> "Trellis coded modulation got this rate up to 50 kilobaud by the 1990s"<p>Not quite, and an interesting story that fits these engineering maxims better than you might think.<p>An analog channel with the bandwidth and SNR characteristics of a landline phone line has (IIRC) a Shannon capacity of 30-something kbit/s, which was closely approached with V.34, which used trellis coded modulation plus basically every other coding and equalization mechanism they knew of at the time to get to 33.6kb/s on a good day.<p>But... by the 80s or so the phone system was only analog for the "last mile" to the home - the rest of the system was digital, sending 8-bit samples (using logarithmic mu-law encoding) at a sampling rate of 8000 samples/s, and if you had a bunch of phone lines coming into a facility you could get those lines delivered over a digital T1 link.<p>Eventually someone realized that if your ISP-side modem directly outputs digital audio, the downstream channel capacity is significantly higher - in theory the limit is probably 64000 bit/s, i.e. the bit rate of the digital link, although V.90 could only achieve about 56000 b/s in theory, and more like 53kb/s in practice. (in particular, the FCC limited the total signal power, which means not all 64000 combinations of bits in a second of audio would be allowable)<p>I worked with modem modulation folks when I was a co-op student in the mid-80s. They had spent their lives thinking about the world in terms of analog channels, and it took some serious out-of-the-box thinking on someone's part to realize that the channel was no longer analog, and that you could take advantage of that.<p>A few years later those same folks all ended up working on cable modems, and it was back to the purely analog world again.
> <i>if your ISP-side modem directly outputs digital audio, the downstream channel capacity is significantly higher</i><p>But why is it higher? It's still an analog channel (the last mile from the ISP to your house), right? Doesn't it get filtered? So isn't it still subject to the Shannon-Nyquist limit?<p>Here's an ASCII drawing of which parts are digital vs analog as I understood your explanation:<p><pre><code> Rest of world<--- digital--->Telco<---digital--->ISPmodem<---analog--->HomeModem
</code></pre>
Suppose you're saying that the link between the ISPmodem and the HomeModem is a bare unfiltered copper wire. In that case, I have a different question: Couldn't you send data at megabits per seconds over a mile long copper wire without using modems at all (using just UARTs?).<p>I hope you can clear up my confusion.
> Couldn't you send data at megabits per seconds over a mile long copper wire<p>Yes, but you need the bare copper wire without signaling. We operated a local ISP in the 90's and did exactly that by ordering so-called "alarm circuits" from the telco (with no dial tone) and placed a copper T1 CSU on each end. We marketed it as "metro T1" and undercut traditional T1 pricing by a huge margin with great success to the surrounding downtown area.
No, it’s more like HomeModem ←A→ Exchange1 ←D→ Exchange2 ←A→ ISPModem. The digital parts were all inside the telco’s networks that connect the exchanges to each other.<p>> Couldn't you send data at megabits per seconds over a mile long copper wire without using modems at all (using just UARTs?).<p>No. The exchange is sampling the analog signal coming in over your phone line at 8kHz and 8 bits per sample. They just designed modems that sent digital data over that analog link, in a way that would line up exactly with the way the exchange will sample it.
It's ISP←A→Telco←D→Telco←A→You<p>Traditionally both the ISP and you pay for analog phone lines from the telco. The telco uses digital internally (remember you and your ISP probably aren't at the same exchange), which puts a hard limit on data rate - there is no trick you can do to get more bits through than the bits used in the digital part of the call.<p>If you (as the ISP) buy enough lines you can get them delivered in digital format. A T1 is designed to carry 24 simultaneous phone calls, acting virtually as a bundle of 24 analog phone cables. So the obvious next stage was to have a modem that can handle 24 simultaneous connections on one cable.<p>Now you have ISP\_modem←Ax24→ISP\_muxer←Dx24→Telco←→Telco←A→User<p>The ISP's modem generates analog signals for up to 24 simultaneous incoming calls, and they pass into a multiplexer that connects 24 analog lines to a T1 line and they go through the telco digitally to users. The maximum bandwidth is still as before - the modem has to generate an analog signal that will still be receivable at the other end after A2D and D2A conversion. Even though the digital bandwidth for the digital part is 56kbps, the maximum achievable bandwidth through this digital-bottlenecked analog call was found to be 33.6kbps.<p>But the industry had an idea: by convincing the telco to install the modems into the user's exchange, the analog portion would only be between the telco and the user, without a digital segment in the middle of it, and therefore wouldn't be bottlenecked the same way. The same digital backhaul from the ISP through the telco was used, but instead of transmitting a digitised analog modem signal and therefore causing degradation of quality, it transmitted your actual internet traffic bits, up to 56kbps. The analog signal was made at the user's side of the telco and didn't have to fit within 56kbps when digitised.<p>Pedantically, the digital circuits are 64kbps but one bit in some bytes is used for call status signaling, which is okay for voice, but the ISP equipment can't predict which bytes have a bit overwritten (and it could be multiple if there are several hops) so it just used 7 bits in each byte.
<i>Not quite, and an interesting story that fits these engineering maxims better than you might think.</i><p>Huh? The SotA for dial-up modems in the 1990s was indeed 56 kbps, achieved with various encoding and compression hacks in both domains including trellis coding.<p>Weirdly enough, you can apparently still buy them from US Robotics: <a href="https://www.usr.com/products/56k-dial-up-modems/" rel="nofollow">https://www.usr.com/products/56k-dial-up-modems/</a>
Law 20 seems to express the state of most startups these days:<p>> "A bad design with a good presentation is doomed eventually. A good design with a bad presentation is doomed immediately."
Imagine that you're a highly intellectual, highly technical, and highly responsible person in control of large sums of governmental or corporate money. You don't want to waste the money, you want stellar results (in spacecraft industry, maybe literally so).<p>Would you assign a large sum of money to a group that cannot present their design clearly, neatly, and concisely? If they are struggling even with that, would you trust them to be good at actually designing a spacecraft soundly, economically, and in a reasonable time?<p>"If you can't explain it to a five-years-old, you likely do not understand it yourself", said one of the greatest modern scientists, who also was notoriously good at explaining things.
What often happens next is, another party comes up in the middle to manage the interaction between both of you (with the proper bump in the ask price), because there's not only so many decision makers looking for neat presentations and whatnot but also there's only so many teams willing to do the actual work.
I think that Feynman was talking about preparing a freshman-level lecture for Caltech-standard freshmen, but maybe you have somebody else in mind.
I would say: You either don't understand your subject, or don't understand your audience, if you can't explain your subject to your audience, at the highest level they can understand, coherently.<p>The average person can understand anything ... at some level. Being able to match that level is positive evidence (but not proof) of competence.<p>Duality: Being unable to match that level is positive evidence (but not proof) of incompetence.
I think if you understand something <i>really</i> well (anything: the law of gravity, the Curry-Howard isomorphism, electrolytic dissociation, general relativity,...), you can find a bunch of comparisons, or metaphors, or other ways to explain it so that an interested five-years-old will get a rough idea. A very rough idea indeed, but one that could allow them to ask qualitatively reasonable questions, and that forms an intuition which helps during a real study.
The "interested" part does a lot of lifting though. It's really hard to explain things to uninterested people.<p>If the person you are explaining your project to is not interested in the technical side, presumably under the rather confused but popular theory that technical aspects are not relevant to technology ventures, you'll not be making headway. It's much better to just make up some dollar numbers and run with that.
That's what happens when the likely success scenario is selling out to an existing company rather than growing to be a genuinely large and long lived company.
There are beef companies and milk companies, depending on the way they plan to use their cash cow.<p>(Most VC funding is used to quickly produce a beefy market share, and sell it to those who think they can milk it, or to profitably butcher it.)
I'd add to that: If you recognize a good design presented poorly, be the one to stand up and present it well, otherwise you will be stuck with the bad one.
Hiring salesmen to talk to other salesmen is always the sleaziest part of doing anything productive. You could say the same thing about opening a restaurant.
> The biggest commercial success is not the best technical design: Nokia N95 versus the first generation iPhone<p>That’s not a good example. Neither is Beta vs VHS. The most they illustrate is a different law I am coining right here:<p>Canyon’s Law of Design Optimization: you will inevitably choose to optimize for different metrics than your customers would wish. Don’t try to convince them they are wrong.
That's not a good example, but for a different reason: the N95 outsold the original iPhone.<p>The original iPhone was a promising proof of concept. It got the form factor and the interface right, but the actual device was underwhelming. It had no 3G, no GPS, no third-party apps, and a weak camera. iPhone 3G added all the features competitors already had (apart from a good camera) and became a much bigger commercial success.
> <i>That’s not a good example</i><p>It’s a great example. One can list a litany of technical specifications on which the N95 is superior. That, however, doesn’t make it a superior product.
I had to look up the N95. Yeah, Wikipedia goes to pains to rattle off things that made it better than the iPhone, but then I looked at a photo of the device and it was clear why the iPhone "won".
I has a Nokia N95. The phone itself was great. The problem was the dearth of apps. Nobody developed anything for windows mobile OS. Maybe Ballmer was not so crazy when he was running around on stage screaming “Developers, developers, developers”.
The last and most important slide is missing.<p>"Ignore all the advise above and do the right thing
Subtext: This will take multiple lifetimes to accomplish"<p>This is particularly important considering that some of the advice is at odds with each other and engineering is an unending juggling of tradeoffs. It's also by far the hardest to achieve both technically and socially but worth striving for.
While this PDF might be new, Akin's Laws of Spacecraft Design dates back to 2003.<p><a href="https://web.archive.org/web/20031101212246/https://spacecraft.ssl.umd.edu/akins_laws.html" rel="nofollow">https://web.archive.org/web/20031101212246/https://spacecraf...</a>
The first law already gives a good reason why software "engineering" is rarely actually engineering.
Almost all the laws are great guidance for Software dev to be honest and idiomatic to a lot of what's banded around today as good practice.
Oh god it can be worse when the suits try to measure everything. Some things are measurable and others are taste / gut and that is OK.
Agreed (unfortunately). It's also a good reminder IMHO to think and do engineering so we can one day be worthy of the title. Be the change you wish to see in the world. :)
I've been quoting von Tiesenhausen's Law of Engineering Design for over a decade, since it is a great summary of why I switched from engineering to product design mid-career. That law is the one that says <i>engineers always wind up designing the vehicle to look like the initial artist's concept.</i> I didn't engineer spacecraft, but on web projects I noticed that whoever made the documents furthest upstream had a ridiculous amount of influence over the outcome of the product. Even just being the one taking notes in the first meeting gives you leverage in a process which, despite claims of being agile, is definitively path-dependent most of the time.
These are awesome. I took that capstone spacecraft design course at MIT in 2003, and can't recall encountering this list (based on the archive snapshot, he would have shifted on to UMD more than thirty years prior). The project was satellite-focused rather than launch vehicle, so maybe his instructions were imbued implicitly into the course?<p>• Much of the design-conservative ethos permeates aerospace development. It's unsurprising that astronautics evolution has been slow at least until Elon came along. I wonder how Elon / SpaceX folks would respond to these laws esp. #39 (avoid designing launch vehicles).<p>• Also the one that was conspicuously maintained at the end across the different archived versions:<p>"Space is a completely unforgiving environment. If you screw up the engineering, somebody dies (and there's no partial credit because most of the analysis was right...)"<p>It's notable that the wayback machine's first crawl of Akin's list is late 2003 (so presumably when the source page went live) the year in which the Columbia disaster took place.
<i>> I wonder how Elon / SpaceX folks would respond to these laws esp. #39 (avoid designing launch vehicles).</i><p>Perhaps, based on specific context, let the Law's vote!<p><i>> Law 11: Sometimes, the fastest way to get to the end is
to throw everything out and start over.<p>> Law 16: The previous people [...]
did not have a direct pipeline to the wisdom
of the ages. There is therefore no reason to
believe their analysis [was optimal].<p>> Law 31 (Mo's Law of Evolutionary
Development): [...] understand the fundamental
limitations of [the existing] technology/approach.</i><p>IMHO: Law 31 is the kicker. That triggers Law 11. With general support from Law 16.
> <i>> I wonder how Elon / SpaceX folks would respond to these laws esp. #39 (avoid designing launch vehicles)</i><p>SpaceX has designed 3 (four if you include Falcon 1) launch vehicles: Falcon 1, Falcon 9, Falcon Heavy and Starship. The first three run the same engine family and are arguably in the same vehicle family.<p>SpaceX has absolutely embraced avoiding designing new launch vehicles (and even major components) until the fundamental limitations of the existing approach are exceeded.
“Avoid the temptation to believe a completely
new product will always be better than an
evolution of an old product”.
Buildsjets laws of spacecraft design:<p>The propellant storage shall be designed and located such that a catastrophic failure of propellant storage will not damage the passenger compartment.<p>The propulsion system shall be designed and located such that a catastrophic failure of the propulsion system will not damage the propellant storage or the passenger compartment.<p>The launch system shall be designed to ensure a minimum of two survivable abort alternatives at each phase of the flight. Each abort scenario shall be validated by a flight test before certifying the system for general use.<p>The re-entry system shall be designed so there are no single points of failure. If single points of failure are unavoidable, a method pf inspection or surveillance shall be developed to detect the failure prior to de-orbit.<p>In-orbit repair procedures for foreseeable types of damage shall be developed and validated prior to certifying the system for general use.<p>Yeah, this is all 20/20 hindsight. But we really need to avoid developing ANYTHING similar to the STS in the future. I truly believe it set us back by 50 years.
I like systems that are maintence free and easily replaceable. My experience so far in software engineering is that technologies die, so it should also be easy to replace the tecnology, like the hardware it runs on, the platform/os, the programming language and the framework.
Last archived working original: <a href="https://web.archive.org/web/20250808213459/https://spacecraft.ssl.umd.edu/akins_laws.html" rel="nofollow">https://web.archive.org/web/20250808213459/https://spacecraf...</a>
Adapted to writing:<p><a href="https://en.wikipedia.org/wiki/Wikipedia:Akin%27s_Laws_of_Article_Writing" rel="nofollow">https://en.wikipedia.org/wiki/Wikipedia:Akin%27s_Laws_of_Art...</a>
> <i>Ask any industrial engineer about MBAs</i> (in Law 7)<p>In the Navy-nuke world there was a saying, purportedly by Admiral Rickover, about non-technical "managers": There but for the grace of God goes God.
Many of these laws can be applied to any discipline. Business, politics, etc.
I’m Henshaw (#37). AMA.
What's the story with the Avro C102 (per law 20)? What's the connection with "A bad design with a good presentation is doomed eventually. A good design with a bad presentation is doomed immediately"? I'm intrigued.
Two references come to mind – the first is that the Avro C102 was beat by the de Havilland Comet, which was an example of a <i>bad</i> design with a good presentation. It had a series of mysterious hull losses that were eventually attributed to repeated stress on the corders of a square cut-out for an escape hatch.<p><a href="https://en.wikipedia.org/wiki/De_Havilland_Comet#Comet_disasters_of_1954" rel="nofollow">https://en.wikipedia.org/wiki/De_Havilland_Comet#Comet_disas...</a><p>The second, more depressingly, is that the Avro C102 was cancelled to redirect funds to the Avro Arrow, Canada's mythical first and only jet fighter, of which only a handful were produced before the program was cancelled, and Avro wound up.
> Bhargava’s Law: Only 1 out of 10 research ideas make it into industrial practice.<p>Not sure of the source for this. Nevertheless, this is ridiculously high percentage of projects that ever see an industrial angle, at least in basic sciences. Perhaps, this is restricted to engineering.
> Quality thinking is more important in this profession than fast thinking.<p>Feels true, particularly in an era where LLMs make fast thinking cheap.
Law 20: A good design with a bad presentation is doomed immediately<p>I definitely struggle with this. I run a math education site and I usually focus heavily on technical accuracy but underestimate the presentation.<p>Hard lesson that being "right" isn't enough if the delivery is clunky.
I have learned to think about this problem thus: there's reality, and then there's perceptions, and communication is a task of persuading somebody else's perception to somehow align with yours. When you both view reality clearly, it's best to present simple facts and their implications. The other three cases are education, bullshitting, and nonsense, and it's best to involve a professional.
Something I've learned, loosely related:<p>Adding "just a few fundamental equations" to a presentation won't make your case more compelling to business stakeholders. You lose roughly 10% of engagement for every greek letter in a slide.
care to share the site? :)
yeah, i learned this the hard way. I tried producthunt, indie hackers, linkedIn, twitter even instagram thinking distribution was the issue.<p>Turns out it was mostly audience and presentation. Math/educational content just doesn't travel well on those channels unless it’s tailored very carefully. Being right wasnt enough if the delivery didn’t match how people consume it.
I have these taped up above my desk at work. I've been able to point to at least half of them from one time or another
> Bhargava’s Law: Only 1 out of 10 research ideas
make it into industrial practice<p>It's a nice reflection, but what is the origin of this? Can't find another reference to this "law" online.
Not all laws are hard science laws in the sense of the second law of thermodynamics; they are, however, good approximations based on experience and, in the right context, make a bit more sense.<p>Notes from the original author:<p>> I've been involved in spacecraft and space systems design and development for my entire career, including teaching the senior-level capstone spacecraft design course, for ten years at MIT and now at the University of Maryland for more than a decade. These are some bits of wisdom that I have gleaned during that time, some by picking up on the experience of others, but mostly by screwing up myself. I originally wrote these up and handed them out to my senior design class, as a strong hint on how best to survive my design experience. Months later, I get a phone call from a friend in California complimenting me on the Laws, which he saw on a "joke-of-the-day" listserve. Since then, I'm aware of half a dozen sites around the world that present various editions of the Laws, and even one site which has converted them to the Laws of Certified Public Accounting. (Don't ask...) Anyone is welcome to link to these, use them, post them, send me suggestions of additional laws, but I do maintain that this is the canonical set of Akin's Laws...
> The schedule you develop will seem like a complete work of fiction up until the time your customer fires you for not meeting it. (Law 23)<p>So will Musk finally be fired in 2026?
I'm not sure if the nokia example works. When the nokia launched the screen technology, SoC horsepower, battery tech, etc just wasn't there to make an iphone. Even when the 2007 iphone launched it was a bit of mess, with the first gen not being 3G when other phones were and no app store, but instead devs were told to write web apps.<p>If anything, some of these early smartphones were pushing a lot of limits considering the hardware restraints. Its just by the time the iphone came out, these restraints were lessened and Apple did a good job using these technologies.
I have found that my best designs, few and far between, enter a period where they get simpler as they are completed. And my worst or failed designs keep getting more complex as I go on.
This is the way.<p>"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius — and a lot of courage to move in the opposite direction."<p><a href="https://www.goodreads.com/quotes/14789-any-intelligent-fool-can-make-things-bigger-more-complex-and" rel="nofollow">https://www.goodreads.com/quotes/14789-any-intelligent-fool-...</a>
My contributions/spins on this:<p>The most general problem cannot be solved. (If you don't limit scope, you will never finish. You won't even finish the design.)<p>If you want it bad, that's how you're going to get it. (That is, rushing a project means you get crummy results. This may be "Hanka's Law", because I first heard it from Steve Hanka, but it may not be original to him.)
Lol, see the recent HN post on TRIZ and my comment there pointing folks to this. And here I complete the cycle.
My general issue with this is that it is overly hardware centric and not as accurate when it comes to Aerospace Software<p>Law 4 Bhargava’s Law: Only 1 out of 10 research ideas make it into industrial practice is wrong anecdotally particularly when it relates to software.<p>Law 13 is flat-out wrong in that there is a huge amount of potential SWaP trades & innovation trades to be made, and the changing requirements environment where it is easy to predict where a requirement will be, despite a space program with a legacy requirements baseline.<p>An example of Law 13 errors would be the JPSS security redesign campaigns, and a less ideal retrofit
TL;DR:<p>Minimize negative(painful) notions as much as possible, ideally approaching zero, while maximizing positive (pleasurable) notions.<p>Minimize negative(painful) notions:
Uncertainty,
Risk,
Chaotic behavior,
Randomness,
Non-deterministic,
Instability,
Cost,
Energy losses,
Time consumption,
Resource usage,
Excessive complexity,
Failure modes,
Noise<p>Maximize positive(Pleasure) notions:
Reliability,
Efficiency,
Deterministic,
Predictability,
Precision,
Accuracy,
Verification,
Validation,
Safety,
Stability,
Simplicity (lower complexity),
Robustness,
Redundancy
I can think of a few SaaS products in the document scanning and OCR space whose UIs are not efficient or simple, while being time consuming and, to my mind, chaotic.<p>There should be an Akin Exit Clause from said 3-year contracts. They have zero incentives to fix or improve _anything_ during those years of servitude.