Hacker News new | past | comments | ask | show | jobs | submit login
Russell Kirsch, inventor of the pixel, has died (petapixel.com)
130 points by Yetanfou on Aug 15, 2020 | hide | past | favorite | 17 comments

Kirsch co-authored - with an undergraduate intern! - one of the foundational papers of my field of cheminformatics. Then never touched the topic again, as he was more interested in computers and art.

He worked on the SEAC computer at the National Bureau of Standards. Quoting the Wikipedia page (which in turn cites Kirsch), SEAC "is claimed to be the first fully operational stored-program electronic computer in the US".

The NBS had this great little machine - the first computer available to the US government, at least for general use - and it got applied to all sorts of projects.

Marvin Minsky, for example, learned to program on it, https://www.webofstories.com/play/marvin.minsky/29 , where they tried to get the computer to recognize letters. (Minsky commented "my friend Russell Kirsch ... was the kind of person who could say well, everybody's having their machines crunching numbers, can we have one that can recognize patterns.")

The SEAC team talked with the US Patent office, which had a thorny problem of how to handle chemical patents. They needed some way to search chemical structures by substructure. Ray (the intern) and Kirsch implemented subgraph isomorphism search on the SEAC, in 1956, published the following year, and still often cited (though less often read).

Kirsch humorously points out in the video at http://www.computerhistory.org/revolution/birth-of-the-compu... ,

> 21:09: This solved the immediate problem for the Patent Office, although in a very ponderous way. An amusing incident is this: The program that we wrote demonstrated this possibility, but it was so ponderous - so slow - that we decided 'gee, I hope we don't have to use our program in the Patent Office'.

> 21:29: Well in those days, a tool that you hoped you wouldn't have to use was the hydrogen bomb. So we decided to call our program the H-BOMB code. When the coding sheets got misplaced, the FBI came in, and it took an awful lot of explanation to explain this was just a name; it had nothing to do with the hydrogen bomb.

As another example of the humor they had, see http://www.ams.org/journals/notices/200711/tx071101502p.pdf where:

> The staff of the NBS computing lab declared us “Heroes of the SEAC”, a title awarded in those days to programmers whose programs ran on the first try—a rare event—and for some while we had to go around wearing our “medals,” which were drawn freehand in crayon on the back of used teletype paper. (The word “hero” was in parody of the practice in the Soviet Union of declaring persons “Heroes of the Soviet Union” for this and that accomplishment.)

For more about the machine, see https://www.nist.gov/news-events/news/2010/06/pioneer-nist-c... .

I wrote a letter to him a few years ago, to ask questions about his chemical substructure work. His wife, Joan Kirsch, kindly replied and said his dementia was too far progressed. My heart goes out to her and her family.

This got me curious when I saw the news item on national Swedish news (!) a few hours ago, so I did some wikipedia reading.


> The word pixel is a portmanteau of pix (from "pictures", shortened to "pics") and el (for "element"); similar formations with 'el' include the words voxel and texel. The word pix appeared in Variety magazine headlines in 1932, as an abbreviation for the word pictures, in reference to movies. By 1938, "pix" was being used in reference to still pictures by photojournalists.

> The word "pixel" was first published in 1965 by Frederic C. Billingsley of JPL, to describe the picture elements of scanned images from space probes to the Moon and Mars. Billingsley had learned the word from Keith E. McFarland, at the Link Division of General Precision in Palo Alto, who in turn said he did not know where it originated. McFarland said simply it was "in use at the time" (circa 1963).


> The first digital image was produced in 1920, by the Bartlane cable picture transmission system. British inventors, Harry G. Bartholomew and Maynard D. McFarlane, developed this method. The process consisted of “a series of negatives on zinc plates that were exposed for varying lengths of time, thus producing varying densities,”. The Bartlane cable picture transmission system generated at both its transmitter and its receiver end a punched data card or tape that was recreated as an image.

> In 1957, Russell A. Kirsch produced a device that generated digital data that could be stored in a computer; this used a drum scanner and photomultiplier tube.

