Projects to find the limits of what is possible right now?

[Nominal Animal]

Have you ever done an electronics project just to see if it is possible to do that specific thing better than anything already existing on the market?  What was it?  For what definition and value of "better", other than cheaper/smaller BOM/easier to manufacture?


In the programming world, one of my interests is finding out ways to speed up certain types of simulations (high-performance computing stuff).  It is not a commercially viable approach, because it is much easier to spend $10x in hardware than $x in wages.  However, there is a fraction of programmers that do this constantly, just the same.  See git as an example.  It was not a case of NIH as some believe, but more borne out of frustration of lack of proper tools that could handle the Linux kernel code base distributed over dozens or hundreds of developers.

(I don't mean NIH is not a thing with us code monkeys, because it is.  This is, however, unrelated to NIH: this is about doing stuff better than before, for some definition of better that is not related to commercial value.)

Recently, I asked a question here about a similar situation with USB wall warts and LED lights. Essentially, I was wondering what is currently technically possible and feasible; ignoring what makes sense commercially.  I noticed that all answers basically seem to hint that a thing makes sense if and only if it makes commercial sense, and that if a thing does not make commercial sense, there is no reason to think about or discuss such thing.  (At least that is the way I read the answers; but do note me fail English often.)  I am wondering whether that is a prevalent way of thinking in the electronics world, or whether there are designers and hobbyists that do like to push the envelope just to see how far it stretches, ignoring any considerations for commercial viability.

Don't get me wrong: it is not a bad thing, to consider only viable stuff.  It is a very rational way of looking at things.  I'm just asking, in this particular question, who and how electronics designers and hobbyists "push the envelope", so to speak.

Note that the "maker" world is a different thing altogether, because they are trying to solve some problem, or achieve some specific purpose, and not just to see if the thing can be done better.  (You could say that git is then not a very good example. True. It was just the closest thing I could come up with, other than my own experiments and computational schemes for molecular dynamics simulations. I could show the way I devised for MCMC simulations using Metropolis-Hastings algorithm, allowing efficient parallelization and simultaneous communication and computation, without compromising overall detailed balance; but it's extremely dull for anyone not particularly interested in that, and even the simplest version uses 21/64 sub-steps for each cycle with a sliding 3D window to the data, with detailed balance retained over full cycles only.  It's too deep into software engineering to fit into computational papers, and too much about a specific data structure and use case to interest any computer scientists.  So there is no hope for even publishing it.  Yet, it beats the shit out of current widely-used simulators in efficiency and speed in comparison.  But, because the code is about two orders of magnitude more complex, it is not really viable even in the simulation world.  As an example, I have still not proven, mathematically, that detailed balance is truly retained. I can only show it numerically, in the statistical sense. So, objectively speaking, it's just me waving my hands around vigorously.)

[David Hess]

I have designed and prototyped low noise DC amplifiers which were comparable to the best that you can find for sale.  Noise was so low that low values of resistance could be measured from the Johnson noise of the resistance.  What was more interesting was verifying this level of performance with basic bench instruments.

I have designed some non-traditional audio power amplifiers which were better than my ability to measure but honestly that was not saying much.

I once replaced the voltage regulator on my car with a custom designed unit which regulated the battery voltage to a change of less than 10 microvolts over a 20 amp load change.

The untuned field strength receiver shown below combines a single stage wide range AGC circuit with a temperature compensated square law detector to operate over an incredible range of signal levels from ambient RF to 10s of watts.

[rstofer]

Essentially, I was wondering what is currently technically possible and feasible; ignoring what makes sense commercially.  I noticed that all answers basically seem to hint that a thing makes sense if and only if it makes commercial sense, and that if a thing does not make commercial sense, there is no reason to think about or discuss such thing.  (At least that is the way I read the answers; but do note me fail English often.)  I am wondering whether that is a prevalent way of thinking in the electronics world, or whether there are designers and hobbyists that do like to push the envelope just to see how far it stretches, ignoring any considerations for commercial viability.

Basically, if a thing is worth doing, it is worth doing for money.  Sure, I'm up for that!

Projects that tweak intellectual curiosity are also important but they don't pay the rent so maybe they get pushed to the back burner.

Then too, it depends on where you work - if you work for Bell Labs, you can do pretty much anything you want!  Nine Nobel Prizes kind of gives a hint of their scope.  I suspect the janitors have Master's Degrees and a desk job probably requires a PhD.

https://en.wikipedia.org/wiki/Bell_Labs

There are many jobs that pay you to play with stuff.  Or you can do it on your own time, with your own equipment, and see what happens.  The problem, of course, is that you might not have cutting edge tools.

Universities are a good choice but the fun stuff is usually left to the professors and their prodigies.  It would be an exceptional outsider to get into much of the fun stuff.

One area that is wide open is computer science.  The cost of entry is low, the playground immense and the need for improved algorithms is vast.  AI is the new 'flashy thing' and we see applications with self-driving vehicles.  As soon as they get the 'swarm' component working, it may actually be possible to improve traffic flow.  Of course, we will have to wait for the dinosaur cars to wear out...

[tpowell1830]

Just to be clear, I am old, 65 years. I have tried a couple of times to do what you are suggesting when I was young, can't remember exactly what the projects were now. However, the thing that kept me from completing those projects was the fact that when I get home from working an engineering job all day, the last thing that I want to do is spend a lot more time in my head trying to solve some serious equations or inventing new ways of doing things.

Like many here, doing engineering work for a company is very taxing because there are some serious thought processes that have to occur in order to do what I do for a living. I find myself thinking about doing things that you are talking about occasionally, but I find the wear and tear of everyday problems at work to be too much, and all I want to do in my spare time is watch some educational videos from time to time and tinker with electronics that I am curious about occasionally, outside the infrequent repair that sometimes pop up.

The type of grind that I am now seeing in my job is typical of my entire career (although it seemed much worse when I was a nube back in the Paleolithic period, sooo much to learn). I have often wondered if other engineers had the same outlook as I do, however, I never asked.

[Electro Detective]

MOST real hackers, hobbyists, tweakers and backyard/farm cobblers do not concern themselves with petty "commercial interests" (GREEDtardery) when they are busy improving on a design or tool or software/firmware etc
be it out of user frustration with an item, 
or just sensing/knowing it can be done better with a bit of head scratching, elbow grease
and perhaps a bit of trial and error destructive testing thrown in along the way   

I've improved and or re-invented half baked stuff over the years that many might appreciate, but being a realist I know that I don't have the right commercial interests 'connections' 

i.e. unless you have those 'connections' it's a waste of time and money trying to get anywhere,
and if the 'better' product interferes with a marketed inferior product, which is usually the case, the commercial interests won't be rolling out any welcome mats for yours... 

That said, if it wasn't for many good spirited people 'having a go' at doing something 'better' over the last few centuries, 

the so called "commercial interests" of today would be busking on the streets or flogging favours in back alleys for a feed 

MOST "commercial interests" only know how to exploit no matter what their clever apologetic jargon and spin BS is,
they are parasites that feed off financially clueless inventors and workers, and not doing the general population any favours

Good useful products sell themselves, they don't need suspect "commercial interests" playing go-between and taking an unfair cut of the profits and stifling R+D
'just because they can...' 

[sundersoft]

In the programming world, one of my interests is finding out ways to speed up certain types of simulations (high-performance computing stuff).  It is not a commercially viable approach, because it is much easier to spend $10x in hardware than $x in wages.

The commercial vendors seem to invest quite a bit of effort into speeding up their algorithms (e.g. for electromagnetics; not sure about molecular dynamics). If you sell a product as a finished piece of software or as middleware with an easy to use API then no one will care how complicated the implementation is.

It was just the closest thing I could come up with, other than my own experiments and computational schemes for molecular dynamics simulations. I could show the way I devised for MCMC simulations using Metropolis-Hastings algorithm, allowing efficient parallelization and simultaneous communication and computation, without compromising overall detailed balance; but it's extremely dull for anyone not particularly interested in that, and even the simplest version uses 21/64 sub-steps for each cycle with a sliding 3D window to the data, with detailed balance retained over full cycles only.  It's too deep into software engineering to fit into computational papers, and too much about a specific data structure and use case to interest any computer scientists.  So there is no hope for even publishing it.  Yet, it beats the shit out of current widely-used simulators in efficiency and speed in comparison.  But, because the code is about two orders of magnitude more complex, it is not really viable even in the simulation world.  As an example, I have still not proven, mathematically, that detailed balance is truly retained. I can only show it numerically, in the statistical sense. So, objectively speaking, it's just me waving my hands around vigorously.)

If you can create an implementation for a specific use case that people outside of academia are willing to pay for, then it might be commercially viable and having more complexity is better anyway because it will prevent competitors from copying your software. If you published a paper or convinced someone else to implement your algorithm then you wouldn't make any money anyway.

[Nominal Animal]

Then too, it depends on where you work - if you work for Bell Labs, you can do pretty much anything you want!

I just flew off at the handle at a "HPC consultant" for suggesting that using async I/O with MPI is somehow bad or risky (deadlocks my ass) at StackOverflow.  (Sure, not doing communication while you do computation is good for not overheating the CPUs, and really good if you want to ensure every cluster has an InfiniBand network.. but in software engineering terms, that sort of argument is akin to claiming that shit tastes good, because this paper here undersigned by so many Big Names says so.)

My understanding is that slick people with good people skills and reliable, productive but not very creative development skills are what most employers are seeking.  They want profits, and have reasonable expectations of profit, and that excludes purely exploratory investigations into what is possible.  I think the only true exploratory think tanks left are ... well, in someones garage, after they've won the proverbial lottery, and not have to worry about money or publishing, and just do what the heck they want.

Universities are a good choice but the fun stuff is usually left to the professors and their prodigies.

I'm still in an University that does HPC on their own clusters, but at least here, it looks like development and maintenance is being outsourced.  When the number of articles is your primary measurement, you know your operations will tend to optimize that to the detriment of any other consideration.  When your only tool is a hammer...

One area that is wide open is computer science.

Exactly.  Now that it is not that rare for a hobbyist to have a pretty darned full-fledged electronics lab, I'd expect there would be a people with the skills, equipment, and interest to do that sort of stuff in electronics, too.

I've improved and or re-invented half baked stuff over the years that many might appreciate

Why not talk about them, here?

I mean, if anyone were actually interested in some of those schemes I've developed (that are too complex to implement in a real, maintained simulation; but are a significant step in computation speed wrt. real world time taken), I'd be happy to talk about it.  It's just that in my case, it's so niche nobody is interested.  Here, with respect to electronics projects, I am asking and interested!

The commercial vendors seem to invest quite a bit of effort into speeding up their algorithms (e.g. for electromagnetics; not sure about molecular dynamics).

The things I've made most headway with are at the core of what e.g. LAMMPS and Gromacs do.  I think that if I were given a year, I could implement the core and at least a few different potential models in a form that would be useful.
However, even then, if you look at who the sponsors of those two pieces of software are, I'd have a vertical wall ahead of me trying to explain why my simulator would be better.  Even a 2x to 5x real-world speed increase means nothing, when the "competition" is well cited, well funded, and commonly accepted as a method of obtaining numerical results for articles in even the most respected scientific journals.

I mean, I know for a fact that both of those are actually very inefficient, and we can do much better.  But it is the same situation as with USB wall warts and LED lights: they are perfectly good enough as is, so there is zero demand for anything better.  I personally am just much more interested in doing better, and making even better tools, even if not needed or commercially viable. My mind is b0rked that way.

[T3sl4co1l]

This current-limiting electronic fuse, which can dissipate 600W for 150ms, while drawing a mere 300uA from a 9V battery, giving months of battery life:



In fact it's self-powered during a limiting event, but that's such a small duty cycle that it doesn't really matter.

To realize the same solution with off-the-shelf ICs, would require more than twice the supply current.  No one makes low current gate driver ICs!

If its basic design were integrated, it could probably be 1/3 to 1/10th the current consumption still, but of course I don't have that as an option.

Tim

[sundersoft]

The commercial vendors seem to invest quite a bit of effort into speeding up their algorithms (e.g. for electromagnetics; not sure about molecular dynamics).

The things I've made most headway with are at the core of what e.g. LAMMPS and Gromacs do.  I think that if I were given a year, I could implement the core and at least a few different potential models in a form that would be useful.
However, even then, if you look at who the sponsors of those two pieces of software are, I'd have a vertical wall ahead of me trying to explain why my simulator would be better.  Even a 2x to 5x real-world speed increase means nothing, when the "competition" is well cited, well funded, and commonly accepted as a method of obtaining numerical results for articles in even the most respected scientific journals.

I mean, I know for a fact that both of those are actually very inefficient, and we can do much better.  But it is the same situation as with USB wall warts and LED lights: they are perfectly good enough as is, so there is zero demand for anything better.  I personally am just much more interested in doing better, and making even better tools, even if not needed or commercially viable. My mind is b0rked that way.

There are probably some types of simulations that aren't tractable currently and might be with improved algorithms (although it would probably require a 100x speedup or a more along with good scalability on a GPU cluster). There might be some commercial applications in automatically finding drug candidates that target a particular active site of a protein and automatically checking for side effects, for example, but that would probably require very high speedups over existing implementations to be feasible.

[Nominal Animal]

There are probably some types of simulations that aren't tractable currently and might be with improved algorithms (although it would probably require a 100x speedup or a more along with good scalability on a GPU cluster).

I suspect that there is an underlying perceptual barrier between current high-performance computing algorithms and the operation of the underlying hardware.  Essentially, the paradigm is that your code operates on data, with all of the data accessible all at once.  And that is just not how computers work.  Data flows.  You have caches and bottlenecks, and the time taken to access a specific piece of data depends on a large number of factors.  Those factors cannot be eliminated (although things like caches and InfiniBand networks were developed to make them as small as possible).

Those using both GPUs and CPUs for computation have a much better grasp of this, because the transfer of data between system RAM and dedicated GPU RAM is a bottleneck they are well aware of.  My suspicion can be simplified into something like "to get the most out of the system, we must understand those bottlenecks, and ensure we balance things so we never or only rarely hit any particular bottleneck, even when it is hard to work with such complex designs".  It is not a popular view among programmers, because doing stuff that way is hard, and a lot of programmers cannot handle the complexity.  It's just too hard to do right.

In a very real way, some of the old games were pretty good examples of how to push the envelope.  One excellent example is Starflight, from 1986.  It fits in 128k of RAM.  Yet, the game is huge, with a real storyline.

I do not mean to belittle current games, by the way. They are just focused on a completely different envelope, and push the boundaries of that instead. A good example is the recently published game Vane; I watched the walkthrough video on Youtube.  It is odd, definitely entertaining, and some of the things that the game engine must do, are technically very interesting.  But, overall, it is not the technological envelope that it pushes against.

What I haven't seen except in this thread, and occasionally by mention (especially as things in the past, being way ahead of their time in their capabilities), is the electronics equivalents.  I am pretty sure the old greybeards do talk about that sort of projects amongst themselves, and would love to hear more of those.


Perhaps a bit of reasoning why this is so interesting is in order.

Consider the technology in the early 1960s.  Who knew what that tech level was actually capable of?  I see the manned Moon landings as very much similar push outside the envelope, albeit with all sorts of undercurrents and contingency plans at the highest levels that are not exactly clear to me.  But if we rule out the politicians and career bureaucrats, to the rest of the people involved in it, I believe that it was very much an exercise in seeing if they could do it.

Where is that sort of curiosity and inquisitiveness and creativity today?  I do see hobbyists doing it in programming, and makers building all sorts of gadgets for fun and to solve existing problems.  Is there anyone trying to see what our electronics are really capable of now?  To see where the boundaries really are, instead of living by the limits dictated by commercial or practical viability?

[cdev]

I think one of the more interesting projects is trying to improve how we function cognitively, in a safe, legal, fun, food based manner.  I've been into that for a long time. It does seem to work in certain domains. Others, not so much, and we do get older.

But all in all I wouldn't do it any other way.

[CatalinaWOW]

Different folks have different motivations.  I find little interest in pushing a capability to the limit just for bragging rights or whatever.  I do like making something that was hard easy, or even just possible.  But once that has been done it is hard to motivate myself to go further. 

In your archetype example, I couldn't for the life of me motivate myself to invent a better simulation compute engine or element.  UNLESS it was a step towards doing a simulation I wanted to do, that couldn't meaningfully be done with current technologies.  That interest could push me to try.  Of course trying is not guarantee of success.

Other examples of the difference.  Your wall wart example.  Existing wall warts meet all my needs so I can't get motivated to work on them to make them better by any metric - cost, size, regulation, current capacity, adjustability, safety or anything else I can think of.  There are design challenges in all of those directions, but for me, electronics and software are a means to an end, not the purpose of my intellectual activity. 

[Nominal Animal]

I think one of the more interesting projects is trying to improve how we function cognitively, in a safe, legal, fun, food based manner.

Cognition is one of those things that change as you age, no matter what you do.  It is an interesting challenge to steer the changes in an useful direction.  Food -- and not just its composition, but also the intestinal bacteria it best supports and the calorific intake patterns -- is definitely one way to do it.  I am lazy on that front, so I tend to use puzzles and abstract intellectual challenges myself.

I find little interest in pushing a capability to the limit just for bragging rights or whatever.

It is very much the "whatever" part, by the way.   I am absolutely not interested in being popular or in bragging rights; it's not about that.  It is about being genuinely interested in where the limits of what is possible are.

This probably has to do with personality types.  I've always been an analytical crazy scientist type; there are family stories about that dating back to preschool age.  I often spend time creating better tools just because even minor deficiencies in existing tools annoy me when I know how those deficiencies can be fixed/overcome/avoided.  I have had to learn the engineering approach the hard way; making things good enough, but not waste resources overdoing it.  That is still very hard for me, as I'd much rather make things as good as I can.  Not because of bragging rights or anything like that, it is more an innate need of some sort.  Could be a variant of OCD.

[cdev]

Probiotics and prebiotics (fiber like inulin) definitely do help in that regard, as do phytonutrients like berberine. Two that are particularly good for a healthy gut are trimethylglycine. (also called betaine) and its close relative choline. Choline is very good for both the brain and your little brain in your gut. Also, cold water fish oil is too. The brain contains a lot of DHA, an n-3 fatty acid found in large amounts in ocean fish and it needs lots of it when its growing, also later in life for its integrity. (Also eating lots of fish oil is very good for the heart see https://academic.oup.com/ajcn/article/83/6/1477S/4633237 )

Infants and children as well as people at risk for heart disease and older people in particular should get a lot of Omega 3s. For vegetarians, Flaxseed oil is a good nutritional source of Omega 3s as is alpha lineolic acid. However they may not be as good for the heart as is fish oil.

Two more nutrients that are helpful as we age are Aceytyl-l-Carnitine and alpha lipoic acid. They help keep your mitochondria in good shape.

Cognition is one of those things that change as you age, no matter what you do.

Thats true but we can influence this much more than many think.

It is an interesting challenge to steer the changes in an useful direction.  Food -- and not just its composition, but also the intestinal bacteria it best supports and the calorific intake patterns -- is definitely one way to do it.  I am lazy on that front, so I tend to use puzzles and abstract intellectual challenges myself.

  This is really true and taking probiotics like yogurt and prebiotics like fiber both are really good for your health as you age.  Acetyl-l-carnitine in particular is also really good for diseases that impact the gut as well as the brain.

[coppercone2]

most process controllers you can do better then what is available, like most things with feedback loops including amplifiers. Filters too most likely.

The biggest problem for most people IMO with this kind of stuff is the damn expensive laser trimmed dividers, HV dividers.

[Nominal Animal]

most process controllers you can do better then what is available

Any anecdotal stories you could share?  (I have a feeling that perhaps you or your colleagues have ditched some process controller because it really wasn't that good, and replaced it with something homebrew that worked much better.)

[coppercone2]

Ypu wont have the same level of shielding, immunity, mechanical robustness stability or precision with mass produced products because saving five cents on some extra decoupling caps makes sense for a designer but if you even think about doing that for a single unit you are wasting your time.

Or a nice sealed welded steel chassis with all the shielding dodads and shielded lcd etc. No one wants to pay for the bythebook solution

Then when you do it right the fun begins with transducer hysterisis and aging and shit being the dominant effects, then your interest shifts to custom sensor design that can challange your electronics.

In general if you dont know what you stand to get rid of your design will age better or be more stable with age or fail more gracefully.

What you wont get from yourself is good software though. At least mine never was.

And obviously i cant give examples on this forum lol

Its obvious if you want to make money you find the minimum working solution to a spec people will buy. Why would a company make a process control with capabilities there is no process for? Usually they take such suggestions as gambling with their money and trying to bait scientists. Everyone finds a market so they can make a sellable spec.

Or the research applications for such a device are absurd and useless. Mad scientist shit.


I am not saying their bad or dont work but their not exactly trying to paint the mona lisa. Its never really gonna be that good.

But good luck trying to beat 8.5 digit meter . How to defeat this behemuth? Or microwave stuff. I can see someone with enough machinery getting extreme quality rf plumbing at home though.

[CatalinaWOW]

 I often spend time creating better tools just because even minor deficiencies in existing tools annoy me when I know how those deficiencies can be fixed/overcome/avoided.  I have had to learn the engineering approach the hard way; making things good enough, but not waste resources overdoing it.  That is still very hard for me, as I'd much rather make things as good as I can.  Not because of bragging rights or anything like that, it is more an innate need of some sort.  Could be a variant of OCD.

Deficiencies are in the eye of the beholder.  Is something defective because it can't be operated in the middle of a blast furnace?  Is a socket defective because it cannot apply 1000 N/m of torque, or is it defective because the walls are too thick to fit in tight places?  There is a reason why both impact and normal versions are available.  They have been optimized in different directions.

While you may have invented methods to improve the tools you are using, are you sure you didn't make them less perfect for another user who has a different application or a different approach to the problem?

None of this is intended to belittle your efforts, or to stop you, just pointing out that there are many different itches to scratch out there, and not all of them match yours.  And maybe to direct some of your OCD into a "perfect" understanding of all of the users of a given product and their requirements and desires.  Which is necessary to "perfect" a tool for a market broader than one.

[coppercone2]

Usually people will agree that its done for the sake of being cheap. If its done to be light or small or something its usually noticable. Market cost.

Usually very obvious to another designer.

You see a product and know the gist of how it works you can imagine rather quickly how it would change if optimized for some peculiar reason. If you cycle through the usualy suspects in your head like repairability, robustness, size, weight etc and your  noct seeing a design phillosophy that usually meane someone got greedy or the market is shit.

I.e. its not hard to tell the rammifications of storage, intraoptability or others. Usually its cheap but there are clever designs related to cleaning ease or weight or storage that can be tricky and lumped as cheap but really result in more cost to design.

[CatalinaWOW]

Usually people will agree that its done for the sake of being cheap. If its done to be light or small or something its usually noticable. Market cost.

Usually very obvious to another designer.

You see a product and know the gist of how it works you can imagine rather quickly how it would change if optimized for some peculiar reason. If you cycle through the usualy suspects in your head like repairability, robustness, size, weight etc and your  noct seeing a design phillosophy that usually meane someone got greedy or the market is shit.

I.e. its not hard to tell the rammifications of storage, intraoptability or others. Usually its cheap but there are clever designs related to cleaning ease or weight or storage that can be tricky and lumped as cheap but really result in more cost to design.

Sometimes it is that simple. 

But to give a counterexample, integrated circuits.  They are not repairable and are not designed to be.  Yes, low cost comes along with that.  To extend, if you want repairability and maintainability you would eliminate BGA and other high density packaging techniques.

Or another example - is it good design to make something robust and maintainable in a rapidly changing technology area?  Do you really still want to be using your old brick analog cell phone, or is it fine that it is thrown away without repair after a few years?  Yes that does serve the salesmen and such who just care about sales, but it also has allowed the product to develop much more rapidly than it would if new products had to wait until the older generations had finally been repaired too often to make further repairs feasible.

[Nominal Animal]

I often spend time creating better tools just because even minor deficiencies in existing tools annoy me when I know how those deficiencies can be fixed/overcome/avoided

Is something defective because it can't be operated in the middle of a blast furnace?

I think it is, if and only if I suspect I would like to operate it in the middle of a blast furnace, and I know how to make it safe to operate in the middle of a blast furnace.

Please, do not confuse this with NIH.  It also isn't just OCD or perfectionism.  I only perceive defects when the defect limits or negatively impacts me, and I know how to avoid that defect.

While you may have invented methods to improve the tools you are using, are you sure you didn't make them less perfect for another user who has a different application or a different approach to the problem?

The pool of tools is not a limited one.  The tools I create do not "supersede" or replace anything, so your question does not apply.

Elsewhere, I do keep harping about maintainability (readability and modifiability of a tool) and the importance of documentation, and comments describing developer intent in code.  So, if you meant something along the lines of "how do you ensure users are not confused by choice, and know when to choose a tool of your design, rather than something else?", my response is "by documenting the tool well, and noting the important differences when showing the usage information".

just pointing out that there are many different itches to scratch out there, and not all of them match yours.  And maybe to direct some of your OCD into a "perfect" understanding of all of the users of a given product and their requirements and desires.  Which is necessary to "perfect" a tool for a market broader than one.

That assumes humans are rational, and are always willing to adopt a better tool.  They are not.  You see, I have in fact done that sort of workflow optimization and tool customization as well.  It is fun and rewarding, until you get hit by the stupidity and inertia in doing things the old way by typical human beings.

Thing is, humans are surprisingly attached to the ways of performing tasks they have learned to do by rote, and see any change as dangerous.  Job Safety and all.  Often, when they say something is performing poorly, they mean they have read about this new fad and would like you to find them an excuse to partake in that fad.

If it wasn't for that, I'd probably be doing that sort of workflow optimization/problem solving/tool customization stuff for a living.

It might be different in businesses elsewhere in the world, but this is unfortunately the situation in Finland, and I do not want to emigrate.  (As an example, it is accepted as normal for large IT projects to fail without any end product or repercussions to the consultants or providers; so much so that only a third or so of large IT projects actually produce anything.  That is not just with government stuff; it applies just as well in the business world.  It is a very good time to be a CEO of a big IT consulting company, with a small group of flashy consultants to make a good initial impression, a group of technical writers exclusively writing bids, and offshore the actual production to recent graduates in India and Indonesia.  As long as you keep your public image shiny and keep the politicians well lubricated, the money keeps flowing in like crazy.  Pity that I look like a potato, and would rather live as a hermit in the back woods, than do that sort of a job again.)

[Nominal Animal]

another example - is it good design to make something robust and maintainable in a rapidly changing technology area?

On the software side:

Yes, iff you make it modular too.

The idea is to assume change, and that changes will be needed at some point down the line.  Trying to cater for all possible needs up front is a bad idea, though, because guesswork is unreliable.  Instead, you look at the real existing needs, and cater for those, and try to do it in a way that makes it easy to adjust the behaviour later.  Usually that means splitting the task at hand into several different small tools or libraries you use in combination.

Unix philosophy, the KISS principle, and minimalist approach to computing seem to work.

I don't know how any of that translates to electronics, though.

[coppercone2]

imo 'not designing something right because of changing technology' is basically punishing an existing industry.

1) you cant tell the future (so your boss is a gambler)
2) its low quality regardless
3) when you make something nice people use it and it resists change or is incorporated into something else.

I don't believe in it but I am sure it does not fit with your corporate quarterly culture. What you don't see is that dave jones is going to get a heart attack reviewing that garbage produced with this philosophy.


I can see it being done as a custom request for someone saying we are migrating to this process in 2 years and we need a stopgap measure etc. But don't design a 'all arounder' , market it as a all rounder, and then build it for one particular industry. Like, I gotta talk to the sales guy to find out its a piece of shit designed to keep some hack running? 'we built it at reduced cost because our main customer...' (when it breaks on you) yea then just market it to your main customer privately. I should never know about it then and I should not be able to buy it. Otherwise its deceptive sales practice. The fuck do I care about your intended market? i gotta study the entire industry to figure out if you are trying to sell some low quality shit because someone might change something in 2 years because of something you overheard at some trade room toilet stall?

[Nominal Animal]

Well, there is nothing inherently wrong in designing a short-lived gadget that barely does what its sales pitch promises.

There is also nothing inherently wrong in using human hair to make a soy sauce analog, or extracting proteins from human fecal waste for use in food production, as long as the end result is clean and of acceptable quality for human consumption.

[coppercone2]

that might get you killed