LETTERS TO THE EDITOR

In the first section of this page, immediately below, are some "letters to the editor" that I wrote and were published.  (These have been, or may soon be, removed from those publications' websites.  They are preserved here for your reading pleasure.)   In the second section of this page are some letters that I wrote to various publications that were not published, and which I consider deserving of a readership.

A.  PUBLISHED LETTERS:

Alternate Endings for Tinseltown's Digital Drama

Business Week, August 4, 2003, referring to an article about declining CD sales (supposedly due to piracy) and the future threat to DVD sales:

The fact that CD sales fell during the period when digital audio technology became widely available does not prove causality. CD sales were kept at "unnatural" levels from 1983 to 1999 by the baby boomers, who gradually repurchased, in the new format, music already owned on vinyl. But by 1999, boomers had repurchased all the material they ever wanted, including many overlapping compilations containing many unwanted duplicates. (I count no less than eight copies of Hey Paula in my collection!)

Malcolm Hamer
New York

This was a rather-slimmed-down version of the letter that I sent them, the full text of which was:

In “Stealing Hollywood”, July 14, you stated that the music industry is “nearly paralyzed by piracy” and you imply that the burning of CDs and exchanging of MP3 files is central to the fall-off in CD sales in the last four years. However, the fact that CD sales have fallen over the same period that digital audio technology became widely available does not prove causality. Based on timing, one might also argue that digital audio technology has caused an increase in terrorism. The music industry’s assertion that a causal relationship exists is self-serving and should be questioned. If we could re-run history without digital audio technology it is quite possible and, I believe, likely that the fall-off in sales would have happened anyway, for two reasons.

First, younger CD buyers have always tended to buy as a group – sharing their purchases and making “compilation” tapes (now CDs) for one another. As a group they have a certain number of dollars that they are prepared to spend on music. This is more or less fixed, regardless of the extent of sharing within the group. Digital sharing technology has not significantly changed this situation. However, other demands on these buyers’ budgets have caused them to reduce the amount allotted to CD purchases, notably mobile phone charges and, to some extent, DVD purchases.

Second, CD sales were kept at “unnatural” levels from 1983 to 1999 by the baby boomers, who gradually re-purchased, in the new format, music already owned in vinyl format. The industry has known this all along and it has devoted great energy to releasing new combinations of old material, with just enough new-to-CD tracks and remixes to get the boomers to hit the “buy” button. But by 1999 the boomers had re-purchased all the material they ever wanted, including many overlapping compilations containing many unwanted duplicates. (I count no less than 8 copies of “Hey Paula” in my collection!) That the boomers’ CD-buying has fallen off dramatically is obvious if you compare the top-twenty charts of 2003 with, say, 1998.

While the music industry and Hollywood can, and should, take legal action against organized pirates, their obsession with trying to use technology to protect material sold to normal consumers is misplaced. It will not stop organized pirates, who have the resources to crack any technical protection, and it will take away our rights to fair use (as the industry tried, unsuccessfully, to do via the courts when the VCR appeared on the scene in the early 1980s).
 

MPEG-4 And The HD-DVD Standard

Widescreen Review, November 2002

Dear Gary,

I was pleased to see that in your October (Issue 65) editorial you made mention of MPEG-4 (in connection with the hoped-for HD-DVD standard).  It seems to me that MPEG-4 is being treated by some players in the video technology market as some sort of rogue technology, rather than the outcome of years of serious work by the industry’s respected group of experts (i.e. the MPEG).  Steve Jobs and Apple seem to be the only players openly supporting and praising MPEG-4.  Other players seem to want to brand it as “that thing that video pirates are using to exchange ripped videos over the Internet”.  And Microsoft’s message to the MPEG regarding MPEG-4 seems to be “Thank you for all your hard work, guys, but we’ll take it from here.”  It is not clear what deals Microsoft is striking with the other players, particularly the studios, but it seems possible that they are along the lines of: “We will build rock-solid copy protection into the compression algorithm and you will support the use of a Microsoft algorithm so we can make billions of dollars in licensing fees, just as we did with Windows.”  