So Kirsch was a pioneer in creating a device that could store images in a computer. That is something he should be remembered for.

This "inventor of the pixel" thing is just weird, though. Pixels can clearly exist outside of computers, and did so before 1957. Not to mention https://en.wikipedia.org/wiki/Mosaic#History .

"Inventor of the pixel" is a feel good story title.

I remember reading a fun story about Russell (a fellow Portland resident) a few years back. Worth a quick read.


Rest in peace, Mr. Kirsch.

One of the amazing things about Russell Kirsch is that he got excited about the possibility of digital imagery when he was 21. Kirsch created the very first digital image in his late twenties and it was a black and white image of his three month old son. He continued to work on digital imaging for a total of 70 years!

He and I frequented the same bakery several years ago. The poor fellow’s brain wasn’t coping with age well, but he was always cheerey, and one time made a pretty fun comment about how my pocket computer’s (phone’s) camera had him to thank. Sad to hear of his passing.

Why aren't we using variable shaped pixel?

The image quality seems so much higher than the square pixel used today.

See the short but interesting video in the linked article.

The image quality is higher because it has a lot more information in it; it's not accurate to compare the pixel count in a rectilinear array to the count of custom-cut tiles in a mosaic.

We use rectilinear images because that's the most convenient form for addressing the picture elements in memory. Not only from the point of view of getting to a particular pixel in a frame buffer as in y * scanline_width + x. Image processing algorithms of all kinds depend on the data being a rectilinear grid of samples: that the sample rate is constant across the scan line, and vertically.

We can recover image quality with higher resolution and anti-aliasing techniques.

We can also use vector techniques to represent a mosaic's tiles accurately (yet compactly, for file storage and network transfer), and render them at any resolution using square pixels. If there are enough of those pixels that the human eye is well beyond giving a darn, then the issue is pretty much settled.

We can prove with information theory that if the pixels are small enough, their shape does not matter. After a certain resolution level, we no longer add detail: we only improve the smoothness of anti-aliasing, and there is only so smooth it has to be relative to the frequency response of the eye's optics and retina.

Link for the lazy: https://vimeo.com/22179638

It's worth noting that many OLED displays used in smartphones don't have square pixels but are actually made up of a "pentile subpixel matrix" (do an image search for those keywords to see what I mean). That said those non-square red/green/blue subpixels are fixed and Russel Kirsch suggestion seems to be to somehow display differently shaped pixels arbitrarily. I'm not sure the kind of display technology he thinks could achieve that. Maybe E-Ink displays used in eBook readers like Amazon Kindle might be able to achieve this when the magnetic field is applied in a certain way? I'm not sure. Also I came across this quote from his son:

> Walden Kirsch said he regrets his father was unable to appreciate how prevalent that technology has become.

For somebody who had Alzheimer’s disease, he seems very lucid in that 2011 video. The world hasn't changed that much in the last 9 years so I think it's pretty clear that he understood the impact his work had on the world.

Makes me think of an old tee-shirt I had: https://idata.over-blog.com/1/72/23/10/images-2/pixel_mort.j... ("yet another dead pixel"). RIP Russell Kirsch.

Has anyone done any work on variable shaped pixels? I can see the currently strong counter arguments against it, but it seems like one of those weird ideas that might find a home someday.

Interesting. I am a bit disappointed though about the caption of that 172x172px image. It goes into details about superimposing two scans to achieve a 172x172 b/w image. But obviously the image shown is an interpolated and post-processed version of a photograph. It is neither b/w nor 172x172. Probably because the original data wasn't archived, only a printout (1957!). Still, a slightly misleading caption.

Ceci n'est pas une pipe.

Can we do this with all "has died" posts? You know, a brief summary what the person is known for.

90% of the time I have no idea who the person is. So sometimes I look the person up and realize there's no way I would have known who they were. Maybe that's the goal with the "${person-most-people-have-never-heard-of} has died" titles though.

Part of the HN philosophy is that it's good for readers to have to work a little. Not too often nor too much, but enough to engage the slower circuits a bit.



Interesting. It's always nice when a made up theory is validated.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact