilab succeeded in breaking Kodak’s virtual monopoly, and was ultimately the innovation that commoditised both film, and retail processing.The Frontier SP-1000 minilab comprised a conventional film processor, a digital scanner and an LED-based photo printer. The introduction of digital technology into the process meant that images could easily be previewed and corrected for basic image issues such as exposure and colour balance. Negatives were pre-scanned and the results displayed on a monitor so that the operator could correct colour balance and exposure settings before the images were sent to the laser printer, which used LED lasers to expose conventional photographic paper. The exposed paper was then developed in a conventional ‘wet’ photographic printer.Among other features, Fuji’s Frontier recognised the DX barcode below the sprocket holes of almost all film, and adapted the scanning and output profile for each film type.Fuji didn’t just offer competition, the Frontier minilab attacked Kodak’s core brand values. The Frontier minilab was the stroke of genius that enabled Fuji to better even Kodak’s brand promise, adapting to not just the film type, but compensating for the exposure and colour characteristics of an individual image also. The Fuji Frontier promised better results from almost any brand of film, under almost any conditions, lowering the ‘rejection rate,’ making for happier customers.The new, smaller scale and lower operating costs of the next-generation mini-labs also enabled Fuji to broaden their distribution of retail outlets in a way that nobody seems to have anticipated. Fuji quickly achieved mass market penetration, offering 60-minute in-store processing, soon picking up distribution deals with the likes of Wal-Mart and Walgrens pharmacy chains. Fuji soon dominated the OEM and generic film sales.At a shareholders meeting in 1978, Kodak had demonstrated technology that digitised film to produce high-quality prints, and was not far behind Fuji with a digital minilab of their own, but were naturally wary of attacking their existing distribution chain. Fuji of course had nothing to protect and everything to gain, and ultimately it was Kodak’s protected customer base that was under attack.Even in the digital era, for years there was nothing that came close to traditional photographic printing for the resolution, quality or traditional look and feel of it’s prints. Twenty years later, photographic prints are all digital — either from scans of film, or digital source files. Traditional ‘wet’ photographic printing has all but given way to inkjet technology in so called ‘dry’ minilabs.Inevitably, digital minilab capability became all about software. Just one company, Noritsu, finally ended up making minilab hardware for Kodak, Agfa and Fuji!If the future of film was starting to look a little ‘pixelated,’ Kodak had far from thrown in the towel. In 1995 Kodak applied to the Court Of Appeals to terminate both the 1921 and 1954 consent decrees that had restrained the company in the market. The purpose of the original decrees had long since been served. Competition had become well-established, and Kodak now held a much less than commanding share of the marketEven the Appeals Court agreed that Kodak no longer had any real market power. By 1995, the Court noted, Kodak was just one of at least five companies, including Fuji, Konica, Agfa, and 3M, that had all established successful photographic businesses in America. By that stage Kodak’s share of the worldwide market was 36%, and Fuji had raised it’s stake to 34%.In America, Kodak still commanded a 67% market share, against Fuji’s 12%, although Fuji also supplied house brand product for K-Mart and Wal-Mart, who were thought to retail half of all film sold, albeit at significant discounts.The strength of Kodak’s brand meant that Kodak still commanded 75% of film sales by value, but Kodak’s dominance, and it’s price premium, had clearly been decimated. In the USA Kodak’s “small and declining market premium” had been slowly eroded to only an average of 4.5% over Fuji, but the court believed that was evidence of Fuji’s aggressive discounting strategy, rather than any perceived strength in the Kodak brand.The court found that Kodak was often unable to achieve any kind of premium in the mass market, whereas just five years previously they had achieved an average of 6.3%. The court believed that if Kodak were to raise it’s US prices by just 5% they would likely lose a further 10% market share. Elsewhere, it was Kodak that was forced into a much more aggressive posture. Around the World, Kodak product was cheapest in Japan, where the company held only 10% market share against a dominant Fuji.After nine days of deliberations, the Court unshackled Kodak and terminated the consent decrees, but that was really only a pyrrhic victory for Kodak at that point.It looks like a distraction now, but in 1996 there was yet another photographic format in market. The Advanced Photo System was an initiative between Kodak (Advantix), Fuji (Nexia), Agfa (Futura), Konica (Centuria), Nikon (Pronea), Canon (EOS IX) and Minolta (Vectis), based on a smaller frame size, with a correspondingly smaller, lighter cameras. Beginning in 1991, the thinking apparently went that the 35mm format had become ‘stale,’ and that some innovation might stimulate demand for film, for cameras and for photography in general.This was a US$500 million dollar initiative, that somehow remained the industry’s best-kept secret for almost 5 years. The project was even awarded the Project Management Institute “Project Of The Year Award” in 1997.The secrecy was critical. As depressed as the photographic industry may have been, news of a new format could have stalled the market entirely.The 24mm-wide APS film stock covered only 58% of the image area of a conventional 35mm film, which in itself impacted on quality enough that it never captured the attention of serious users. A lot of “innovation” in film had failed on exactly the same point— think the 126 or 110 formats, or Kodak’s disc-format camera. Reduced negative sizes did not deliver comparable, or even adequate quality prints.APS came with some novelty features however. APS film cassettes recorded camera data which could be used in photo processing, or even printed on the back of the print. APS film cassettes had no leader, and automatically threaded the film through the film transport mechanism of the camera. When the negatives had been processed, they were actually returned to the film cassette for safekeeping, and the prints indexed with a serial number that corresponded with the cassette and frame number for re-printing.Canon, Minolta and Nikon all created new families of lightweight, compact SLR cameras to cater for the new format, sometimes including smaller, lighter lenses. Most manufacturers, like Canon and Nikon, produced relatively conventional scaled-down SLR cameras, as well as smaller, lightweight ‘novelty’ cameras, and some pocket, point-and-shoot models as well. There was no shortage of hardware options to support the new format.The APS format had all the hallmarks of a category breakthrough, but the cameras were neither functional or rugged enough to engage serious photographers, or novel enough to attract new customers to photography. APS cameras predominantly found favour with point-and-shoot customers who appreciate their compactness and light weight.The bigger issue of course was that digital cameras had started to capture mindshare.Investing in film at this point must have seemed like a regretful spend. Nikon ceased production of their Pronea in 2000, Canon stopped production of their EOS IX in 2001, Kodak stopped making even simple APS cameras in 2004, and both Fuji and Kodak discontinued APS film production in 2011.Meanwhile, Kodak had well and truly exploited their proven DCS 200 digital architecture. After four years in market, a lot had been learned, and it was time for an overhaul.Kodak engineers decided that new base architecture would be based on CCD sensors as before, although brand-new sensor designs and materials improved the colour balance, and a PowerPC chip was added to handle JPEG processing on the fly. The bulky and unpopular SCSI interface made way for the new, faster and more compact IEE 1394 ‘FireWire’ standard connection. Other changes included a colour LCD display with a new graphic menu.First cameras to adopt the new architecture were Kodak DCS 5XX and Canon EOS DXXX models. Canon supplied modified EOS 1N bodies with the film transport mechanism removed, and both Canon and Kodak marketed their respective models. The 2-megapixel Kodak DCS 520 and Canon EOS D2000 were identical, for example, as were the 6-megapixel DCS560 and Canon EOS D6000 models. Kodak launched the DCS 520 at the annual PMA show in New Orleans in February 1998. If the price of digital technology was falling, the US$14,995 list price of the DCS520 was still well beyond the reach of the mass market.Also in 1998 Kodak launched the DCS 315. Based on Nikon’s smaller, lighter Pronea APS-format film camera, the DCS 315 was a 1.5 megapixel machine that was the first digital camera to feature an on-board LCD image display, and built-in JPEG image processing. Like other DCS models, the 315 was based on a simple substitution of the Pronea camera back for the Kodak module, which more than doubled the bulk of the original Pronea. The Pronea was the functional equivalent of Nikon’s F70 camera, intended for a different price point in the market, but the DCS 315 still retailed for US$4,995 on release. The DCS 330, released the following year, featured 3 megapixels, but did not support JPEG compression.The DCS 315 and DCS 330 deserve mention now only because they were the forerunners of the modern APS-C format digital cameras. Nikon, Canon, Fuji and most others manufacturers have long produced lightweight digital SLR cameras that take advantage of smaller CMOS sensors, in the same way that the Nexia, Pronea and EOS IX took advantage of the APS format. While these crop-sensor cameras lack the wide-angle or low-light versatility of the full-frame cameras like Nikon’s D5 or the Canon 1D Mk X, their light weight and smaller profile found favour with a new market of amateur photographers.At the 1999 PMA exhibition in Las Vegas, Kodak launched their all-new DCS 620 camera.For the first time, the DCS 620 was a properly integrated, all-in-one digital camera, based this time on Nikon’s F5 film camera. Nikon supplied an F5 camera body adapted to Kodak’s design, and the electronics were fitted at Kodak’s Rochester, N.Y, facility. Already a staple of the press and professional markets, the rugged and sophisticated F5 gave Kodak’s digital cameras a new legitimacy. Kodak even promoted the DCS 6 Series as the, “Affordable, high-performance NIKON F5 SLR.”It seems likely that Kodak rushed their DCS 620 to market to try and capitalise on what demand premium they could, before Nikon and Canon, with their obvious cost advantages, entered the market on their own. It wouldn’t have taken a crystal ball to see that coming. It can’t have escaped Nikon's attention, for instance, that their F5 could be bought retail for US$1,850 while the DCS 620 was US$14,995.In a similar vein, on the 31st of January 2000, Fuji announced their FinePix S1 Pro DSLR.The FinePix was based on Nikon’s F60/N60 film camera body, and featured Fuji’s own CCD sensor that cleverly extrapolated a 6-megapixel image from a 3.4-megapixel CCD sensor. The FinePix Pro was rated with sensitivity from ASA 320—1600, and produced JPEG or 8-bit TIFF files in-camera.The execution of the FinePix also looked rushed. It needed no less than 3 battery types — two CR123A Lithium batteries to power the camera, 4 x AA cells to power the digital electronics, and a CR2025 button cell to maintain camera settings like date and time.Users rated the FinePix for it’s natural results, but ultimately it seems that Fuji expected to sell on price. Unlike any of the other digital SLRs on the market at that point, the FinePix Pro was unashamedly plastic. The first announcements pegged the price at US$4,000, but on launch it was US$3,500 and within months of launch could be bought for US$3,000.Few people doubted that Nikon would have their own digital camera in market, and it had become a question of when, not if. At the 1999 PMA exhibition, Nikon finally offered a glimpse of the future, offering a non-working mock-up of the camera that would become the D1. The machine was locked in a glass display case, alongside F5 and F100 models, with whom it shared an obvious and reassuring familial resemblance.Nikon representatives on the exhibition stand were giving nothing away with regard to the specification of the final product, or a delivery date, but the intention could not have been more clear. Nikon would soon have their own, likely more affordable, ‘digital F5’ in market in the near future. Nikon said that it would continue to honour it’s agreements with Kodak, but the announcement of the Nikon D1 effectively stalled Kodak’s market.If Nikon’s PMA exhibit had just the one talking point, Kodak’s exhibition stand was a little busier. Aside from the DCS 620, Kodak seemed intent on taking the fight to Fuji, and exhibited their RFS 3570 film scanner, a digital multi-printer production system and the second-generation Professional LED II Production System. Kodak also demonstrated some increasingly mature software products. such as ‘Kodak Portraits & More.’Finally however, Nikon ended all speculation. On the 31st of August 1999, at the annual Seybold desktop publishing conference in San Francisco, Nikon unveiled details of their much-anticipated D1. Offering 2.74-megapixels, 50 milliseconds shutter latency, a 4.5 frame/second shooting rate, and a 21-frame buffer, the D1 appeared to be a mature camera, usefully differentiated from it’s competition. First deliveries of the new camera were anticipated to be in September.Elsewhere, an event so inconspicuous it likely went unnoticed, actually marked the beginning of a trend that would eventually transform all photography forever.Friends Kamran Mohsenin and Lisa Gansky decided that digital photography had the potential to change the way that people could store, view, edit, share and print their photos. They launched their new, on-line service, called Ofoto in June 1999.Digital users could upload their photos directly, using Ofoto’s free software, while film users could mail in their rolls of film and have them processed, scanned and uploaded for just US$3.95 per roll. Users could edit their photos, or customise them with Ofoto’s digital frames and cards. Conventional silver-halide photographic prints were available for anything from US$0.50 to US$20 each, depending on print size, and CD-ROMs were available for US$10 each. The service was free, provided users made at least one purchased every 12 months.Ofoto piqued the interest of investors, raising US$16 million in venture capital in 2000, and then US$41 million in another round of funding, before being acquired by Kodak in April 2001. By that point, Ofoto had 1.2 million registered members and employed 121 people. An independent analysis of on-line photo services by research company ARS Inc. rated Ofoto’s service best in terms off ease of use, service and price competitiveness. Ofoto was even endorsed by Microsoft, being one of three photo processing services bundled with WindowsXP.At least two of Ofoto’s competitors were other Kodak initiatives, such as Kodak-sponsored "You've Got Pictures,” offered via America OnLine, Kodak PhotoNet Online, and the Kodak Picture Center online at CVS.com. According to Neilsen research, these sites were averaging a total of 782,000 unique visitors in the three months leading up to the Ofoto acquisition.In a press release, Willy Shih, president of Kodak's Digital and Applied Imaging unit, said, “The acquisition enhances our leadership in the growing market for online photo services. By combining Kodak's and Ofoto's technology, marketing and distribution assets, we will be able to deliver the most comprehensive, easy-to-use online photography services to customers and consumers. This will accelerate Kodak's growth and drive more rapid adoption of online photography. This move will result in greater speed to market for new features and reinforce Kodak's position as the market leader.”Kodak re-branded the service 'Kodak Gallery,’ by 2003 had expanded the service to the United Kingdom, and added Kodak’s Perfect Touch image optimisation software to the options available to users.Kodak also appeared to realise the connection between connectivity and their digital services, and in 2005 launched the WiFi-enabled EasyShare One camera, although the point-and-shoot model didn’t last long in Kodak’s line-up.Ofoto and it’s peers were simply a little ahead of their time. Today of course there are more photographs created, edited and distributed on-line than by any other means. Smartphones and increasingly also cameras are being network-enabled, able to capture, edit and upload photos virtually in real-time.Following Kodak’s divestiture in 2012, it’s on-line services ended up being a part of Shutterfly.Elsewhere too, Kodak was doubling down on digital. Intending to get out of the professional market at some point, Kodak was not yet ready to give up. Late in 1999, the DCS 5 and 6 Series received a firmware upgrade that made more intuitive sense of the user interface. (Kodak would continue issuing firmware updates for their DCS range until they exited the SLR business in 2005).By 2001, the DCS 6XX series was altogether superseded by the 7-Series. The DCS 7-series was effectively Kodak’s third generation of digital technology. The entire architecture had been optimised to get the sensor closer to the circuitry, lowering the signal-to-noise ratio and increasing the camera’s speed and sensitivity to light. A Texas Instruments digital signal processor had been added to complement the upgraded 75 MHz PowerPC chip, and dual PCI card slots supported a variety of storage options. The 7-Series was supposed to support JPEG compression, but was finally shipped with only RAW format support.The DCS 720 was a 2-megapixel model, but the real breakthrough was the DCS 760, which featured 6-megapixels, and listed at just US$7,995. The 7-Series wasn’t the only news. Kodak shipped it’s brand-new Professional DCS software, offering photographers much greater flexibility and control over their raw images. The DCS 760s earned an enviable reputation for it’s reliability and image quality, and second-hand DCS760s still trade at reasonable values on the internet.To some extent Kodak seemed to be playing to it’s heritage, offering a ‘photographic system’ — the camera, software, and it’s legacy network of processing and printing. The DCS 760 finally put digital photography within reach of a whole new market segment, and more importantly, was much closer to the price point that Nikon was expected to hit when their new D1 finally hit the market.While Kodak’s digital technology had been revolutionary, the key to great images was still actually the camera, in which Kodak had literally no IP or expertise. Adapting Canon and Nikon models had been fast and simple to do. Kodak was able to leverage the installed base of lenses and accessories for these cameras, and address the professional market most likely to value their digital technology.No doubt Nikon and Canon both learned from their exposure to Kodak’s technology, but the honeymoon couldn’t last forever. The drive to digital was absorbing capital in Nikon and Canon’s traditional customer base. Every camera Canon or Nikon sold to Kodak, was absorbing perhaps 7-10 times as much revenue from their own customers. Ultimately both Nikon and Canon would have to plug the leak in their market with digital cameras of their own.The Nikon D1 and Canon 1D were intended to be those cameras, but Kodak was proving to be a determined competitor.In September 2002 at Germany’s Photokina exhibition in Cologne, Kodak debuted the DCS Pro 14n camera. At 13.89-megapixels, the Pro 14n was the highest-resolution digital camera available to that point, and only the third camera ever to offer a full-frame 24 x 36mm sensor (after Canon’s 1Ds and an unsuccessful Contax camera).Based on Nikon’s N80 camera, the Pro 14n was a departure from all of Kodak’s previous digital cameras in that it featured a CMOS sensor, developed and manufactured exclusively for Kodak by the Belgian firm FillFactory. The Pro 14n offered a 2-frame/second exposure rate, an 8-shot burst (later upgraded to 18 shots) and could transfer data via it's Firewire interface at 12 Mb/ps.Kodak also finally offered qualified on-board support for JPEG compression. The issue with JPEG compression is that it achieves reduced file sizes at the expense of image data. That in itself is not a problem, but it destroys the latitude available for adjustment of the images in post-processing. Kodak offered an extended-range (ERI) version of JPEG that preserved a full two stops of latitude in exposure and colour control. From the Kodak press release, “The ERI-JPEG format provides professional photographers ease-of-use of JPEG files with the image quality and colour/exposure control of Kodak's highly regarded DCR format raw camera files, to create the best quality images.” The intent was clear — Kodak was well aware of the issues associated with JPEG files, but also anxious to offer some of the advantages of image compression.Kodak had also seized on the flexibility of it’s software-based digital architecture. “Because the features of the DCS Pro 14n are based on Kodak Professional firmware and not hard-wired within the camera, the camera can be enhanced and easily upgraded with free firmware downloads from Kodak Professional. Free firmware upgrades essentially give photographers a "new" camera whereas other manufacturers require the purchase of an entirely new camera system to receive the latest enhancements.” Previous upgrades had added menus, additional languages and so on to the user interface of the DCS 5, 6 and 7-Series cameras.The DCS Pro 14n was a game-changer. Kodak led the market on innovation — and finally also, on price. The Pro 14n debuted at US$5,000, it’s nearest competition being Canon’s 11-megapixel 1Ds that launched at US$8,995. If the standard N80 was a modest camera, lacking the speed, exposure and focus options of the F5, there was nothing that could touch the DCS Pro14n.By 2001 there were reportedly 15 million digital cameras in America, and in 2003, sales of digital cameras finally outstripped film cameras. The traditional film-and-paper photographic industry generally is thought to have peaked at about this time. In September 2003, Kodak announced a new strategy, intended to “broaden our digital presence in consumer, commercial and healthcare markets. These three ‘pillars’ represent the foundation of our business, and are areas where Kodak already has a base from which to grow. We also announced we would select future business opportunities, notably in the display and inkjet markets, that build on our core competencies and our solid base of intellectual property.”By 2004 the Pro14n had been superseded by the Pro SLRn, still based on the Nikon N80, but with much improved low-light sensitivity boosted all the way up to ISO 1,600, and exposures up to 60 seconds possible. The really good news for Pro14n owners was that their cameras, for US$1,495, could be upgraded at Kodak to SLRn spec, losing only the new power management and some of the processing speed of the new SLRn.The key to the SLRn was a re-designed sensor, with a dye-based colour filter rather than the dichroic types common to other sensors. The change helped to deal with the chromatic aberrations that were sometimes visible, with particularly wide-angle lenses.Kodak’s relationship with Canon seems to have soured following the release of the EOS 1D. Canon’s original contract with Kodak is rumoured to have been exclusive, and actually precluded Canon from developing their own DSLRs. Whatever the truth of that, Kodak’s Canon-compatible equivalent of the Pro SLRn, called the SLRc, was based on Sigma’s SD9 camera, and launched in 2004.Kodak’s Pro SLR cameras received rave reviews. The only real criticisms concerned the bulk of the digital machines, and the control layout (particularly the SLRc, which was not as familiar to Canon users as the layout of the N80 was to Nikon’s customers). It would be 2007 before Nikon would have a full-frame camera in the market, and so the Pro 14n and SLR models had been enthusiastically received by Nikon owners. In one review, CNET noted, “Superb resolution; 35mm-size sensor obviates focal-length conversions; significantly improved high-ISO performance; improved battery life; powerful software and user-installable firmware.” Their conclusion; “this dSLR offers outstanding resolution to portrait, school, commercial, and architectural photographers who own Nikon F-mount lenses.” Layout issues aside, the image quality of the SLR/c also compared favourably with Canon’s more expensive 11-megapixel 1Ds.By the third quarter of 2005, Kodak’s digital products earned more money than it’s film products for the first time. By then however, on May 31st, Kodak had already announced that it would cease manufacture of it’s Pro digital cameras.It’s an interesting decision. Having created a competitive product, and at a price that seemed well received in the market and which promised some real sales volumes, it seems like a curious time to pull the pin. The Pro SLR range were ‘halo’ products that validated Kodak’s leadership in digital technology. It would be another 2 years, for example, before Nikon had a full-frame sensor in market, with their D3 model — and even that was only a 12.2-megapixel machine. It looks now like a reaction to Kodak’s first-ever losses.There were a number of factors in play however.For almost 20 years, Kodak had developed and manufactured it’s own CCD sensors, but by the time Kodak started development of the DCS Pro range, it was clear that the next ‘leap’ in performance would need to be based on CMOS technology. While Kodak led the design of the sensor for the DCS 14n, manufacturing it’s own CMOS sensors would require a quantum leap in chip manufacturing technology that Kodak was not prepared to make. More pixels means finer manufacturing tolerances and bigger die sizes, and a lot of semiconductor companies had already failed to deliver CMOS technology reliably in volume. Kodak partnered with FillFactory and IBM to manufacture CMOS sensors to it’s designs.It wasn’t a foregone conclusion that Kodak would need to manufacture their own sensors to remain competitive. Nikon for example, has only ever made a few of their own sensors, typically relying instead on Sony sensors.Neither was it necessarily a deciding factor that Kodak didn’t make their own cameras. Traditionally, some of Kodak’s most famous 35mm cameras, like their Retina and Retinette, were assembled in Germany, with key elements such as shutters and lenses sourced from third parties. Nikon’s camera bodies were a relatively small part of the overall price of a digital camera. Just $400 over the counter would get you an N80, when the DCS Pro SLR sold for $5,000, although Nikon also manufactured the specially adapted magnesium chassis that Kodak needed for it’s electronic components. Sigma started making cameras primarily to sell their lenses, and third-party manufacturers such as Seiko and Copal already manufactured the key shutter mechanisms for other manufacturers.With the benefit of 20/20 hindsight, pulling the pin on the Pro SLR range seems premature, but Kodak’s heritage is in the mass consumer market. The high-end professional market is a very small part of the overall photographic market, and in the digital age increasingly less likely to print anything. The real silver was buried in the mass market, and Kodak’s decision to support it’s retail distribution network with affordable cameras should have supported it’s processing chemistry and photographic paper business.Far from ignoring digital technology, Kodak seemed intent on continuing to lead. An “Image Sensor Solutions Group” was created to concentrate on sensor development, manufacture and supply to the OEM market. Kodak acquired the imaging business of National Semiconductor, and then also engaged IBM in a partnership to build CMOS sensors.Kodak made a case for it’s CMOS sensors with a unique design, called “Pixelux.” In these sensors, Kodak introduced a fourth “transfer gate” transistor to the circuit design, making it possible to electrically isolate the actual photo-sensing diode, and enable the processor to sample each pixel more than once. It also enabled the processor to read each pixel separately, or to read two or four pixels at a time. This technology gave Kodak’s sensors better sensitivity and improved colour in low-light situations that are particularly important to consumers.By 2005, the joint venture was starting to bear fruit. The first new KAC-3100 and KAC-5000 CMOS sensors were being manufactured at IBM’s Burlington, VT, chip plant. Those were miniature 3 and 5-megapixel sensors, focussed on the consumer video and smart device market, but Kodak was quoting US$15 on quantities of 1,000 chips.The sensor is only half the story with CMOS technology however. A CMOS sensor requires an image processor, and here again, Kodak partnered with Texas Instruments to deliver their OMAP 2 image processors. “Kodak has a strong reputation for delivering high quality digital imaging devices, and we are excited that a reference design is now available that combines Kodak's new CMOS image sensors with TI's industry-leading OMAP processors," said Avner Goren, marketing director of TI's Cellular Systems Business. “With this powerful combination, manufacturers will be able to bring new features and capabilities to cell phones as we see the imaging, communications and the electronic markets converge.”The strategy appeared to be paying dividends when in 2006 Kodak reached an agreement with Motorola to supply the imaging technology for Motorola’s next generation of mobile pones. Taiwan semiconductor also licensed Kodak’s CMOS sensor design.In spite of Kodak’s efforts to have the Pixelux CMOS sensors adopted by other manufacturers, the first Pixelux-based product reached the consumer in July 2007, when Kodak rolled out the first-ever sub-US$100 CMOS digital camera. The new EasyShare 513 featured 5-megapixel resolution, a 3x optical zoom lens, and a 2.4-inch LCD screen. Novel for the time, the EasyShare 513 also offered a digital image stabilisation feature, intended to reduce image blur caused by camera movement. Another novel feature for the time was audio-video capture.With the EasyShare 513, Kodak had really taken digital technology full-circle, from a nascent lab experiment, through specialised products, to mass-market consumer solution, in just 30 years. Kodak also focussed on it’s mass-market ‘EasyShare’ digital cameras, and even manufactured a range of digital cameras for Canon.Kodak's CCD sensor business was also still growing, but only slowly. Kodak persevered with development and manufacture of CCD sensors. Aside from numerous industrial and process applications, Olympus’ digital cameras featured Kodak’s CCD sensors. The legendary Leica M8 and M9 cameras both carried Kodak CCD sensors. The Leaf and PhaseOne medium-format digital cameras both featured Kodak CCD sensors, as did the Pentax 645D. These were all somewhat exclusive products however, and the CCD sensor business was fast becoming marginalised by the relative speed and lower cost of CMOS devices.In November 2005, Nikon debuted their D200 model, built around a 10.2-megapixel Sony sensor, which would prove to be Nikon’s last CCD camera.The impact of digital photography was coming sharply into focus against a background of turbulent times in the photographic industry. Polaroid was declared insolvent in 2001, and then again in 2008. Konica and Minolta merged in 2003. Agfa sold it’s photography business in a management buy-out in 2004, which was then forced to file for bankruptcy in 2005. Kyocera announced that it would stop making cameras that year, including their Konica and Contax brands, hoping to salvage something from their photography businesses when Konica-Minolta announced a deal to jointly develop digital cameras with Sony.Kodak also suffered it’s first-ever losses in 2005. It was the beginning of a tide of red ink. As fast as the company retrenched it’s International operations, and exited some of it’s less strategic businesses such as inkjet printers, revenues continue to decline.The intellectual capital of the digital era was not chemistry, but silicon manufacturing techniques. After working for years to develop their own CCD sensors, Sony had developed a huge business supplying CMOS sensors to other manufacturers, along with a nice line in SLR, mirrorless still and video cameras. Samsung and Panasonic also had their own sensors and drove hard for volume to recoup their costs. Kodak still had great brand identity, and was working hard to complement their products with services, on-line and at retail, but digital photography was generating nothing like revenue or margins associated with the heyday of the film industry.Like most businesses invested in research and technology, writing patent applications had been a full-time business at Kodak. The current stock ran to more than 11,000 patents, of which about 1,000 patents related to digital photography. A lot of those patents were preliminary patents, filed in the early stages of sensor and camera development, and they were very broad. Contemporary estimates assumed that 85% of digital cameras were using some of Kodak’s patents. Companies such as Olympus, LG, Sharp, Matsushita Electric (Panasonic), Motorola, Samsung, Sanyo, Sony and Sony Ericsson all licensed Kodak’s patents.Pentax had used Kodak patents and technology, prior to being taken over by Ricoh, and was one of a number of companies that Kodak began suing to assert it’s rights. Ricoh ended up shelling out US$75.8 million to settle Kodak’s claims, the victory helping to seal the validity of Kodak’s patent rights. In January 2012 Apple filed suit against Fuji alleging infringement of five of it’s patents after the Japanese had failed to reach agreement, and at the same time filed suit against smartphone maker HTC. At about the same time, Apple reluctantly entered into a license agreement for five Kodak patents covering aspects of Apple’s iPad 2, iPhone 3G, iPhone 3GS, iPhone 4, iPhone 4S, and the iPod touch (4th generation) products.It had started to look like the portfolio would earn more through licensing that it would realise through a sale. Between 2003 and 2011 Kodak had earned Kodak US$3 billion in patent licensing deals and settlements, but by the end of 2012 the fate of Kodak and it’s patent portfolio was in the hands of it’s bankers.In 1997 Kodak’s share price had hit an all-time high of US$94/share, but by 2004 the company was removed from the Dow Jones Index when it’s shared had plummeted to US$25.38. Restructuring and divestments meant the company was shedding staff, expenses and revenue. The payroll that had once numbered more than 145,300 employees, was now down around 13,100 people, but the payroll wasn’t shrinking as fast as revenue. Kodak was profitable only once between 2004 and 2012. By January 2012 Kodak stock had become a “penny dreadful,” selling for just US$0.36¢/share. Time had finally run out.Kodak finally filed for Chapter 11 protection on the 18th of January, 2012.By any measure, this was a significant insolvency, involving about US$6.75 billion in debt, and by Kodak’s estimates, about US$5.1 billion in assets. There was a relatively small unfunded liability (US$700million) in the company’s American 401K pension plan, but the biggest creditor was the UK pension scheme, which was unfunded to the tune of about US$2.8 billion.Under the terms of it’s insolvency, Citigroup advanced Kodak a loan of US$950 million to enable the company to continue to operate, while it restructured it’s business.Among the conditions of that loan was that Kodak arrange the sale of it’s patent portfolio. An independent assessment, based on discounted cashflow analysis of licensing revenues, valued Kodak’s patent portfolio at about US$2.6 billion. As the business continued to tank, the patent portfolio started to look like one of the company’s few remaining assets.The fatal blow came just days before the patent auction was due to commence.The US International Trade Commission did an about-face and ruled against Kodak in a patent infringement case that the company had brought against Apple and RIM. The patent, known as “218” covered a method of capturing still images while previewing motion. The ruling was right out of left-field. The same patent had already been re-examined more than once, and had also been licensed by at least LG and Samsung (reportedly for almost US$1 billion). While licensing deals are typically irrevocable, the new USITC ruling undermined the value of the balance of Kodak’s portfolio, and there was simply no time, or money, to sort it out.Potential bidders had organised themselves into two camps: Microsoft, Adobe, Apple and Facebook squared off against a consortium that included Amazon, Google, HTC, Samsung, and the photo-printing website Shutterfly. The problem was that while Kodak was leading the negotiations, the company no longer had any equity. Both parties knew the the company had lost it’s autonomy to it’s lenders, who simply needed to cover their loan exposure.Naturally, neither consortium wanted to pay a cent more than necessary to secure the patents they needed, and neither did any of the parties need Kodak’s entire treasure trove of digital patents. Kodak put the portfolio up for auction on the 8th of August 2012, and had expected to announce a winner on August 20th, but the low-ball bids came as a surprise. The two consortiums are believed to have opened their bidding at around the US$150 - US$250 million mark. Kodak delayed the announcement until the 30th of August, then the 7th of September, and then again until the 19th of September, but this was clearly a buyers’ market, and Kodak no longer had any leverage. The court-sanctioned delays only made the company’s creditors more anxious.Checkmate finally came when the two consortiums joined forces, and added Fuji, Huawei and RIM to their ranks in the process. Everyone who might have been interested in Kodak’s patent portfolio was now effectively represented by a single bid, and inevitably, a deal was finally struck. In December 2012 Kodak announced that an expanded patent portfolio — not just the 1,100 digital photography patents — had been sold to the consortium for US$527 million. The consortium had negotiated to pay just enough to force Kodak to complete the sale, but the low-ball bid meant that Kodak would need to dig even deeper and realise additional assets.The Personal Imaging and Document Imaging businesses were sold to the British pension plan (knows as KPP) which settled the US$2.8 billion claim against Kodak and produced another US$650 million in cash and non-cash consideration. That business became Kodak Alaris.Kodak’s on-line services were sold to Shutterfly for US$23.8 million. IMAX acquired Kodak’s patents for laser projection. Kodak had to unravel a US$74 million sponsorship deal for naming rights at the Academy awards venue — now known as the Dolby Theatre. The gelatine business was sold off. The Japanese Brother company bought some of Kodak’s imaging assets.The disappointing result from the patent auction obviously contributed to a shortfall however, and unsecured creditors shared only 4-5% of the US$3.25 billion they were owed.The reorganisation plan was filed with the Bankruptcy Court in New York on 30 April 2013, and was formally approved on the 20th of August, 2013. In a press release, chief executive Antonio Perez said, “We have emerged as a technology company serving imaging for business markets - including packaging, functional printing, graphic communications and professional services. We have been revitalised by our transformation and restructured to become a formidable competitor - leaner, with a strong capital structure, a healthy balance sheet, and the industry's best technology.”It’s tempting to look for simple answers to even the most complex situations, but the evidence doesn’t support some of the most common perceptions. While it’s certainly true that Kodak management may have struggled to understand the impact of digital, the fact remains that Kodak, almost in spite of itself, demonstrably led the development of digital imaging, all the way through not just one, but two generations of sensor technology.The other factor that can’t be denied is that Kodak’s technology represented the pinnacle of CCD sensor technology. Kodak’s sensors were buried deep within the most prestigious products, such as cameras from Leica, Olympus, Pentax and PhaseOne, and the Kodak-designed sensors in the company’s own Pro SLR range were considered industry leading.Some of the markets that Kodak had once practically owned, and which made Kodak one of the World’s most recognised brands, simply don’t exist today. Traditional silver halide photography, although it still exists and is even seeing a small resurgence, is a tiny proportion of today’s photography market.Unprecedented collusion between Kodak’s competitors robbed the company of the better part of the value of it’s research and development. With practically all of Kodak’s potential licensees in a single consortium, determined to take advantage of the company’s insolvency, there was no chance for the beleaguered company to redeem almost any of the value of it’s research.Kodak retained the same rights to it’s intellectual property as the successful bidders in the insolvency auction, and therefore still has one of the richest caches of intellectual capital of any company. The new management is working hard to capitalise on innovation that Kodak never originally delivered to market, and interestingly is still a US$2 billion business, focussed primarily on printing and packaging, and some legacy photographic opportunities, such as movie film stock, which will continue to be in demand for some time yet.At Rochester, N.Y., Kodak had created their own “River Rouge” — a vertically integrated plant that turned wood, cellulose, gelatin, silver and other chemicals, into photographic film, paper and chemistry. By late 2013, most of that complex was surplus to even Kodak’s requirements, and some of the oldest buildings had to be demolished. Today, the area is known as Eastman Business Park, and still comprises about 1.5 million sq.metres (16 million sq.ft) of manufacturing, distribution and office space, spread over a 485 hectare (1,200 acre) campus area, housing more than 60 companies, employing more than 6,600 staff — along with what remains of Kodak.More recently still, Kodak has pioneered the use of blockchain technology to secure digital photographs, essentially doing for photographers and the photographic industry what Apple’s Digital Rights Management technology did for the recording industry. The Kodak systems will allow photographers to register work that they can license and then receive payment. Securing the property rights of photographers and graphic artists everywhere is a curious twist given Kodak’s history, but the company may yet feature prominently in the future of commercial imaging.

AI in Law: Current Applications – Insights Up FrontBased on our assessment of the companies and offerings in the legal field, current applications of AI appear to fall in six major categories:Due diligence – Litigators perform due diligence with the help of AI tools to uncover background information. We’ve decided to include contract review, legal research and electronic discovery in this section.Prediction technology – An AI software generates results that forecast litigation outcome.Legal analytics – Lawyers can use data points from past case law, win/loss rates and a judge’s history to be used for trends and patterns.Document automation – Law firms use software templates to create filled out documents based on data input.Intellectual property – AI tools guide lawyers in analyzing large IP portfolios and drawing insights from the content.Electronic billing – Lawyers’ billable hours are computed automatically.Next, we’ll explore the major areas of current AI applications in law, individually and in-depth:(We’ve done our best to place companies into the category that best represents their product offering, but it’s important to note that there is overlap on many of the groupings we’ve chosen.)Due DiligenceOne of the primary tasks that lawyers perform on behalf of their clients the confirmation of facts and figures, and thoroughly assessing a legal situation. This due diligence process is required for intelligently advising clients on what their options are, and what actions they should take.While extensive due diligence can positively impact long-term shareholder returns (according to a study by the City University of London), the process can also be very time-consuming and tedious. Lawyers need to conduct a comprehensive investigation of meaningful results. As such, lawyers are also prone to mistakes and inaccuracy when doing spot checks.Kira SystemsNoah Waisberg, a former M&A lawyer who founded the software company Kira Systems, thinks that due diligence errors by junior lawyers often occur for a number of reasons. These include working very late at night or on the eve of a weekend, forgetting to perform due diligence before the end of the workweek, and failing to act on it when a deal structure is completely revised.He adds, “Many associates are in a certain negative mood about the efficacy of manual due diligence. Lawyers, being human, get tired and cranky, with unfortunate implications for voluminous due diligence in M&A.”Kira Systems asserts that its software is capable of performing a more accurate due diligence contract review by searching, highlighting, and extracting relevant content for analysis. Other team members who need to perform multiple reviews of the content can search for the extracted information with links to the original source using the software. The company claims that its system can complete the task up to 40 per cent faster when using it for the first time, and up to 90 per cent for those with more experience.LEVERTONLEVERTON, an offshoot of the German Institute for Artificial Intelligence, also uses AI to extract relevant data, manage documents and compile leases in real estate transactions. The cloud-based tool is said to be capable of reading contracts at high speeds in 20 languages.In 2015, IT firm Atos sought the help of real estate firm Colliers International, which used LEVERTON in performing due diligence of a company that the former was about to acquire. Through the use of LEVERSON’s AI, information such as payable rent, maintenance costs and expiration dates were extracted from thousands of documents and then organized on a spreadsheet.eBreviaHowever, lawyers can be burdened by reviewing multiple contracts and they may miss important edits that result in legal issues later on. This is the same problem that Ned Gannon and Adam Nguyen, co-founders of eBrevia, experienced when they were still working as junior associates. They built a startup in partnership with Columbia University with the intention of shortening the document review process.eBrevia claims to use natural language processing and machine learning to extract relevant textual data from legal contracts and other documents to guide lawyers in analysis, due diligence and lease abstraction. A lawyer would have to customize the type of information that needs to be extracted from scanned documents, and the software will then convert it to searchable text. The software will summarize the extracted documents into a report that can be shared and downloaded in different formats.JPMorganOther organizations such as JPMorgan in June 2016 have tapped AI by developing in-house legal technology tools. JP Morgan claims that their program, named COIN (short for Contract Intelligence), extracts 150 attributes from 12,000 commercial credit agreements and contracts in only a few seconds.This is equivalent to 36,000 hours of legal work by its lawyers and loan officers according to the company. COIN was developed after the bank noticed an annual average of 12,000 new wholesale contracts with blatant errors.ThoughtRiverOther AI industry players includeThoughtRiver, which handles contracts, portfolio reviews and investigations for improved risk management. Its Fathom Contextual Interpretation Engine was developed together with machine learning experts authorities at Cambridge University.The company states that it designed the product to automate summaries of high-volume contract reviews. While users read content extracts, they can also read the meanings of clauses provided by the AI. The system is also said to be capable of flagging risky contracts automatically. The company provides a brief tour of their product in the 3-minute video below, including a detailed look at the user interface and basic functions of the software:LawGeexLawGeex claims that its software validates contracts if they are within predefined policies. If they fail to meet the standards, then the AI provides suggestions for editing and approval. It does this by combining machine learning, text analytics, statistical benchmarks and legal knowledge by lawyers according to the company.In this video, LawGeex CEO Noory Bechor further explains how his product can cater to legal services.The company also claims that with their tool, law firms can cut costs by 90 per cent and reduce contract review and approval time by 80 per cent (though these numbers don’t seem to be coupled with any case studies). The firm lists Deloitte and Sears among some of its current customers.Legal RobotOn the other hand, Legal Robot, a San Francisco-based AI company, currently offers Contract Analytics, its answer to the growing contract review software market. Currently, in beta, the company states that its software is capable of changing legal content into numeric form and raising issues on the document through machine learning and AI.A video presenting how the software works state that it builds a legal language model from thousands of documents. This knowledge is used to score the contract based on language complexity, legal phrasing, and enforceability. With the issues flagged by the software, it then provides suggestions on improving the contract’s compliance, consistency, and readability by evaluating it on best practices, risk factors and differences in the jurisdiction.Ross IntelligenceEvery lawsuit and court case requires diligent legal research. However, the number of links to open, cases to read and information to note, can overwhelm lawyers who have limited time doing research. Lawyers can take advantage of the natural language search capability of the ROSS Intelligence software by asking questions, and receiving information such as recommended readings, related case law and secondary resources.BakerHostetler, in what seemed like a break in tradition, employed ROSS in its bankruptcy department, 100 years after the law firm’s founding. The law firm’s chairman, Steven Kestner, explained in an interview that they decided to employ the software to work on 27 terabytes of data. A Forbes report describes ROSS’ function in the law firm’s operations: “ROSS will be able to quickly respond to questions after searching through billions of documents.The company claims that lawyers can ask ROSS questions in plain English such as “what is the Freedom of Information Act?” and the software will respond with references and citations. Like most machine learning systems, ROSS purportedly improves with use.CasetextOn the other hand, Casetext’sCARA claims to allow lawyers to forecast an opposing counsel’s arguments by finding opinions that were previously used by lawyers. Users can also detect cases that have been negatively treated and flagged as something that lawyers may deem unreliable.Casetext claims large law firms such as DLA Piper and Ogletree Deakins as its clients.Other Assorted ApplicationsOther software products also combine machine learning and legal analysis to assist lawyers with legal research but with limited coverage. For example:Loom features win/loss rates and the judge ruling information but only for civil cases in select Canadian provinces. In an interview with Stanford Law School, Mona Datt (co-founder of Loom Analytics) elaborates:“Instead of performing open text searches looking for personal injury precedents, a lawyer could use Loom’s system to see all personal injury decisions that were published in a given time span and then break them down by the outcome.Instead of combing through individual decisions looking for ones written by a particular judge, Loom’s system can show all decisions authored by that particular judge and provide an at-a-glance snapshot of their ruling history. In short, we’re providing quantitative metrics on Canadian case law.”Judicata, on the other hand, only serves California state law as of this writing. Its software, Clerk, is said to be capable of reading and analyzing legal briefs. It also evaluates their pros and cons and then assigns a score for each brief based on arguments, drafting and context.The recommendations intended to reduce content mistakes are listed as part of the action items for the user. In the company’s blog, Product Manager Beth Hoover explains, “Clerk helps reduce these errors by identifying the quotations in a brief and cross-checking them against the cited case to ensure the text is identical and the page numbers are correct.”Potential Bias ConcernsIn a recent paper by Susan Nevelow Mart of the University of Colorado Law School tested if online legal case databases would return the same relevant search results. She found out that engineers who design these search algorithms for case databases such as Casetext, Fastcase, Google Scholar, Lexis Advance, Ravel, and Westlaw have biases on what would be a relevant case that their respective algorithms will show to the user.For example, newer databases such as Fastcase and Google Scholar have generated less relevant search results compared to older databases such as Westlaw and Lexis. Mart argues that search algorithms should be able to generate redundant results on whatever legal online database is used since lawyers need only the most relevant cases. However, because these engineers have biases and assumptions when developing their algorithms, users are recommended to use multiple databases in order to find out the cases that fit their needs.By no means does a single paper (or a dozen papers) imply that legal AI tools shouldn’t be used, but rather that their pros and cons are yet to be weighed out thoroughly. Bias is by no means specific to the legal field, as machine learning systems are always influenced by the data that they’re trained on.Readers with an interest in AI bias and the ethical considerations of discrimination by machines may benefit from reading our recent article in collaboration with the IEEE: Should Business Leaders Care About AI Ethics?EverlawThere has been a growth in the number of e-discovery product manufacturers that harness AI and machine learning. Everlaw uses its predictive coding feature to create prediction models based on at least 300 documents that were classified before as relevant or irrelevant by the user.The AI looks into the contents and metadata and uses such information to classify other documents. The company claims that the prediction model’s results can help users easily identify which documents are most relevant. It also recommends actions on the part of the user on how to improve the software’s predictive accuracy of the model.You can check a demonstration of Everlaw’s Prediction Technology feature in this video:DISCODISCO claims to deliver faster results using its cloud technology for document search on large data volumes. Similar to Everlaw, it also employs prediction technology to suggest which documents are most likely to be relevant or irrelevant to the user.The AI works by assigning scores on tags (on a scale of -100 to +100) in order to improve its prediction results. The software displays its search results with each document’s score and suggests which material is most likely useful for the reader. In the promotional video below, Dr Alan Lockett (DISCO’s Head of Data Science) explains the company’s technology in simple terms:CatalystDenver-based Catalyst markets its Automated Redaction product to help lawyers and legal reviewers remove sensitive and confidential information on documents. “Manual redaction”, as the company claims, is cumbersome considering the amount of time that a reviewer spends on locating content on a digital document and then applying black boxes on these statements.Their tool allows users to convert a document to digital format and then perform multiple sets of redactions for a single document by searching for a word or phrase. Users can also set patterns such as social security numbers on the software to be redacted. An overview of the Automated Redaction feature for e-Discovery introduces its uses.ExterroExterro’s WhatSun claims to combine the functions of a project management software with the capabilities of performing e-Discovery. In other words, users can perform their legal research and then collaborate with others using the software.According to one of Exterro’s law firm clients, they were able to cut down on redundant workers from 100 lawyers down to 5 when they started to employ the system. The software was able to perform the e-Discovery tasks of the lawyers at a 95 per cent cost savings according to the law firm. The company claims AOL, Microsoft, and Target among its marquee customers.Brainspace DiscoveryBrainspace Discovery clusters and sorts documents to match closely to a user’s document search. When finding documents, the AI employs concept search (searching for documents that are similar in concept but not necessarily in words or phrases), term or phrase extension (instructing the software to remove terms incorrectly associated with the results), and classification (specifying another category to refine the search). The topic of document digitization and search is explored further in our article on Document Digitization in Finance. The company claims that by combining these three features, the software can better deliver document search results closer to a user’s needs.Other AI-powered contract review platforms that cater to due diligence for legal professionals include:iManage’s RAVN whose M&A Due Diligence Robot is designed for M&A documents to automate the review process and extract data from cluster sets;LitIQ, which capitalizes on computational linguistics technology to reduce contract-related disputes (Gary Sangha, founder of LitIQ shares his thoughts on the relationship of machine learning and law in this interview);LegalSifter, which claims to cut time and financial costs through its AI software that looks for specific concepts in documents such as general terms and conditions and confidentiality agreements;Seal, whose software is used by Dropbox, PayPal and Experian, was able to reduce the time spent to 48 hours from 255 days by a utility company by searching for thousands of contracts with specific clauses according to their case study; andLuminance, which claims to be the only tool that searches and ranks unusual and anomalous documents and clauses for lawyers.While there has been a growth in the use of e-Discovery tools, its application has become a public issue in states such as California. In 2015, the State Bar released an amended Proposed Formal Opinion, requiring lawyers to have a decent knowledge on the to use the e-Discovery system or they will warrant discipline after being proven to have committed intentional or reckless acts. The State Bar also suggests that if a lawyer is incompetent on the facility, he should learn the skill, hire someone who’s knowledgeable, or just simply decline representation.Prediction TechnologyIn 2004, a group of professors from Washington University tested their algorithm’s accuracy in forecasting Supreme Court decisions on all 628 argued cases in 2002. They compared their algorithm’s results against a team of experts’ findings. The statistical model by the researchers proved to be a better predictor by correctly forecasting 75 per cent of the outcomes compared to the expert’s 59 per cent accuracy. Although in a separate industry with its own separate problems numerous additional use-cases of predictive analytics can be found in our article on Predictive Analytics in Banking. This article illuminates AI programs utilizing predictive analytics to solve real-world issues.Expanding the coverage from 1816 to 2015, Prof. Daniel Katz of Michigan State University and his two colleagues achieved a 70.2 per cent accuracy on case outcomes of the Supreme Court in their 2017 study. Similarly, Nikolaos Aletras of University College London and his team used machine learning to analyze case text of the European Court of Human Rights and reported a 79 per cent accuracy on their outcome prediction.Prof. Daniel Kantz, in his 2012 paper, stated, “Quantitative legal prediction already plays a significant role in certain practice areas and this role is likely to increase as greater access to appropriate legal data becomes available.”IntraspexionIndeed, several AI companies have ventured into this field such as Intraspexion, which has patented software systems that claim to present early warning signs to lawyers when the AI tool detects threats of litigation.The system works by searching for high-risk documents and displays them according to the level of risk that the AI has determined. When a user clicks on a document, risk terms as identified by subject matter experts through the algorithm are highlighted. According to the company, users can which documents put them at risk for litigation when they use the software.Ravel LawAnother tool, Ravel Law, is said to be able to identify outcomes based on relevant case law, judge rulings and referenced language from more than 400 courts. The product’s Judge Dashboard feature contains cases, citations, circuits and decisions of a specific judge that is said to aid lawyers in understanding how the judge is likely to rule on a case.The firm’s CEO, Daniel Lewis, affirms such claim in this interview when he explained that the Ravel Law can aid in litigation strategy by providing information on how judges make decisions.Lex MachinaLex Machina’s Legal Analytics Platform has a variety of features that are said to assist lawyers in their legal strategy. For example, the Timing Analytics feature uses AI to predict an estimated time when a case goes to trial before a specific judge.The Party