While I agree with you that the present red laser approach, with an average bit rate of 5 Mbps, combined with the best possible compression algorithm, is not going to give us the HD quality that we want (Warner’s approach), those other manufacturers that are proposing to use Blu-ray discs in combination with MPEG-2 must be crazy.  Why would anyone opt to use MPEG-2 after the MPEG spent almost a decade developing the improvements and additions to MPEG-2 that resulted in MPEG-4?  If the industry had followed similar logic in moving from laserdiscs to DVDs, then DVDs would be using analog video!  Surely the HD-DVD standard should combine the best available hardware and software elements, namely 20 to 30 Mbps bit rate Blu-ray discs with the MPEG-4 algorithm?  In very rough terms this would give us an overall quality improvement factor versus DVDs of something like 16 to 24 (based on bit rate improvement of 4 to 6 multiplied by an estimated compression efficiency improvement factor of 4).

In Stephen S. Hall’s interview with Ewan Birney [“Journey to the Genetic Interior”], Birney deprecates the term “junk DNA” for sequences whose function we do not know, but he is conservative on how much of nonprotein-coding DNA may be functional (“between 9 and 80 percent”).  The fact that the entire genome is copied at every cell division suggests that close to 100 percent of DNA must be functional.  Had any significant portion of DNA been nonfunctional in the past, evolutionary pressure to develop an editing-out mechanism to increase the cell’s energy efficiency would have been tremendous.

Birney also uses the conservative term “regulation” to describe how the 98.8 percent of nonprotein-coding DNA interacts with the 1.2 percent of protein-coding segments.  It is more useful to describe the entire genome as software: instructions for cells to build copies of themselves and assemble cells into life-forms.  In this view, the protein-coding segments are thought of as fixed-value strings within the code.

If we could send a personal computer with Microsoft Excel and a copy of its source code back in time to Alan Turing, it seems unlikely that on comparing the screen output with the source code, Turing would conclude that fixed values such as “File,” “Edit” and “View” were the essence of the software and that the other 99 percent merely “regulated” the operation of the fixed values.

Malcolm Hamer
New York City

Note: this was the third letter that I wrote to Scientific American on this topic.  The first two (not published) appear immediately below.

_____________________________________________________________________

B. LETTERS THAT WERE NOT PRINTED BY THE PUBLICATION TO WHICH I SENT THEM:

The Unseen Genome: Gems among the Junk           PDF version of this letter
Sent to Scientific American in 2003

The publication of the article “The Unseen Genome: Gems among the Junk”, by W.Wayt Gibbs (October 2003), may turn out to be a critical step in facing up to the possibility that almost the whole of every DNA strand in a cell is potentially-executable “code”. The original “junk” hypothesis – that only the sections of code that form a template for protein construction represent “valid” code – was hastily arrived at, and led to researchers treating these sections of code as uninteresting. Now the hypothesis is being tested and found to be, at least in part, wrong. Perhaps it is entirely wrong.

A useful analog for DNA might be a computer program. Suppose that we knew absolutely nothing about a programming language, but had studied samples of code and noticed that it contained strings of characters enclosed in quotation marks, and that these strings could be matched with words in the program output, like “RESULTS” and “ ACHIEVED” in:
  1000 PRINT "RESULTS ";
  1100 PRINT X1,X2,X3;
  1200 PRINT " ACHIEVED"
  1300 GOTO 700

Looking at this code, would we say that everything other than the printable strings is “junk” that does nothing? Surely not. By analogy, we should think of protein-defining sections of DNA as being like “the stuff in quotes”, and we should expect the rest of the DNA to be just as important – possibly more important. This is a much more logical conclusion. If the “junk” were really junk, and cells were able to recognize and ignore junk when “executing” the DNA, then cells that edit out the junk while copying the DNA would have evolved and prospered, being more energy-efficient. Yet the “junk” is still there; so it is probably not junk.

Of course, not all of the code in the DNA may be executed in a particular cell at a particular point in its life. As in a computer program, we should expect the code to contain conditional statements that cause branching, such that only certain sections are executed in a given situation.

If researchers were to approach the task of understanding how DNA as a whole operates, with the assumption that the majority of DNA is “potentially executable code” that plays a vital role in cell construction (and multicellular lifeform assembly), then they would be more likely to find out how it all works.

Malcolm Hamer
New York

Junk DNA Isn't Junk After All           PDF version of this letter

Sent to Scientific American in 2004

The article “The Hidden Genetic Program”, by John S. Mattick (October 2004), is the latest in a series of articles which reveal the growing recognition that introns (also known as “junk DNA” – the large sections of DNA in between the protein-defining strings) is not, after all, “junk”.  It is becoming clear that intronic DNA, through a number of mechanisms which we are only starting to observe and understand, plays a role in assembling proteins into operating structures.

Viewing DNA as something analogous to a computer program, rather than a series of protein-defining strings separated by junk, is much more likely to lead to a full understanding of how DNA works: it implies that we accept that at least some, and possibly most, of the intronic DNA consists of instructions about how to build a cell, or at least how to split one cell into two cells.  Of course, not all of the intronic DNA code may be “executed” in a particular cell at a particular point in its life.  As in a computer program, we should expect the intronic DNA “program” to contain conditional statements that cause branching, such that only certain sections are executed in a given situation.

One point that has not yet been covered in the articles on this topic is the need to search for two distinct classes of “program” within the intronic DNA of eukaryotes.  The first class consists of programs that play a vital role in intracellular processes, particularly during cell division.  The second class consists of programs that play a role in multicellular lifeform assembly.  Overall, we should expect at least half of intronic DNA to be concerned with intracellular processes, rather than multicellular lifeform assembly.

Multicellular lifeforms, such as human beings, are clearly seen to be complex when studied with an optical microscope.  However, the information required to assemble roughly 100 trillion cells into a human being (given a means of generating cells) is no more than, and is possibly less than, the information required to assemble about 100 trillion molecules of various compounds into a cell.  (We are reminded of the remarkable complexity of cells only occasionally by molecular biologists.)

Congenital diseases, and susceptibility to other diseases, can arise from three distinct characteristics of a person’s DNA: (A) errors in protein-defining strings, (B) errors in the intracellular-process-defining intronic DNA, or (C) errors in the multicellular-lifeform-assembly intronic DNA.  Diseases related to Type (A) errors have been easy to identify.  However, in order to understand the relationships between Type (B) and Type (C) errors and the diseases that they may cause, or make an individual susceptible to, it is first necessary to identify which parts of intronic DNA are related to intracellular processes, and which parts are related to multicellular lifeform assembly.

Malcolm Hamer

New York

Eliminating Flicker With 1080p/24 and Related Matters

Sent to Widescreen Review in 2003

Dear Gary,

The article “DVI and HDMI” by Alen Koebel in Issue 69 gave an excellent review of the state of the DVI, HDMI and IEEE 1394 standards.  I was particularly interested in Alen’s comment about the advantages of 1080p/24 in delivering movie frames without 3:2 pulldown – a point which I believe you have touched on in earlier issues.

Alen went on to say that direct display on a CRT of a 1080p/24 signal would produce objectionable flicker.  This is true.  But flicker at 24fps is, in fact, not specifically a consequence of using a CRT.  Any display, including that of a film projector on a screen, will be perceived to flicker if the frame rate is lower than 45fps – generally reckoned to be the “threshold of flicker perception”.  The pioneers of cinematography (at that time using 16fps) quickly hit on a trick to eliminate flicker perception.  A rotary disc shutter was introduced with three blades in order to add two intraframe blanking periods, in addition to the basic blanking period that is essential during the movement of the film to the next frame.  In effect, each frame of the film was thus displayed three times, creating an artificial frame rate of 3 x 16 = 48fps.  Later, when film speeds were increased to 24fps, the number of blades was reduced to two, thus retaining an artificial frame rate of 48fps (although some 16mm projectors still retain three-blade shutters, so that they can be run without flicker at 16fps when projecting old silent films).

To eliminate the perception of flicker on a video display device such as a CRT, when the content is being played at 1080p/24 (e.g. from a disc), all that needs to be done is to increase the frame rate to at least 48fps at the time of display.  This is done by displaying each frame twice (or more, as desired).

Because 48fps is quite close to the threshold of flicker, most human beings still perceive some degree of flicker in a movie theatre, particularly when viewing bright scenes.  Sadly, projector manufacturers did not reintroduce three-blade shutters for 24fps film projectors once projector lamps could be made bright enough to compensate for the loss of average brightness caused by the extra intraframe blanking period.  But various film formats have attempted to “raise the bar” on real and artificial frame rates, such as Cinerama (2 x 26 = 52fps), Todd-AO (2 x 30 = 60fps) and, most recently, Super Dimension-70 (2 x 48fps = 96fps), which is absolutely free of flicker perception, no matter how brightly lit the screen.

In trying to eliminate flicker perception at home, we are not constrained by anything like the shutter arrangement of a film projector.  But we do need to make sure that HD devices allow us to push up the artificial frame rate as high as we want (within reason).  I suggest that we should be looking for the ability to have an artificial frame rate of at least 72fps, and ideally 96fps, through frame repetition.

Having noted that use of 1080p/24 is no more of a flicker problem than 24fps film, and that flicker perception can be prevented by the same technique in both cases (frame repetition), we can now look at how 1080p/24 measures up as a standard for storage and transport of content, versus other possible formats like 1080p/30.  In the following I consider only 1080-line formats because these are, I think, true HD.  Anything less is not.  Also I consider only progressive scan.  Interlacing was a less-than-ideal expedient to eliminate flicker within the technical constraints of 1950s analog technology and really has no place in the processes of video display, transport, or storage today.

The storage of content (on disc or tape) and the transport of content (via cable, satellite, or broadcast) both consume scarce resources (such as space on a single disc, or bandwidth).  In choosing a format for storage and transport we do not want to insert repeated frames.  Repeated frames waste space or bandwidth.  Of course, the basic signal will generally be compressed before storage/transport (using MPEG-2, or better still, MPEG-4, or a further development of MPEG-4).  In comparing a repeated frame with the previous frame the compression algorithm will represent the second frame as a “zero difference” P-frame, which takes up a relatively small number of bytes.  Nevertheless, this is still slightly wasteful.  Also, using a pre-compression signal with repeated frames gets in the way of the user selecting the repetition rate himself or herself at the point of display.  It makes a lot more sense to store/transport only the distinct frames that appear in the original content.

Most of us will spend a lot more time watching movies on our HD equipment than watching made-for-TV content.  Today, movies are generally in 24fps formats.  It therefore makes a lot of sense to make 1080p/24 the standard for HD-DVDs, so that each frame on the film becomes a frame stored on the disc.  There is already a lot of support for the notion of making 1080p/24 the “Common Image Format” (CIF) for movies, to avoid multiple telecine runs for each movie.  If this CIF is made the standard for HD-DVDs, users with HDMI displays can directly display the 1080p/24 signal (with frame repetition to prevent flicker perception).  This removes 3:2 pulldown from the end-to-end process entirely, as Alen pointed out.  Responsibility would be moved to the HD-DVD player for doing any frame-rate conversions that are needed to provide 30fps NTSC or 25fps PAL analog outputs.  Made-for-TV content (the minority of our HD viewing) that has not been mastered in the 1080p/24 “CIF” would be converted to 24fps before being placed on HD-DVDs.

The Super Dimension-70 standard opens up the possibility that at least some content will start to be available with an actual frame rate of 48fps.  In order to accommodate this we should demand that the HD-DVD standard be able to handle both 1080p/24 and 1080p/48 content, in the latter case preserving all the distinct 48 frames each second.  This could be accomplished by simple flags in the bitstream to indicate where 48fps is in effect.  Note that using 48fps will reduce the number of minutes of content that will fit on one HD-DVD, but not halve it.  The differences between adjacent frames will be smaller at 48fps than at 24fps, so the ratio of the 48fps post-compression bit-rate to its 24fps bit-rate equivalent will be significantly less that 2-to-1.  (It would be interesting and useful to conduct tests to determine the exact ratio.)

I believe that we should regard an actual frame rate of 48fps as part of our long-term HD goal.  The “threshold of continuous motion perception” is generally reckoned to be 16fps (not to be confused with the 45fps threshold of flicker perception: the two are independent perceptual phenomena).  However, the 16fps threshold is for a static camera and does not take into account the effects of panning.  A movie does not become completely free of all perceived motion artifacts until around 48fps (actual frame rate).

Although I have referred to 1080p/24 and 1080p/48 mainly in the context of HD-DVDs, ultimately the use of these formats might be extended to cable and satellite transport.  Their use for HD-DVDs would probably encourage the operators to head in this direction.

Moving on to what happens within our homes, we can consider what we, as consumers, should ask of the HD industry for HDMI interconnectivity (player-to-monitor, cable-box-to-monitor, cable-box-to-recorder, and so on).  In contrast to the scarce resources used in storage and transport, bandwidth on the cables that interconnect devices has practically zero marginal cost.  So, while we seek to optimize the efficiency of storage and transport (for example, using 1080p/24 for standard movies), we should seek the option to move the content between devices at any frame rate we want.

This takes us back to the point about flicker.  My view is that the interests of the consumer will be best served if display devices are flexible (that is, able to handle inputs at 24/25/30/48/50/60/72/96 fps), but which place control of the frame rate in the hands of the sending device (such as the HD-DVD player, D-VHS player, or cable box).  Thus, if I am playing an HD-DVD containing content stored at 1080p/24, I would like to be able to set my player to output that content (over an HDMI) at any reasonable multiple of 24fps that I choose – at 24fps (so I can experiment with how my perception of flicker varies between different display devices); at 48fps (for a “normal” cinematic experience); at 72fps (if I like the “feel” of this mode); or at 96fps (if I want the ultimate in no-flicker viewing).

When I select, for example, 72fps, the player would simply be pumping out each frame it reads off the disc three times.  The display device would be faithfully displaying each instance, without trying to do anything clever itself to manipulate the frames.  Such an arrangement future-proofs my expensive display device.  By contrast, if the manufacturers of display devices include signal processing circuitry which tries to take control of the display process (e.g. deciding when to repeat frames to eliminate flicker, or attempting its own interframe interpolation, or dropping repeated frames), our display devices will be less future-proof and the manufacturers will take away our ability to control our viewing experience.

In summary, since most of us will spend a lot more time watching movies than watching made-for-TV content on our HD equipment, we should strongly support 1080p/24 as a basic storage and transport standard, especially on HD-DVDs (because this removes 3:2 pulldown from the end-to-end process entirely).  Second, we should support an “enhanced format” of 1080p/48 to handle content that has been filmed at an actual frame rate of 48fps (or up-converted to 48fps by computer-based interframe interpolation).  Third, we should not think “flicker problem” when we read “1080p/24”.  Rather, we should think of flicker management as something that is handled at the time of content display.  And fourth, we should ask manufacturers, as they implement HDMI, to place control of the frame rate in the hands of the sending-end device, not in the display device itself.

Malcolm Hamer

New York

Those Dirty Discs

Sent to Widescreen Review in 2004

Dear Gary,

I was very pleased to read your editorial about the problems caused by discs being “naked”.  I am always horrified by the state of the DVDs that I rent, even ones that have been in circulation for only a couple of weeks.  What on earth do people do with their rented DVDs?  They are almost always covered in scratches, finger marks, and other gunk.  I generally have to wipe them thoroughly before they will play properly.

The naked design of the audio CD, in 1982, went against two decades of development in consumer media, during which cassettization became a universal approach to protecting the information-bearing surface (audio cassette, mini-audio cassette, floppy disks, VHS, Video 8, MiniDisc, and so on).  The designers of the CD ignored this trend, probably because of a desire to minimize the cost of CD manufacture, knowing that the error-correcting code would mask read errors due to dirt and minor scratches.  Unfortunately, this decision set the scene for VCDs, and then DVDs, to be naked also.  In 1993, to make it possible to use CD-production plant and drives for VCDs, the designers of VCD technology stuck with the naked disc, in spite of the likelihood of a significant rental market for VCDs and the consequent less-careful handling that VCDs would be subjected to.  During the next three years the designers of the DVD settled on a naked disc, presumably because of their desire to make DVD players able to play CDs and VCDs (and possibly because they thought that consumers would expect a DVD to look like a CD).

From the remarkable and largely unexpected success of the DVD, it is clear that consumers will gladly pay for quality and convenience.  It is also clear that, contrary to initial fears, nobody really minds having two, or even three, separate video-playing boxes in their video-viewing room (a DVD player, a VCR, and a TiVo).  Had the designers of the DVD known what a runaway success the DVD would become, they would have been much less timid about breaking away from the physical design of the CD.

The opportunity now exists to make the right choice for the design of the HD DVD, given that (a) the much higher information density on HD DVDs will make them even more vulnerable to the effects of scratches and dirt, and (b) consumers will probably be happy to add a fourth video-playing box to their pile of boxes if it delivers high quality and greater convenience.  (An HD DVD design that places the disc in a plastic enclosure will certainly be much easier to handle than a naked disc.)

I, for one, would appreciate the convenience of not having to wipe the surfaces of rented HD DVDs before playing them; and I am certain that the rental companies would welcome much lower damaged-disc losses.

Malcolm Hamer

New York

GSM is global, not "European"

Sent to Business Week in 2003

Several U.S. Corporations have had spectacular successes in the last thirty years as global leaders in technology, establishing standards that have been adopted worldwide (for example, IBM, Microsoft, and Cisco).  However, U.S. corporations can never be expected to score 10 out of 10.  Yet Business Week sometimes seems to shy away from mentioning cases where the U.S. is a technology follower and not a leader.  One such case is mobile telephony.  The global standard today is GSM and has been for almost a decade.  The U.S. cellular service providers and equipment builders made a huge mistake in deciding not to join the “GSM club” early on.

GSM is more than just one of several options for the design of the radio transmission components of a mobile phone.  Adoption of GSM by a network provider means that the provider can join the “global GSM club”, so that the provider’s customers can roam internationally in other GSM countries and overseas visitors can roam via the provider’s network.  Outbound and inbound roaming bring effortless incremental revenues to the provider in both cases, especially the latter.  Joining the global GSM club also means that the provider’s customers can exchange SMS messages with any GSM network subscriber globally, regardless of whether the sender or recipient is in his or her home country or is roaming in another country.

A large number of BW’s readers travel internationally from the U.S., as I do, or live in GSM countries.  They must be as puzzled as I am when they read, almost weekly, in BW the phrase “Europe’s GSM standard” (most recently in “Go East, Young Chipmaker”, December 30/January 6 issue).  Your message to readers seems to be that there are many regional and local standards and it is therefore quite understandable that the U.S. providers have their own standards.

GSM is not “Europe’s standard”.  It is the most important global standard, adopted from the early 1990s in all major markets except the U.S. and Japan.  It does not matter if theoretically more efficient standards exist.  Like VHS (versus Betamax), it is where the world is at.  Get over the fact that the U.S. was not the global leader on this one.  In future, write “GSM, the global standard that the U.S. failed, initially, to adopt”.

In spite of the initial failure of most U.S. providers to recognize the importance of GSM, at least one brave provider saw GSM’s potential – VoiceStream (T-Mobile).  VoiceStream, as most users still know it, built the first national GSM-compatible network in the U.S., using the GSM standard, but operating in the 1900 MHz band (presumably because the 900 and 1800 MHz bands were already used for other purposes).  They then persuaded the major phone manufacturers, such as Ericsson and Nokia, to add the 1900 MHz band to many of their models to create “tri-band” phones that could roam between the U.S. and the rest of the GSM world.  Once these phones hit the market around the world, VoiceStream effortlessly tapped into lucrative international roaming revenue, as overseas visitors switched on their phones on arrival in the U.S., locked on to VoiceStream, and started making domestic and international calls.

AT&T Wireless has now seen the wisdom of VoiceSteam’s strategy and has embarked upon a rapid roll-out of a national GSM network, providing its customers with promotional discounts to migrate from TDMA to GSM.  This is a bold and visionary move which deserves praise and mention in BW.  And please do not forget to say that they are migrating to the global GSM standard.

Malcolm Hamer

New York