Tuesday, April 30, 2013

Open Science: Inverting Metagenomics Pipelines - 10 things you may not know about RPS-BLAST

1 May 2013- Singapore. This is an Open Science post intended to spark ideas and conversations with other scientists.


I was thinking about some of the assembly problems that have been described in large metagenomic (e.g. soil) samples by @ctitusbrown, @phylogenomics and many others.

To start with - I am not an “NGS” metagenomics expert. Iam a “FGS” metagenomics guy. We published log-odds scores for detecting over 100 species by amino acid composition in 2002. (http://www.biomedcentral.com/1471-2105/3/39). I know a thing or two about large scale bioinformatics pipelines and optimizations.

So. Here are my current thoughts about inverting the metagenomics assembly pipeline.

In a nutshell:

Short read -> RPS-Blast -> PSSM Hierarchy Bin -> Bin Assembly -> Bin contigs -> RPS-Blast hybrid ejection -> Function Assignment -> Phylogenetic assignment -> Taxon Bin -> Assembly. 


Inversion? How is this possible?

First - the "small database effect".
What is the small database effect? It is the inverse effect of the large database problem. Google "cancer" and you get (today) 644 million hits. The larger the database, the more hits you get. A smaller database gives you correspondingly fewer hits.

Consider that the number of protein functional classes (PSSMs or HMMS) is smaller than the number of possible species. Organisms are built on common protein building blocks, and there are more ways to organize the blocks and encode them (i.e. species) than there are blocks themselves.

In principle, searching for the function of a short read should outperform a search for the taxon of a short read, in terms of both hit quality and number. Can this work in practice? I think so. Here’s why. (I have tried out steps 1-3 they work from the command-line RPS-BLAST, which got me started on this idea).
  1.  RPS-BLAST can find PSSMs that match sequences of 9-12 amino acids. Change the threshold E-score to around 100. Multiple PSSMs may be hit in the process, and some of these may be false positives. See point 3 about reducing the number of hits by up to an order of magnitude, and points 6 about false positive hits.
  2. Short reads on the order of 30bp can be input directly into RPS-BLAST as nucleotides. Select the appropriate genetic code. Assume everything encodes protein and close your eyes for the moment.
  3. CDDs are not independent, rather they are curated. The PSSMs are organized into a hierarchy of evolutionary families and superfamilies. Most hits are in fact within a branch of the hierarchy. The collection of over 40,000 PSSMs is itself redundant and a last-common family or superfamily can be readily found by traversing the ancestral hierarchy. This can collapse the number of candidate PSSM hits to the short sequence considerably. In some cases, by an order of magnitude. For example 50 PSSM hits can collapse into 2-5 superfamily CDD PSSMs.
  4. With the assignment of lowest common ancestor CDD PSSMs to the short read sequence, one can use the resulting PSSM / CDD identifier as a database key for binning. A short read sequence can be assigned to multiple PSSM bins without harm at this point. By grouping together all the short-reads into PSSM based bins, one can subdivide the metagenomics dataset into potentially related protein functional units by putative peptide sequence.
  5. Once the reads are binned, one can assemble the sequences in each PSSM bin into contigs independently, by feeding the binned read set into a conventional assembler.
  6. One can remove erroneous hybrid contigs from the bin by a second RPS-BLAST scan against the parent PSSM or CDD PSSM branch hierarchy. This is a tiny fraction of the entire PSSM set. At this stage the successful contigs should represent species fragments with coding regions based on nucleotide and reading frame overlap, with minimal interference from random overlap. The resulting set of contigs is functionally related to the PSSM, so the assignment of function is already in hand.
  7. Digital normalization could be applied to the contigs within the PSSM bin to remove redundant information for further genome assembly. There will be a pile of un-binned reads as well that do not map to any PSSMs. These could be assembled/digitally normalized separately from the binned sequences, and merged for the final assembly pass.
  8. The PSSM itself could be used a scaffold for sub-assembly. That will require some code, but it is probably not necessary.
  9. The processed contigs can be sent to another system in the pipeline for phylogenetic assignment (e.g. BlastX ). Final assembly can performed assuming species specific sequences are grouped better by the preceding steps.
  10. PSSMs are divisible and thus can be cut up and subdivided into smaller PSSMs. If there are unmatched portions of the original PSSMs in a set of contigs, they can be chopped into mini-PSSMs, related in the existing CDD hierarchy to the parent PSSM we cut them from. This now changes the search statistics. A new custom RPS-BLAST database can then be compiled out of these leftovers. The unassigned or discarded short reads can be searched against this. This mini-PSSM database will be much smaller in size than the original. This could further extend the previously found contigs by assigning new, weaker hit contigs into the existing PSSM bins up the hierarchy. 


Feel free to substitute HMM for PSSM in most of this. However I have no idea whether implementations of protein family HMM searching can use short nt sequences as input (points 1,2), or how to achieve the hierarchical reduction in number of hits without an evolutionary model for the entire set, as curated by CDD (point 3), or whether a HMM can be simply divisible (point 10) without a lot of re-processing.

Tuesday, October 23, 2012

The mouse trap, redux.

The Hook of Job: 
History reduces the irreducible.

tldr? - scroll down to Fig 5 to skip the details.

            The historical provenance of the mouse trap's unique design back to 1847 reveals its inventor, Job Johnson, and that it is reducible to a functioning single part animal trap, the fish-hook.



 Figure 1. A modern Victor Brand mouse trap with bait-pedal up,
showing the vestigial profile of the fishhook, from which it originated.

          In his book Darwin’s Black Box1, and follow-up The Edge of Evolution2, Biochemist and Intelligent Design instigator Michael J. Behe uses the mouse trap as the defining example of a device that is irreducibly complex. He explains how it can not function without all of its parts, and that none of its parts, alone or in various combinations, can do the function of the entire trap. And he explains how the trap could not have been created by a small succession of modifications to some simpler precursor that performed the same function of mouse trapping.  Niall Shanks, in his critique of Intelligent Design3, made an effort to address the historical origin of the mouse trap, but could not get past what seems like a popular myth when in fact there is much older supporting evidence for the origin of the mouse trap in both the Patent Office and in antique mouse trap collections.

          The origin story for the mouse trap is important because it maps the progress of a complex idea. It is at the first moment of invention of the mouse trap that marks the start of the idea of its complexity. It is in the original trap that the human intelligent design work was performed. The complex mouse traps designed thereafter are a cascade of copycat follow-ups with incremental changes to the original design. The idea of a snap-style mouse trap is an idea that, once it started, became hugely popular.

          If you could collect one example of each mouse trap produced every year after its original invention and place each trap on a long table side by side in chronological order, how do you think these mouse traps would appear to have changed over time? How far back in time would we have to go to find the original snap-style mouse trap? Would the original snap-style mouse trap even look the same? The long line of mouse traps we see before us would be a record of the legacy of small improvements to the original snap-style mouse trap design. 

          While we have no table of mouse traps handy, there is an excellent record in the U.S. Patent office that documents, illustrates and patents each important improvement in its evolution. For example, today some Victor brand mouse traps4,5,6 are sold “Pre-Baited” with yellow plastic bait parts shaped like little slices of Swiss cheese impregnated with a chemical scent to attract mice, a high-tech improvement eliminating the need to supply and load bait. But this chemical innovation is really just a small incremental change to the previous stamped metal bait tray7. Everything else is the same. If the plastic was soft enough to gnaw on and could be further impregnated with a mouse-specific poison, perhaps the rest of the mechanical parts of the mouse trap would become dispensable, converting it from a trap to a poison. So minor innovations, like the new “Pre-Baited” trap, can accumulate and obscure the details of the original mouse trap design.




Figure 2. Patent drawings of the pre-baited cheese-shaped bait pedal. 

          This is precisely why we must go back in time and conduct some research to try to find the original invention and an authentic artifact or illustration to comprehend its design. The original inventor of the mouse trap had no mouse trap to look at or think about, just a problem to solve: how to trap and kill a wild animal busily gnawing away at his food stores in the cold of winter. Would Behe’s questions about irreducible complexity and the nature of the parts of the mouse trap hold true for the original first mouse trap? Or would the original device betray Behe with answers that would prove to be the undoing of the most fundamental straw-man of the Intelligent Design argument?

          The snap-style mouse trap design is what Richard Dawkins8 calls a meme, which he defines as an original thought or first of a kind idea that has been copied and repeated many times since its origin. The mouse trap idea began with the very first such trap, and it is at the time of origin of this idea where the analysis of the irreducibly complex features must be properly considered, not the countless copies that follow or their incremental changes to the original design.

          Let’s look at the history of the mouse trap to understand this. Now, the modern mouse trap is a very common item, simple enough to be understood mechanically, and very easy to illustrate. Definitive statements can be made about today’s mouse trap mechanism and its complexity. In his argument, Behe asks his readers to imagine how well the modern version of a mouse trap would work missing one, two or even more parts. But we know nothing of earlier designs, so this puts the reader in the position of having to imagine a design history for the modern mouse trap. For us to conclude that there could be no simpler form of the mouse trap, based on the very sparse imaginary history we fabricate in our minds, is a flawed way of thinking. Like the many versions of the Robin Hood story we have seen, starring a broad range of stars, from an animated fox to Russell Crowe, there can be a big difference between an imaginary history and a real history.

          Since old mouse trap designs are obscure, we must look up some of the mouse trap patents and find old mouse traps in collections in order to follow them back to their origin. There are thousands of illustrations of mouse traps filed with the U.S. Patent Office and other patent offices worldwide. By looking back through old patent documents, we can find many forms of mouse traps, some with recognizable snap-style trap features, but also many others with springs in strange places or iron parts where wood is expected. So it is better to consider the information in the patent records and in actual artifacts as original sources of information.

          Who invented the mouse trap? A crowd of inventors might say “I did!” each raising his hand. With over 4,400 U.S. patents on mouse traps, many can claim to have invented a mouse trap. But who was first? The mouse trap’s invention story is remarkably convoluted and obscured by thousands of inventors all trying to “build a better mouse trap”. More problematic is that the most successful American mouse trap company, Woodstream, has perpetuated, in the popular media, a myth about the invention of the mouse trap, attributing the modern mouse trap design mostly to its founder, John Mast.

          In fact there seems to be two separate corporate mouse trap origin myths in publications, one American and the other British. The good folk of Lititz PA would point to their long-running company, Woodstream, makers of the famous Victor brand mouse trap, and credit company founder John Mast with the invention in 1899 which was patented in 19039. Although some popular articles10 correctly place the year of invention of the Victor Brand mouse trap to 1894, that is not the date of the Mast patent. In an interesting parallel, folks more familiar with the British “Little Nipper” mouse trap would credit James H. Atkinson with the invention in 1897. His British patent (GB 13277 of 1899) was sold in 1913 to a company named Proctor for 1000 British Pounds. Both Proctor and Woodstream have been making mouse traps for a very long time, so it is no surprise that each takes credit for the invention, though separated by vastly different markets and an ocean.

          Now, while Mast and Atkinson are the inventor-founders of the two major mouse trap companies, the race to perfect and market a cheap and reliable mouse trap was not new in the late 1890s, the period in which they started their respective companies. Their own patent documents show that they borrowed many design ideas from other trap inventors. One earlier patented mouse trap design11 is more similar to Woodstream’s modern Victor brand design than any other, even closer to it than Mast’s own 1903 patent. It was the 1894 design of William C. Hooker of Abingdon Illinois. Hooker founded The Animal Trap Co. of Abingdon Illinois with his invention. According to mouse trap collector and expert Rick Cicciarelli, The Animal Trap Co. first marketed their unmistakably modern looking mouse trap as the “Out O’ Sight” mouse and rat trap in two different sizes. 

Figure 3.  1894 W.C. Hooker snap style "animal trap".




Figure 4. Hooker 1894 Design Patent Animal Traps - "Out O' Sight Rat Trap and Mouse Trap.
Photo courtesy Rick Cicciarelli.

This is important because the design of the snap-style mouse trap and the rat trap are both identical designs, just scaled copies of the same design, one mouse-sized, and one-rat sized. Cicciarelli tells us that the company that Mast had founded acquired Hooker’s design in 1905. This explains why sometimes the Victor mouse trap is credited with being invented in 189411, the date of the Hooker patent, which became their intellectual property.

So it was Hooker’s design that became the modern mouse trap, but even his design was also predated by earlier traps and patents. And while Mast was indeed a master of turning the idea into a low-cost product, the essence of the snap-trap was not his idea, nor even Hooker’s idea, for that matter. The origin is found further back in time, back when trapping animals for food was far more important than trapping pests like mice and rats.

To find the original mouse trap design idea, let us try to define the essence of the mouse trap. How do we describe the unmistakable part of the mouse trap that can be recognized by its structure as the invention? The snap-style mouse trap is distinguished by a U-shaped bar that travels 180 degrees from one side of a thin rectangular platform of wood, to the other. It is powered by a coiled spring under torque, not by compression or stretch. And it is triggered by the mouse moving a bait platform, releasing a catch bar, and allowing the torque spring to move the U-shaped bar forward with deadly force. A rat trap is just a bigger version of the mouse trap. So we must also consider that the very first design could have been either in the form of a mouse or a rat trap. So let us restate the “Who invented the mouse trap?” question to reflect the details of its successful design idea; “Who invented the snap-style mouse/rat trap with wooden base, bait trigger mechanism, perpendicular torque-spring and U-shaped kill bar traveling over 180 degrees?” Hooker’s patent satisfies this description. Are there older ones?

After examining snap-style mouse trap patents, what varies most in the design, generally speaking, is the trigger mechanism and bait tray. This is the part that holds the cheese that the mouse or rat gnaws on that triggers the trap to snap shut. The bait tray has been changed and refined over time by many inventors, including the modern fake plastic chemical-scented Swiss cheese version. You can see the difference in triggers on the Victor mouse traps for sale today4,7 as compared to Hooker’s patent11. The modern bait tray and trigger is much more sensitive than the original design and stamped from a single piece of metal. We can see a number of small improvements in trigger sensitivity as Hooker’s design worked only when pressed down. Hooker’s bait tray and trigger was made from three separate pieces of metal but the improved single-part bait tray and trigger senses chewing motion in any direction whereas earlier ones could be defeated by clever mice who knew how to chew in the right direction. I imagine that sideways-gnawing mice may have escaped with their lives and a cheek full of bait cheese, and gone on to breed even more sideways-gnawing mice, were it not for these numerous slight modifications to the bait pedal and trigger mechanism over time.

Now let us go back further, as several patents predate Hooker’s 1894 patent. The C.B. Trumble Patent of 189212 and the W. H. Castle patent of 188813 are key examples of what patent lawyers call “prior art”. These older designs retain the essence of the snap trap design as I defined it above, but with variations that include different triggers and cast iron bases instead of wooden bases, and double torque springs instead of a single torque spring, in the case of the Castle patent. John Mast referred to both of these earlier patents in the text of his own 1903 Patent9. In addition, Mast referred to two patents from 1855 that were granted to Lucien B. Bradley for rat trap designs14,15 but he avoids mention of the Hooker Patent11 in his filing, instead drawing attention to the older designs, which helps our quest.

The 1855 Bradley rat trap patents14,15 were powered by springs that were compressed and released and were oriented in a straight line to the bait, rather than in the perpendicular torque spring arrangement. This may have been an effort to alter an earlier torque spring design as a patent workaround. With the 1855 patents of Bradley, we are close to the origin point, and nearly half a century before the date of the Mast patent which is so commonly mistaken for the original.

Rick Cicciarelli is an avid collector and authority on the history of antique mouse and rat traps, and he has read through all the US patents in his quest as a collector. Cicciarelli owned a prized snap-style rat trap marked “JOB JOHNSON BROOKLYN NY PATENTED 1847”, which he bought at an antique store as a boy for only $30. He believes it is the earliest flat snap-trap design. In his correspondence to me, Rick says, “Generally speaking, that flat snap trap design is attributed to Hooker. However, I have a flat snap rat trap which was patented in 1847, and I believe THIS trap to be the earliest flat snap trap design.”



Figure 5. Rat trap, marked 

“JOB JOHNSON 
BROOKLYN NY 
PATENTED 1847”.
Photo courtesy Rick Cicciarelli.


Upon examination, the antique trap of Job Johnson does fully satisfy our question of the essence of the mouse trap design idea, with wooden base, U-shaped kill bar, and torque spring. But the patent of 1847 to Job Johnson16, assigned the very early U.S. Pat. Number 5,256, was not the design of a rat trap. Rather, it was one of the three first novel designs for a spring-loaded fish hook. Johnson’s fish hooks are gloriously illustrated in photographs in a collector’s book on spring-loaded fishing tackle and fish traps by William Blauser and Timothy Mierzwa17. This text of patent number 5,256, being sworn and witnessed statement to the U.S. government, tells us part of Job Johnson’s story:

Be it known that I, JOB JOHNSON, of the city of Brooklyn, State of New York, fish hook manufacturer, a native of England, having been resident more than one year next preceding the date hereof in the United States, and having duly declared my intention to become a citizen thereof, have invented and made and applied to use certain new and useful improvements in the constructive application, arrangement, and combination of mechanical means whereby the bite of a fish at the bait on a hook causes a crooked barb-dart to strike into and hold the nose, head or gills of the fish, independently both of the line and of the person holding the line, and the general arrangement of which, when of a proper size, may be applied to the capture of any kind of fish or of any destructive or ferocious animal, and for which improvement I seek Letters Patent of the United States;16

Figure 6. Spring-loaded fish-hook illustrations from 1847 Job Johnson Patent 5,256.


Job Johnson, as it turns out, was a prodigious inventor. His legacy was nearly lost to history but it has been recently rediscovered by Dr. Todd Larson, a historian at Xavier University. Larson is an expert on Job Johnson’s inventive legacy, which can be found in Larson’s book, The History of the Fish hook in America18. According to Larson, Job Johnson was a prolific American inventor with 38 patents ranging from fishing tackle to elevated railways, demonstrating his very broad creative capabilities. Johnson got rich from a thriving automatic fish hook manufacturing business. This success obviously put him in a position to explore commercial designs for traps for other animals. His experience with springs and wire, and his workshop, filled with springs and triggers from his work on the spring-loaded fish hook, would have been the perfect place to experiment with other forms of traps.

We know, thanks to Dr. Larson’s research, that American farmers needed rat traps and thought about using fish traps to catch rats, as is mentioned in an article in the 1847 edition of The Prairie Farmer. Entitled “A New Fish hook”19, the article described Job Johnson’s fish hook and concluded, “Those who wish to catch rats have got the right machine here.” So how did Johnson make the move from spring-loaded fish hooks to the rat trap? According to Cicciarelli it is pretty obvious to the naked eye that Job Johnson used his patented spring-loaded fish hook trigger design, stuffed the spring-loaded fish hook mechanism in a hole in the wooden base, and rigged the torque-spring with a vicious serrated U-shaped striking bar. In Cicciarelli’s words:

You can't really see the mechanism of the trap very well from the photo, but the bait hook is actually a long fish hook, the end of which goes through the base and comes out at the rear to hook into the jaw when the jaw is pulled back into the set position. It works just like the spring hook.

So the Job Johnson rat trap actually contained a copy of his spring-loaded fish hook as a bait platform and trigger mechanism. As such, it was fully covered by the wording and considerations in U.S. Patent 5,256 and so the rat trap was duly stamped “PATENTED 1847”.

Johnson was already a well-known fish hook maker, having started the first American effort in their fabrication in 1843. While now banned as unsportsmanlike, spring-loaded fish hooks were, in their day, an important way for working fishermen to maximize their catches. Todd Larson devotes an entire chapter to Job Johnson’s inventions and says that he was a man “whose hooks were so good they inspired poetry.” Truly, Job Johnson’s fish hooks are mentioned repeatedly in the 400 line ballad “The Legend of the Great Tautog” written by an anonymous author and published on 23 October in The Spirit of the Times, in 1852. According to Dr. Larson, Job Johnson’s fish traps were known to be capable of catching small game, including rats, simply by hanging them above the ground by a string and baiting them.

Larson’s careful research shows that his patent 5,256 for the spring-loaded fish hook was, in fact, an improvement on the first spring-loaded fish hook invented by a 16 year old boy in 1845 named George Washington Griswold18. Griswold’s design was assigned and patented by entrepreneur Englebrecht and lawyer Skiff in 184617,19, and given U.S. Patent number 4,670. It was the first U.S. patent involving a device to catch a fish17.

Griswold used a flat spring to cause two hooks to close on the mouth of a nibbling fish. Job Johnson improved the Griswold design with a more powerful contractile helical spring, driving the two hooks together. Job Johnson was a natural at spring-making, probably from his early training with iron wire fabrication for fish hooks. So we know Johnson had the skills to design and assemble the parts of the first rat trap. And we know that Johnson contemplated that other animals could be trapped with his spring-loaded fish hook by the words in the patent text itself16. Finally, we know that Cicciarelli’s prized artifact shows that Job Johnson fabricated rat traps containing the exact same spring-loaded fish hook mechanism, with the added torque spring and kill bar, and mounted on a flat piece of wood.

Well, this is where we come to the end of the line of commercialized mouse and mouse traps, where we run out of artifacts and patents. Older inventions may have existed, but they were not spread as ideas. There was a wave of innovation in the mid 1840s involving a proliferation of patents in fish traps, popularized by many articles in the early editions of Scientific American18 as one of the carriers of design ideas in its day. The starting point of all this innovation was from the 16 year old Griswold’s first spring-loaded fish trap.

As I mentioned, collectors Blauser and Mierzwa have a wonderfully illustrated book showing these early spring-loaded fish hook designs17. Their photographs show that Johnson’s first design was a spring-loaded fish hook from 1846 fabricated with three metal parts, three rivets and one spring (page 16) but no corresponding 1846 patent was found matching this artifact. Interestingly, his 1847 patented device was altered to be made with 4 metal parts, four rivets and one spring (page 21). So Job Johnson likely took the Griswold design and made at least two successive, slight modifications to create a superior spring-loaded fish hook design that would continue to be sold into the 1900s. 



Figure 7. Job Johnson 1846 and 1847 spring-loaded fish hooks.
Arrow points to additional part and rivent added in the 1847 design.
Photos courtesy Tim Mierzwa.


There is little doubt that both Johnson and Griswold had other prototypes made in between these that failed to work. While most of the failed prototypes of the earliest fish traps and rat traps are lost and long forgotten, the additional 1846 fish trap of Job Johnson is evidence of a prototyping process prior to the broad spread of the idea and more inventions in spring-loaded fish traps, mouse and rat traps.

Thanks to Cicciarelli, we can conclude that Job Johnson was the earliest known inventor and original spreader of the snap-style rat and mouse trap idea. Yet it was a branch of an idea started by the young Griswold, a simple idea about how to trap a fish, which was modified into a snap-style rodent-killing machine. Like a viral predator, spring-loaded fish traps crossed the species barrier to become rat trap and mouse traps, providing two independent and successful lineages of traps, one trapping food, and the other for getting rid of destructive pests.

So finally, now we can go back to the context of Behe’s steps1 to determine irreducible complexity and apply them to the Job Johnson rat trap. According to Behe, the first step is to specify the function of the system and all the system components. The second step is to ask if all the components are required for the system function. Well, the system function of the rat trap is to lure in unsuspecting rats and immobilize them so they can no longer render havoc on stored food supplies. So the system function is unchanged; it is just adapted for a larger rodent.

Now, we also know that the spring-loaded fish hook used inside the Job Johnson rat trap was itself a standalone animal trap, so we can state conclusively that not all the components are required for the system function. Importantly, this is where the answer changes from yes to no. Not all parts are required for the entire system to function. And in fact we can dispose of all but a single part and still retain system function. The simple one-part animal trap, the fish hook, is clearly visible as the bait holder in the Job Johnson rat trap. So one part of the Job Johnson rat trap is a physical precursor that remains largely unchanged throughout its design: the fish hook.

Griswold started off with a design that duplicated the fish hook into a grabbing mechanism in a configuration like the gripping talons of a bird of prey. Johnson’s intermediate work shows increased part counts in his first two spring-loaded fish hook designs. He added one part and one rivet in a step-increasing complexity in a small increment to provide better leverage for the trigger mechanism. It is clear from the historical accounts that people used spring-loaded fish hooks by themselves to catch small animals. So the piece of wood is dispensable, as is the torque spring and kill bar. It is just as likely that one could catch and kill a rat with a baited barbed fish hook as one could a fish, provided you could stay awake long enough to wait for a rat to bite the hook in the middle of the night.

So we have now uncovered that the irreducibly complex mouse trap is a conclusion made by a convenient omission of a forgotten history. The original invention is reducible to a functioning single-part animal trap, the fish hook, retaining the same system function throughout the transition. The historical lineage of the snap-style mouse trap comprises evidence showing that the mouse trap is an example of reducible complexity, thereby disproving the notion that it was irreducibly complex when it was originally created. The mouse trap voyage through time takes us back to the fish hook. And as I show in Figure 1, on the current Victor mouse trap7, the profile of the metal bait pedestal is, remarkably, still shaped with the same curve as a vestigial fish hook.

Job Johnson is the most likely inventor of the mouse trap design. Why don’t we know more about Johnson? Dr. Larson’s book tells us how Johnson’s last few inventions and investments were considered the work of a crackpot, as he suffered from senility in his 80s18. His inventive reputation suffered greatly as these failures mounted. Johnson’s senility left a poor impression on the historical record of his later years, and his earlier successes were overlooked as things like spring-loaded fish hooks fell out of popularity for being unsportsmanlike. And so the story of the invention of the mouse trap may be obscured by Mr. Johnson’s own illnesses later in life. Yet his highly successful 1847 fish hook design continued well after the patent expired, sold by the Sears and Roebuck and Montgomery Ward catalogues into the early 1900s. Only two examples are known of the Job Johnson rat trap, and there are also very few examples of the original Job Johnson spring-loaded fish hooks in the hands of collectors. Oddly, the spring-loaded fish hook line of inventions starting with Griswold was destined for extinction, while the early diverging line of rat and mouse traps is still successful, and still being used today.

Recall that long table and line of old mouse traps we were going to set up? Well we know it goes back to 1847 and we know that next to the first Job Johnson rat trap we must put the first three spring-loaded fish hook designs, two of Johnson’s, and the first from Griswold. The mouse trap line is a branch from another long line of inventions – the spring-loaded fish hooks. Men invented machines to trap pests at the same time they were inventing machines to trap food. At the very beginning of the two lines of traps lies a single-part animal trap called the fish hook, the manufacture of which was Johnson’s trade, a set of skills passed from father to son.

So let me more boldly ask whether any other device of complexity is, historically speaking, going to withstand the kind of scrutiny we just gave to the mouse trap design? Of course I will not suggest the mouse trap evolved without human intelligence. After all, it is a product of human design. The key point of the historical study is this: when you stop and identify individual intelligent human designers as individuals and you look carefully at how they design things, you see that they always apply their intelligence in small doses of creativity in the context of prior knowledge. Incremental additions to prior designs and prior knowledge are the way humans achieve creative new designs. This is a principle recognized by the Patent Office and the patent process. Small changes are how intelligent human beings get to complexity. The true nature of human intelligent design is that humans design complex systems in a step-wise approach, not in an all-at-once magical fashion. Simple designs, such as the mouse trap bait pedestal/trigger part, are made up from small improvements over time borrowing from earlier ideas. The Patent for the silly Swiss cheese shaped and scented bait pedal5, invented in 1981 and made ornamental in 19896 , makes reference to an earlier cartoonish mouse trap with decorative holes from a design patented in 194821. Again, simple ideas combine into a minor modification4 to an existing mouse trap design7.

Inventions accumulate new parts with variations on existing design memes just as evolving creatures accumulate new genes based on variations of prior ones. The key is that the knowledge is stored, either in memories, in writing or in the form of the artifact itself. In living creatures the memory of the prior prototype is simply the DNA, the gene which encodes the part. Each gene is an accumulated store of information about the successful small innovations in the mechanical parts, the proteins, within a cell. No intelligence is required in evolution because the memory of prior prototypes does not require an intelligent being to extract the information out and copy it and make small modifications to it.

Now, in the case of human design, many other examples exist, and the patent record holds many of the forgotten details. I invite you to scrutinize the patent history of any object foisted as irreducibly complex. So far there are no examples of spontaneous intelligent designs of inherent complexity that withstand proper scrutiny, and if there are any that appear to, it is simply because through the historical evidence of prototypes, trial and error has not been preserved to tell the story. Even the modern design complexity research community itself acknowledges that complex product design is a process of small step-by-step improvements to prior designs22, and that this is the preferred way our own most intelligent engineering teams pool their efforts to approach complex design tasks.

My conclusion is that human intelligent design is itself a process more similar to the evolutionary process than it is different. Only a few design ideas are successful over the long term. Even successful ones, like the spring-loaded fish hook, can become obsolete and disappear in a process rather like extinction, while another related design, the mouse trap, thrives. One of the benefits of having the Patent Office and its process is that these design ideas are captured for all to examine and modify and reproduce. Patents are the collective DNA of our human innovative genius and the genome of the industrial and technological revolutions

References


1.              Micheal J. Behe. Darwin’s Black Box. The Biochemical Challenge to Evolution. 1996 Free Press, New York USA.

2.              Micheal J. Behe. The Edge of Evolution. The Search for the Limits of Darwinism. 2007 Free Press, New York USA.

3.              Niall Shanks. God, the Devil and Darwin. A Critique of Intelligent Design Theory. 2006, Oxford University Press, Oxford UK.

4.              Victor® Brand Easy-Set Mouse Trap (Part No. 19032P), Woodstream, Lititz PA, USA.

5.              USPTO 4245423. 1 Dec 1978. Anthony J. Souza and Joseph H. Bumsted. Animal Trap.

6.              USPTO D300163. 7 Mar 1989. Harper Landell and Donald W. Warren. Bait Pedal for a Mouse or Rat Trap.

7.              Victor® Brand Mouse Trap (Part No. 19032P), Woodstream, Lititz PA, USA.

8.              Richard Dawkins. The Selfish Gene. 1976. Oxford University Press, Oxford, UK.

9.              USPTO 744379. 17 Nov 1903. John M. Mast. Animal-Trap.

10.           Joseph Rosenbloom. Snap, Crackle, Pop!  Inc. Magazine, May 2000. (http://www.inc.com/magazine/20000515/19003.html)

11.           USPTO 528671. 6 Nov 1894. William C. Hooker. Animal-Trap.

12.           USPTO 481707. 30 Aug 1892. Chauncey B. Trumble. Animal-Trap.

13.           USPTO 391118. 16 Oct 1888. William H. Castle. Animal-Trap.

14.           USPTO 12892. 22 May 1855. Lucius B. Bradley. Rat trap.

15.           USPTO 13843. 28 Aug 1855. Lucius B. Bradley. Trap for Catching Animals.

16.           USPTO 5256. 21 Aug 1847. Job Johnson. Improvement in Fish hooks.

17.           William Blauser and Timothy Mierzwa. Spring-Loaded Fish hooks, Traps & Lures. Identification and Value Guide. 2006, Collector Books. Paducah KY USA.

18.           Todd E.A. Larson. The History of the Fish hook in America: An Illustrated Overview of the Origins, Development, and Manufacture of the American Fish hook. Volume I 2007, Whitefish Press, Duluth MN USA.

19.           A New Fish hook in the 1847 edition of The Prarie Farmer, op cit18.

20.           USPTO 4670. 28 Jul 1846. Theodore F. Engelbrect and George F. Skiff. Improvement In Fish hooks.

21.           USPTO D155513. 8 Sep 1948. Charles E. Jones. Design for an Animal Trap or The Like.

22.           Eppinger, S D; Whitney, D E; Smith, R P; Gebala, D A. A model-based method for organizing tasks in product development. Research in Engineering Design 6, 1-13, 1994

23.           Dagg, J.L. Exploring Mousetrap History. Evo. Edu. Outreach. 4:397-414, 2011

Copyright (c) 2012 Christopher W. V. Hogue. All Rights Reserved.

Saturday, February 25, 2012

Fun with Proteins and how Proteins Really Fold...


Today, I spotted a post on Facebook with a photo of some of my cousin's kids playing Twister. I was knee deep into making figures with Ramachandran Plots, and structures for a paper I am working on. I thought it might make a good analogy for protein folding. So I did some artwork. See what you you can spot in the details...
Enjoy.

Sunday, February 12, 2012

The History Delusion: Intelligent Design Gerrymanders Tesla

Last month Vincent Torley, via the Uncommon Descent website, (link) posted a pre-emptive attack against my blog series. His tactics included rolling out anonymous faux-experts doubting my credibility, rants on probability, the Cambrian explosion and even some inspirational poetry. I must thank him for writing exactly what I anticipated. That is, he took a position that the process of human design is solitary, mysterious, and magical, based on the solipsistic autobiography of Nikolai Tesla.


Gerrymandering Tesla

In addition to defending science and evolution, my argument is this. Intelligent Design gerrymanders historical evidence to support its claims against science and evolution. I use the term gerrymander as it aptly describes how ID shifts and rearranges the borders of reason to exclude both historical and scientific evidence. Exposing the bad history within ID propaganda is important because history is easier for laypersons to understand. It does not require an advanced degree to fact-check. Readers can decide for themselves.

In the post about Tesla, Torley spent more time attributing his magical inspiration to God than uncovering the history of inventions made by Tesla’s predecessors. Gerrymandering in this case applies to the sole-sourcing of Tesla’s autobiography and his entertaining eureka story. This is as expected. Torley puts forward a misleading caricature of a Tesla that appears magical, mysterious and not coincidentally, one who attributes his inspiration to God.

How is the ID viewpoint on Tesla historically misleading? Tesla’s AC motor and radio inventions were incremental improvements and insights completely dependent on his experience with the designs and efforts of his predecessors. Changes Tesla made to existing designs of DC generators, which he did while under the employ of Thomas Edison, were incremental changes to an accumulated complex design.

These designs began with Michael Faraday and were incrementally improved by at least seven others over the span of decades. Conceptualizing the changes to make a motor run on AC was solving a small problem of mechanical and electrical redesign. There were already AC generators, as well as DC motors and generators, and the latter two were well known to be reversible systems. A non-incremental complex design by Tesla would have been the design of an AC motor without any prior motor or generator designs, or theory as precedent. That would have been magical and mysterious, but that is not what happened.

Tesla’s contributions to radio were also incremental and replaced by better ideas soon after. The critical thinker can find all the relevant information on Wikipedia and Google Patents to see that a body of precedent designs and training led Tesla’s mind in the direction of invention. Precedent and existing information, far more than divine inspiration, was the key to his success.

Prior trial and error led to electromagnetic theory and physical laws. These, in turn, allow men like Tesla to make predictions and not repeat the trial and error process. The application of ideas is incremental and requires nothing more than disciplined study and practice. If you apply some academic rigor to account for the design history of the first AC motor, or the first voice radio, you get a list of contributors. But the list is even longer than you might think.

Why the Factory Matters

Replication matters because it is the entire process that must be compared. To bring the human design process in parallel to biology, the two must be laid bare, side by side and compared at every stage. One cannot limit the design discussion to only include the invention or idea formation itself. That is not sufficient. One must trace the generations back in history in both realms, from concept to complexity, without the convenient exclusion of evidence.

How do we account for the contribution of human minds to the design of the AC motors or radios you see and use today? The mass production of serial copies of these products is based not only on the mind of Tesla, but the minds of all the engineers and contributors that made the factories to produce these devices. The factories are far more complex. An enormous number of patents describe machines inside factories – ones that we never see. These unseen machines manufacture parts and assemble them into the finished product. So Tesla is not alone. If all who contributed ideas to manufacture AC motors were alive today and assembled in one place, Tesla would be lost in a crowd.

To compare human design to biological complexity, the factory must be considered as the ribosome, transcription, and DNA replication complexes are themselves factories that manufacture biological polymers like proteins, RNA and DNA. Without exception, mass-production and factory design requires the contribution of many individual human minds, each one contributing incremental improvements. Complexity in factory design grows by generational iterations. The best examples of which are from the early glass factories of Owens-Illinois Corning. History shows these factories started from simple ideas like the bicycle pump and the Gatling gun, and grew in complexity. These are examples you won't find in Wikipedia, but are ones I have researched for a book I am writing on the topic.

Touting the solo Tesla story to support the ID analogy gerrymanders away the necessary factory analogy and has an ugly side effect. It misinforms students that that only the chosen few can invent.  Intelligent Design is religion, disguised as science, and identified as such by excluding both research practice and the accumulation of evidence. It is harmful to students and does not belong in science or history class.

Finally, it may be useful for the Uncommon Descent readers to note that Vincent Torley's Ph.D. thesis in Philosophy is extensively based on evolution, complete with sample phylogenetic tress and information from the fossil record. His thesis mentions neither Intelligent Design or Creationism. A few quotes from his thesis:
Prescott considers the appearance of the platyhelminthes in the fossil record (565 to 544 million years ago) to be the next major breakthrough in the evolution of action selection, after the evolution of cnidaria. He cites research by Raup and Seilacher (1969, cited in Prescott, 2007, pp. 9 - 12) showing that trace fossils of meandering foraging trails left by the earliest animals possessing bilateral symmetry can be generated by combining four simple behaviour mechanisms, one of which functions as a centralised conflict-preventing mechanism, of vital importance to an organism with a primitive brain and bilateral symmetry.

Nerve cells are only found in animals. In fact, they are unique to so-called "true" animals (the subkingdom Eumetazoa awhich excludes sponges). The simplest of these "true" animals are the Cnidaria - commonly known as coelenterates, including animals such as jellyfish, sea anemones, corals and freshwater hydra, which possess the most rudimentary nervous systems found in nature. Fossil evidence indicates that cnidaria were present in the Ediacaran period, 635 to 542 million years ago (Prescott, 2007, pp. 3 - 4).

Conclusion 6.2: The appearance of multicellular organisms, a primitive nervous system and a centralised nervous system represent important milestones in the history of action selection. However, none of these milestones entails a capacity for internally generated flexible behaviour, which seems to have arisen later in evolutionary history.

So as regards Torley's question to me regarding the Cambrian explosion, I leave the exercise to him to refer to his own Ph.D. thesis for the answer.

Thursday, January 26, 2012

What Would Woz Do? Inside the Mind of a Designer.

Author Steven Berlin Johnson has stated quite plainly that an idea is a network of neurons (1). These networks can today be crudely but dynamically visualized with advanced brain scanning. Within this network of neurons, transient connections bring together the fuzzy memories of designs we carry.  When we connect  existing designs together, complexity grows from simplicity. Humans encourage their offspring to copy the eating behavior of their tribal family group, and thereby learn what things are edible and safe, by example. These human faculties, both connection-making and copying, contribute to our survival and place as the dominant species on the planet. 

Wozniak’s Long Hunch

Johnson also talks about the “the long hunch” in the mind of the innovator. Innovations in human design are limited by the network of memories we can juggle in our minds. Human memory is notoriously fickle and transient. Even upon making a new connection, vague memories must be recalled into working memory, facts double checked by reading, observing, or testing, and research must be done in order to fill in the gaps. It is a slow process. 

A clear account of a long hunch story can be found in Steve Wozniak’s autobiography entitled "iWoz. From Computer Geek to Cult Icon: How I Invented the Personal Computer, Co-Founded Apple, and Had Fun Doing It." Wozniak tells how he progressively acquired knowledge about technology, starting with an electronics kit in fourth grade. I can spot more than 30 separate complex technologies that Wozniak had to learn and master before he was able to contribute the design of his first production computer, the Apple 1. He tells of at least 12 complex electronic devices he made that had prototype elements of circuitry and ideas that he would eventually integrate into the Apple 1. Some of these devices were copies of others, like his hardware-based Pong-style video game with on-screen obscenities to amuse his friends.  This game helped him learn how to draw letters and dots on a cathode ray tube with digital circuitry. Some of Wozniak’s prototypes connected existing ideas, like a TV and a typewriter keyboard wired together with circuitry to make a cheap teletype terminal. The copy from an original Apple 1 computer advertisement (2), speaks directly to the contribution of this prototype:  

You Don’t Need an Expensive Teletype. Using the built-in video terminal and keyboard interface you avoid all the expense, noise and maintenance associated with a teletype. And the Apple video terminal is six times faster than a teletype, which means more throughput and less waiting.”  

By connecting a microprocessor and RAM into his home-made teletype, Wozniak created the Apple 1. It was the first integration of video, keyboard, microprocessor, ROM boot code and expandable dynamic RAM, and it ignited the personal computer revolution.  In Wozniak’s words: “People who saw my computer could take one look at it and see the future. And it was a one-way door. Once you went through it, you could never go back.” Wozniak’s design spread through the computer world because he promoted it with free design schematics, software listings and demonstrations at the Homebrew Computing Club. Elements of the design were copied and adapted many times, including the successful computers like the Commodore 64, the Apple //, the Macintosh and the IBM PC.

Wozniak’s study of electronics and his continued prototyping of devices filled his own neural networks with the information he drew upon for the Apple 1, recognizing that he could build a computer inside a cheap teletype terminal by including within it a microprocessor and RAM and ROM code. He proceeded slowly to its realization. I recommend Wozniak’s book for the complete story. You can go further and make your own Apple 1 replica if you want a more detailed view of what it took to build the first modern personal computer (2). 

The Ancestor's Typewriter

Now on this same topic, Denyse O’Leary gives us a nano-lesson on the nature of human design in her book “By Design or by Chance? The Growing Controversy on the Origins of Life in the Universe”. As regards the complex nature of human design it is astonishingly thin in detail, as O’Leary writes just this about the computer: “In the same way, your computer did not evolve from a typewriter by a long, slow series of steps. Most of the steps that separate your computer from a typewriter were the product of intelligent design.” O’Leary floats this little analogy about human design without relevant historical  bibliography. Her footnote suggests that there may be some lessons for Intelligent Design in the retention of the QWERTY keyboard. I submit that the pursuit of O’Leary’s footnoted “lessons” from human design history are things her ID colleagues should have taken seriously as a research topic, but as yet have not. 

Let me correct O’Leary’s intellectually lazy analogy as follows. The human design process that led to the Apple 1 computer did in fact involve a long slow series of steps and there is an early identifiable step that utilizes a manual typewriter. I need not comment on the obvious fact that the typewriter is not of a self-replicating variety. My argument is that the human design process is slow and that complexity grows in small incremental steps. If human design is to be foisted as an analogy to some kind of unseen deistic design, I think we should not allow the example to be historically misrepresented by faint mention.  

Human design proceeds in small steps perhaps because the human mind finds connections only rarely, and must consider whether the connection is a waste of time, to be culled, or whether the connection is something of value. The human mind must speculate about how selection agents will react to the design, and bet on an outcome by taking action to gather the materials and parts and develop the prototype. This all takes time. We see Wozniak repeating this process throughout his life.  

Figure 1. Type Writing Machine, 1898. Thomas Oliver. U.S. Patent 599,863.

So what about the manual typewriter? Here in Figure 1, I show the drawing from the U.S. Patent office of the Oliver Typewriter of 1898 patented by Thomas Oliver  (U.S. Patent No. 599,863, and subsequently refined in further patents of the period (e.g. Patents 693,033 from 1902 and 837,611 from 1906) to the Oliver Model 5 typewriter. And in Figure 2, I show the drawing from the U.S. Patent office of the first electronic keyboard of 1909 built by Charles L. Krum from an Oliver Model 5 manual typewriter and granted a U.S. patent in 1915, No. 1,137,146. This "Printing Telegraph Apparatus" is built from the very same Oliver typewriter model, elaborated with solenoids and signal wiring added to the undercarriage. The wiring designs are derived from the telegraph key, and applied in copies to each key on the manual typewriter. This device is noted in the written history of the teletype as the first working prototype, the first keyboard device which signaled key-presses over electric current to another such device, which printed the character pressed. 

Figure 2. Printing Telegraph Apparatus 1909, Charles L. Krum, U.S. Patent 1,137,146.

It led to a long and slow line of innovations in the teletype, which was made extinct in turn by microprocessor-based personal computers. Wozniak himself adapted an ASCII encoded keyboard from a 1970s electric typewriter, which was also a long-modified design improvement over a manual typewriter, and much simpler in mechanical terms.  

There are two signatures of the typewriter and the teletype in the Apple 1 computer.  The QWERTY keyboard and the ASCII code for teletype communication. The former is visible on the Oliver mechanical typewriter keys. This Oliver typewriter is the mechanical ancestor of the modern personal computer keyboard. It has undergone over a hundred years of design changes in its adaptation to today’s use. Legions of different individual human designers contributed to this effort.

Now Wozniak did not bother redesigning the QWERTY keyboard and ASCII code. He employed their existing designs. But both the QWERTY layout and ASCII have been subsequently re-designed. Unicode intelligently expanded the characters of the English dominant ASCII code to represent characters from many languages. The Dvorak keyboard layout intelligently reorganized the older QWERTY keyboard layout for optimal typing speed. 

But note that QWERTY was also an effort to improve typing speed. Between 1867-1878, C. Latham Sholes used trial and error to slowly modify the keyboard layout in his typewriter to minimize type-head jamming, leading up to his U.S. Patent 79,868. Minimizing mechanical jams sped up the typists of the day considerably, but these constraints are now gone and the Dvorak keyboard shows superior typing speed. 

What of the outcome of these ASCII and QWERTY re-designs? One is a success. Unicode has replaced ASCII allowing the representation of all the characters of the world’s languages. The other is a failure. The statistically optimized Dvorak keyboard did not win enough demand from consumers to become commonplace. Typists resisted re-learning how to type, despite the advantages in typing speed that switching from QWERTY to Dvorak might offer. So we live with the QWERTY layout, a suboptimal but working design that is now frozen in time on every computer. Not a superior design, but one trapped in place by consumer selection and laziness. 

Now just because the signature of the original typewriter is trapped in the design of the personal computer does not mean that it originated from a different human design process. It is rather the long accumulation of human typewriter designs by the same slow and incremental process, which you can uncover by looking up the timeline of typewriter patents as I have. If you understand the process, you would not be surprised to see signatures of earlier designs and inventions nested within a complex object.

Designer Myths

Do human designers have some kind of special intelligence that other humans do not possess? Their intelligence may have more working memory, or they may be better at reading comprehension, or their 3D spatial skills may be better. But there is nothing outside the normal human intelligence required. It is not mystical. Wozniak was a voracious consumer of information about technology, and to a great extent he trained himself to be a genius in electronics technology. Certainly he is a genius, but there is no evidence that genius level intelligence is not itself composed of the same mental capacity we all possess. Genius is an expected outlier, an occasional observation within the distribution of normal human intelligence. Just as some people are tall, or have large breasts, it requires no supernatural explanation. Not all humans engage in design, so it is easy to be led to think that human designs are spontaneous, magical creations of intricate complexity, and invented in isolation by a single mind. They never are. Our human efforts at design follows the gradual timeline of our species, accelerating in complexity only after what I would argue is a continuing industrial revolution. 

A recent article in MIT Technology Review by David Rotman entitled “Can We Build Tomorrow’s Breakthroughs?” (3) discusses the invention of new kinds of batteries suggests that human design is now such a complex process, that innovation can only take place in the context of the factory that itself produces batteries. Micheal Idelchik of GE is quoted by Rotman in the article as saying: “You can design anything you want, but if no one can manufacture it, who cares?” The argument is that a lone inventor working in a garage, tinkering on battery design has insufficient information about the production of batteries to make a useful contribution. So the hidden complexity of assembly lines and factories further masks the process to the observer of human design. Yet behind closed doors the innovation process remains the same, the orchestration of Change, Prototype, and Production agents, as I described in my previous essay. The memory of company designs is resident in CAD data files and information systems which are hidden to us. Serial copies of new designs exit the factory only by the mercy and whim of internal Selection agents. 

We may think we see remarkable changes in human designed products. But this is an illusion, and the reality of many small steps and prototypes is hidden inside the company. The story of any object lies in its complete history, and the only intentional design process we can study with historical facts is that of the human design process. We have no examples of intentional deistic or anthropomorphic-alien designed objects and of course no accompanying history. To understand the intentional design process we must study and understand those things that humans have created. When you look carefully you will find that complexity in human design often requires many minds to achieve.


This tight coupling of Change, Prototyping and Production in large teams may have left the individual tinkerer in the dust. But Steve Wozniak reminds us that the individual can still be a successful human designer. Let me close this essay with a few words about how to train your mind for human design.

Prepare for Design

For any form of creative design work, one must take considerable time to build into one’s memory this kind of neural network of information that Steven Johnson describes. Following Wozniak, this process starts from childhood. If you wish to become a great human designer, your challenge is this. You must spend an extraordinary amount of time cultivating your own mind and build your own neural network. You must read, understand and critique a great variety of existing designs, and get your hands dirty with the materials and technologies.


You must understand the tools to develop prototypes, and indeed go further to understand the assembly line processes required to produce serial copies of your design. You must make prototypes as they will feed back and reinforce your understanding about the relation between form and function. Successful human designs must become part of your knowledge. Learn broadly and seed the neural network in your mind with a selection of old and new inventions. Critical thinking is a requirement for this discipline of design, and is not to be avoided. You cannot be intellectually lazy and you cannot submit yourself to the deluded and time-wasting arguments of lazy thinkers. Instead, get your information from original sources.


To polish your neural network of information, it is important to pay attention to opportunities for design connectivity. To do this, make an effort to identify and recall the interface points of each design you encounter. What is an interface point? Steve Wozniak’s 1970’s TV had no video input port on the back. So he opened up the case and used a probe to find the location of the video signal on the circuit board. Then he could connect it to his video circuits. This is precisely what I mean by an interface point. Sometimes it is obviously designed for the purpose, like a video cable connector. Sometimes it is hidden, a signal inside, from which you have to build an interface.  An interface point is that location or opportunity on an existing design to connect to or to construct a connection to another design. Interface points can be mechanical parts, electronic signals, or software functions to which connections can be made. Biosensors often take advantage of chemicals produced by enzymes as interface points for sensor electronics. Chemists recognize certain chemical substructures as interface points for the synthetic reactions required to build complex pharmaceutical molecules.


Copy and Connect to Create Complexity


In all these cases, complexity arises from recognizing the opportunities in connectivity. To best train your mind for design, focus on the identification of these interface points as you learn about general science and technology. Consider it a mental game to find the interface points that are hidden or non-obvious, and think actively about the connectivity opportunities as did Wozniak. Learn by making playful copies of existing designs. Be frugal and eliminate unnecessary components. Make things. And find joy in each of your successes.


Notes:
1. Steven Johnson’s TED talk:
http://www.ted.com/talks/steven_johnson_where_good_ideas_come_from.html
2. Apple 1 Computer Information:
http://www.applefritter.com
3. Can We Build Tomorrow’s Breakthroughs? David Rotman, MIT Technology Review, 2012.
http://www.technologyreview.com/article/39311/