5 ms·

Quality Is Fractal

27 points by kirillzubovsky 12y ago

briantakita 12y ago
The path of mastery is also fractal. You can iterate & zoom in on concepts & practices. Over time, repeating patterns will be seen & seemingly unrelated concepts will be connected in simple ways.
- mattgreenrocks 12y ago
  Absolutely. Mastery is knowing what details to pay attention to. When you're learning, you don't know which ones, so you pay attention to all of them until they progressively become more automatic. This requires a lot of focus and mental energy, but it deeply rewarding in it's own right.
lifeisstillgood 12y ago
I will lay a decent amount of money this is the next meme - it is succinct and yet once you understand applicable everywhere. I will certainly be quoting the phrase "quality is fractal" - so will you soon:-)
mooneater 12y ago
Agreed. The implication is, to change a low-quality X to a high-quality X is a huge amount of work, that scales super-linearly to the complexity of X.
mooneater 12y ago
Agreed. The implication is, to change a low-quality X to a high-quality X is a huge amount of work, that scales super-linearly to the complexity of X.
the_cat_kittles 12y ago
You explain the concept, but where do you offer any evidence to support it? I don't dismiss the concept, in fact I think its an important one, but if you are making such a bold assertion, you would be well served to have something supporting your claim.
- CGamesPlay 12y ago
  Strongly agree. I'd love to see the study or what Nathan Bowers wrote.
  - th3iedkid 12y ago
    googling i ended up here for Nathan bowers article http://web.archive.org/web/20120214085940/http://uxhero.com/ux-theory/quality-is-fractal/ http://web.archive.org/web/20120214085940/http://uxhero.com/... Guess we mentioned fractals
sutterbomb 12y ago
While I mostly agree with the premise, I'm worried many people will internalize the wrong lesson. It seems to imply that quality, as implicitly defined by the examples provided, is a barometer of success. The reality is that there are numerous vectors that you can compete on outside of quality and still be wildly successful. I'd be happy to have 3 lines of subpar code if it gave me an advantage on other competitive angles.
- InclinedPlane 12y ago
  Sure, there are many routes to success. In general though I think quality is one of the strongest and can lead to the most enduring products and companies. Nevertheless, "quality" is an ill-defined concept, and it's easy to be led astray in pursuit of notional quality at the expense of real-world practical improvements to your product. For example, cost and ease of use are not independent factors. Consider VHS vs. BETAMAX video recording as a classic example. BETAMAX had superior recording fidelity, but it was significantly more expensive, and expense plus network effect was a major factor in the adoption of video tape standards. Part of the "quality fractal" are meta-features such as robustness, elegance, usability, performance, and cost. It doesn't matter how theoretically perfect your product is if it's not practical or it's too expensive to serve the market effectively. A Bugatti Veyron may seem in many respects to be a "perfect" automobile, but it's not. The luxurious interior, amazing build quality, and high performance make it easy to overlook the downsides such as the extreme cost, low mileage, low lifetime of many components (such as the tires), high maintenance cost, etc. In most regards a Toyota Camry is a superior car to a Veyron. That's a result that it's very difficult to arrive at if you focus too myopically on the ideal of "quality" without looking at things holistically. Compare, for example, consider the world-wide-web vs. project Xanadu, or Linux vs. Hurd. It's so easy to let perfect be the enemy of good. The concept of "worse is better" is on the surface a refutation of the importance of quality but in reality it's just a reminder that "quality" isn't always what we think it is and practicality is always by far the most important quality.
camdez 12y ago
I don’t find this to be an accurate description. Yes, quality (often) has this one bad apple spoils the whole bunch property. If we have to have a math metaphor we might say that quality is multiplicative, in the sense that one low value in a sequence still impacts the entire product. Or we might say that quality has an absorbing element, by which we mean that any zero value kills the whole set (100 * 100 * 100 * 0 still equals 0). But fractal means that we see self-similarity at all levels. That hardly seems to be the case. According to the original argument, bad software implies bad programmers which implies a bad company. That seems not only incorrect but also non-constructive. It ignores (e.g.) the idea that good employees could release bad products under bad management. Likewise, great programmers can write applications which are terrible to use. There are other skills involved in that process (interaction design, for instance). The criteria for evaluating programmers and software are vastly different and thus it doesn’t make sense to say that there’s a fractal relationship between these two vastly different kinds of entities.
nostromo 12y ago
I like the analogy, but is good advice to software startups? "Finished is better than perfect" is probably a better startup philosophy. I've seen too many friends waste time on the nth refactoring of their code rather than focusing on shipping and proving customer value.
- Swizec 12y ago
  "Finished is better than perfect [as long as it isn't crap]" I've seen too many products become a miserable mess because people believed in "finished is better than perfect". But that's something you learn when you're rewriting your code for the nth time to add the nth feature because otherwise you just can't fit anything new in there. But as I always say on these occasions, please keep shipping bad code. I make good money fixing it.
  - neurobro 12y ago
    Ah, but could your clients afford to hire you to clean up their messes if they hadn't finished something first?
    - Swizec 12y ago
      The clients I've worked with from the get go maintained enough flexibility to finish something every couple of weeks ;)
- jasonpeacock 12y ago
  quality != perfect You can still be fast and good, though it takes more discipline than just cranking out workable but crappy code.
gbaygon 12y ago
About a month ago it was posted an article[1] about why Van Halen requested the infamous no brown M&M's clause in their concert contracts, it struck me as a practical way of measuring overall quality on a complex situattion. [1] https://news.ycombinator.com/item?id=7754334 https://news.ycombinator.com/item?id=7754334
jenntoda 12y ago
Agreed, but I'd argue that striving for perfection in every detail is a bit of first world problem, and it also assumes availability of infinite resources. Starve a man for enough days, then give him a steak on a stick and I'll guarantee that he'll be happy just to tear the meat apart with his teeth caveman style, forget the formalities of napkin, plate and utensils. Same goes for software, esp. startups, where the first challenge is to get to having something that will do a right job however imperfectly, than the wrong job, however perfectly.

The path of mastery is also fractal. You can iterate & zoom in on concepts & practices. Over time, repeating patterns will be seen & seemingly unrelated concepts will be connected in simple ways.

mattgreenrocks 12y ago
Absolutely. Mastery is knowing what details to pay attention to. When you're learning, you don't know which ones, so you pay attention to all of them until they progressively become more automatic. This requires a lot of focus and mental energy, but it deeply rewarding in it's own right.

lifeisstillgood 12y ago

I will lay a decent amount of money this is the next meme - it is succinct and yet once you understand applicable everywhere. I will certainly be quoting the phrase "quality is fractal" - so will you soon:-)

mooneater 12y ago

Agreed. The implication is, to change a low-quality X to a high-quality X is a huge amount of work, that scales super-linearly to the complexity of X.

mooneater 12y ago

Agreed. The implication is, to change a low-quality X to a high-quality X is a huge amount of work, that scales super-linearly to the complexity of X.

the_cat_kittles 12y ago

You explain the concept, but where do you offer any evidence to support it? I don't dismiss the concept, in fact I think its an important one, but if you are making such a bold assertion, you would be well served to have something supporting your claim.

CGamesPlay 12y ago
Strongly agree. I'd love to see the study or what Nathan Bowers wrote.
- th3iedkid 12y ago
  googling i ended up here for Nathan bowers article http://web.archive.org/web/20120214085940/http://uxhero.com/ux-theory/quality-is-fractal/ http://web.archive.org/web/20120214085940/http://uxhero.com/... Guess we mentioned fractals

sutterbomb 12y ago

While I mostly agree with the premise, I'm worried many people will internalize the wrong lesson. It seems to imply that quality, as implicitly defined by the examples provided, is a barometer of success. The reality is that there are numerous vectors that you can compete on outside of quality and still be wildly successful. I'd be happy to have 3 lines of subpar code if it gave me an advantage on other competitive angles.

InclinedPlane 12y ago
Sure, there are many routes to success. In general though I think quality is one of the strongest and can lead to the most enduring products and companies. Nevertheless, "quality" is an ill-defined concept, and it's easy to be led astray in pursuit of notional quality at the expense of real-world practical improvements to your product. For example, cost and ease of use are not independent factors. Consider VHS vs. BETAMAX video recording as a classic example. BETAMAX had superior recording fidelity, but it was significantly more expensive, and expense plus network effect was a major factor in the adoption of video tape standards. Part of the "quality fractal" are meta-features such as robustness, elegance, usability, performance, and cost. It doesn't matter how theoretically perfect your product is if it's not practical or it's too expensive to serve the market effectively. A Bugatti Veyron may seem in many respects to be a "perfect" automobile, but it's not. The luxurious interior, amazing build quality, and high performance make it easy to overlook the downsides such as the extreme cost, low mileage, low lifetime of many components (such as the tires), high maintenance cost, etc. In most regards a Toyota Camry is a superior car to a Veyron. That's a result that it's very difficult to arrive at if you focus too myopically on the ideal of "quality" without looking at things holistically. Compare, for example, consider the world-wide-web vs. project Xanadu, or Linux vs. Hurd. It's so easy to let perfect be the enemy of good. The concept of "worse is better" is on the surface a refutation of the importance of quality but in reality it's just a reminder that "quality" isn't always what we think it is and practicality is always by far the most important quality.

camdez 12y ago

I don’t find this to be an accurate description. Yes, quality (often) has this one bad apple spoils the whole bunch property. If we have to have a math metaphor we might say that quality is multiplicative, in the sense that one low value in a sequence still impacts the entire product. Or we might say that quality has an absorbing element, by which we mean that any zero value kills the whole set (100 * 100 * 100 * 0 still equals 0). But fractal means that we see self-similarity at all levels. That hardly seems to be the case. According to the original argument, bad software implies bad programmers which implies a bad company. That seems not only incorrect but also non-constructive. It ignores (e.g.) the idea that good employees could release bad products under bad management. Likewise, great programmers can write applications which are terrible to use. There are other skills involved in that process (interaction design, for instance). The criteria for evaluating programmers and software are vastly different and thus it doesn’t make sense to say that there’s a fractal relationship between these two vastly different kinds of entities.

nostromo 12y ago

I like the analogy, but is good advice to software startups? "Finished is better than perfect" is probably a better startup philosophy. I've seen too many friends waste time on the nth refactoring of their code rather than focusing on shipping and proving customer value.

Swizec 12y ago
"Finished is better than perfect [as long as it isn't crap]" I've seen too many products become a miserable mess because people believed in "finished is better than perfect". But that's something you learn when you're rewriting your code for the nth time to add the nth feature because otherwise you just can't fit anything new in there. But as I always say on these occasions, please keep shipping bad code. I make good money fixing it.
- neurobro 12y ago
  Ah, but could your clients afford to hire you to clean up their messes if they hadn't finished something first?
  - Swizec 12y ago
    The clients I've worked with from the get go maintained enough flexibility to finish something every couple of weeks ;)
jasonpeacock 12y ago
quality != perfect You can still be fast and good, though it takes more discipline than just cranking out workable but crappy code.

gbaygon 12y ago

About a month ago it was posted an article[1] about why Van Halen requested the infamous no brown M&M's clause in their concert contracts, it struck me as a practical way of measuring overall quality on a complex situattion. [1] https://news.ycombinator.com/item?id=7754334 https://news.ycombinator.com/item?id=7754334

jenntoda 12y ago

Agreed, but I'd argue that striving for perfection in every detail is a bit of first world problem, and it also assumes availability of infinite resources. Starve a man for enough days, then give him a steak on a stick and I'll guarantee that he'll be happy just to tear the meat apart with his teeth caveman style, forget the formalities of napkin, plate and utensils. Same goes for software, esp. startups, where the first challenge is to get to having something that will do a right job however imperfectly, than the wrong job, however perfectly.