Group Editor, on the other hand, allows users to select lawyers and analyze their experience before a judge or a court and the number of lawsuits they were involved in before, among others. In the video below, the product’s user interface is featured and sample analytics results are presented:PremonitionFinally, Premonition, which claims to be the world’s largest litigation database, asserts to have invented the concept of predicting a lawyer’s success by analyzing his win rate, case duration and type, and his pairing with a judge at an accuracy of 30.7 per cent average case outcome. According to the company, the product can also aid in looking at different cases and how long they’re going to take for each attorney.But similar to any analytics platform, AI tools that deal with predictive technology need a lot of data in the form of case documents to fully work according to Kantz. In an article, the model is described as “exceptionally complicated.” That’s because it needs almost 95 variables (with almost precise values up to four decimal places) supported by almost 4,000 randomized decision trees to predict a judge’s vote. Kratz admits that a database that will fully support his product is still not readily available except for a few ones that charge access fees to obtain data.Legal AnalyticsCase documents and docket entries provide supplementary insights during litigation by lawyers. Current AI tools claim that today’s software products are able to extract key data points from these documents to support arguments.Lex MachinaHogan Lovells litigation attorney Dr Chris Mammen uses Lex Machina’s Legal Analytics software to find out “who is the plaintiff, who is their counsel, who have they represented, and who else have they sued.”The software generates data that can be used to analyze an opposing counsel’s likelihood of winning or losing a case. He claims to save time through the analytics results when creating a narrative. “You send an email to the research department, get first results, iterate the process – that usually takes a day or more.”Intellectual property lawyer Huong Nguyen also used the software when she represented a generic pharma company. Using the Legal Analytics tool,, she found out that the judge’s history of ruling cases tends to favour cases like hers. Both parties settled in the end, which was a better resolution according to Nguyen.Ravel LawThe data can also be used in pitching a law firm’s services to potential clients by providing intelligence on the opposing counsel, generating values on the probability of winning the case and identifying litigation trends to use in their marketing campaigns.Apart from prediction technology, Ravel Law’s software also claims to provide lawyers with judges’ data on cases, circuits and ruling on their dashboard, which can be used in landing new clients. Currently, the company is bolstering its data minefield by working with Harvard Law School in digitizing the faculty’s US case law library to be made available on its tech platform.Settlement AnalyticsIt must be noted that legal analytics still has its limitations. Robert Parnell, CEO of SettlementAnalytics, explains in his paper that there is uncertainty in producing results with a high level of accuracy with this kind of technology. This comes in very small sample sizes after filtering, cognitive biases, the tendency to interpret random trends as valid patterns, and having a lot of data noise.Parnell opines, “Overall, the quantitative analysis of legal data is much more challenging and error-prone than is generally acknowledged. Although it is appealing to view data analytics as a simple tool, there is a danger of neglecting science in what is basically data science. The consequences of this can be harmful to decision making.”Document AutomationA McKinsey & Company report estimates that knowledge work automation will most likely be one of the top disruptors in the global economy.A screenshot from McKinsey’s report cited above. See the orange bar to the right of “2. Automation of knowledge work”Some law firms are also beginning to adopt such technology by drafting documents through automated software. Many such software companies claim that the final document, which could take days by manual human drafting, is generated in a matter of minutes. This technology is utilized in finance today; discover real-world banking-sector examples of the concepts outlined in this article.PerfectNDANeota Logic System claims that it's software PerfectNDA shortens the nondisclosure agreement (NDA) process by offering templates selected by AI according to a user’s scenario. The user answers questions and a pre-filled template is then generated. In addition, the software also features document filing and integrated e-signatures to streamline related manual processes involved in NDA drafting.Intellectual PropertySecuring patents, copyrights and trademarks is often best left to a lawyer’s expertise. However, the entire patent application process can be long and arduous. Traditional trademark and patent search, for example, involves looking into hundreds, if not thousands, of results through manual research. This takes so much time, which is ironic considering that patent applications are time-sensitive.According to US patent attorney Patrick Richards, “You only have one year from the first time the invention is publicly disclosed [i.e. sold] to file a patent application; if as a business owner you launched a product within the last year, you need to talk to a patent attorney right away to make sure it is protected.”TrademarkNowTrademarkNow is a company taking on some of the manual knowledge work of intellectual property application with AI. It uses a complex algorithm that is said to shorten weeklong searches for patents, registered products and trademark using the Trademark Clearance platform, which returns search results in less than 15 seconds according to the company’s claims.The system analyzes the results and ranks them according to relevance to the user as identified by the algorithm. As this many e-discovery applications, the solution promises efficiency of spent attention for legal teams.ANAQUA StudioThe cloud-based ANAQUA Studio, on the other hand, is specifically designed for drafting patents and prosecution. The company’s datasheet states that it’s the first patent application-drafting tool for lawyers that save four hours on provisional patent application and 20 hours and non-provisional types. The system is said to be able to detect document errors, circular claim references and formatting defects aside from automatically generating literal claims support.SmartShellSeattle-based TurboPatent released SmartShell to support paralegals performing document reviews, drafting, formatting and identifying issues on patent applications. The software uses AI and natural language processing to assist in creating legal claims.In a case study listed on TurboPatent’s website, two paralegals from the Pacific Patent Group used the software to perform document retrieval, bibliographic data research, examiner remarks review and rejection issues discovery. TurboPatent claims that Pacific’s paralegals were 500-800% more productive in their tasks when using SmartShell (thought the case study isn’t clear what exact tasks were relevant for the software, and which weren’t – we can assume that many paralegal tasks aren’t currently improvable with AI).A brief overview of how the product’s functions and value proposition can be seen in the 1-minute video below:Electronic BillingElectronic Billing platforms provide an alternative to paper-based invoicing with the goal of reducing disputes online items, more accurate client adjustments, (potentially) more accurate reporting and tracking, and reduced paper costs. Firms in the healthcare space are also utilizing AI for medical billing; this concept is further explained in our article Artificial Intelligence for Medical Billing and Coding.BrightflagBrightflag offers centralized legal pricing software that automatically adjusts line-by-line items. It also allows users to centralize the invoice review so that all documents submitted are routed directly to the correct approver. In addition, the AI provides analytics features by tracking and categorizing all pricing data to determine alternative fee arrangements (AFA) and budgets.The company claims that its average client reduces administrative costs related to payment management by 8 to 12 per cent by using the platform’s assisted review feature. The company lists telecom giant Telstra and ride-hailing company Uber among its current marquee clients.SmokeballSmokeball’s cloud-based legal practice management tool automates the recording of time and activities by law firms. One major feature of this tool is the capability to track all activities including emails that are valid for billing. It claims to have automated more than 600,000 forms and managed over 10 million documents according to its website. The video below explains Smokeball’s software: