14 ms·

Sure, and evolution would be much more efficient if every mutation were positive and the species adapted that specific one instead of going out on their own cre

by croon 10y ago

Sure, and evolution would be much more efficient if every mutation were positive and the species adapted that specific one instead of going out on their own creating meaningless mutations. Who creates the best/bulletproof solutions? You're arguing that we collectively should know which one is the best, among all our different opinions and habits. I'll instead argue that we move forward through natural selection and that everyone learns from trying, failing, trying, succeeding and learning what the best solutions are by experience. I'm not saying it's theoretically optimal, or even practically, but I would suggest it's only natural that we've landed here. That said, I'm all for trying to improve, if you or someone else every comes up with a better route.

jacquesm 10y ago
Evolution is nothing like engineering. Evolution by its very nature makes forward and backwards movements relative to what an outside observer would consider to be an optimum, and selection pressure is what eventually gives rise to a perceived movement in the 'right' direction. Software is - in principle - an engineering domain, but we tend to be much more liberal in applying engineering norms to software production schedules and reliability. This need not be the case. If on the other hand you wish to argue that software is produced in a random fashion and selection pressure is what decides what survives then I rest my case, that view definitely does explain a lot of what I'm seeing.
- croon 10y ago
  Sure, your points on their differences are correct, I agree with that. But I still think my comparison holds. Basically the comparison was made to convery Sturgeon's Law; that 90% of everything made is crap. In that sense it is like evolution: A numbers game. While not explicitly random, distributing risk among groups going in different direction could be an efficient way forward, when counting for what would be lost with some central governance. You then select the optimal direction by results.
  - jacquesm 10y ago
    90% of all bridges and buildings, powerstations, cranes and heavy machines are not crap. Engineering is a responsible job, it's nothing at all like making clay vases or plates where if you throw away 9 out of 10 of the things you make isn't going to cause anybody to suffer other than yourself.
    - croon 10y ago
      Sure, but that's because of safety necessity, not because of optimal solution. If building bridges, powerstations, cranes, etc etc were essentially free outside of drawing/planning them, and could be deployed outside of "production" infrastructure, I'm sure that industry would be lightyears ahead of where it is.
      - jacquesm 10y ago
        Software is rapidly becoming just as critical, and at some point may be more critical. The ability to copy a product is a thing that works to your advantage after you've created it, the fact that each bridge and each building is engineered with slowly changing techniques and standards is what makes engineering a solid profession. I think that we probably have had too much change in too short a time thrown at us in the software related professions, if our hardware and other engineered products would perform at the same reliability level and with the same degree of resource consumption that our software routinely gets away with - not even looking at the risks associated with deploying crappy software - we would have returned the product to the manufacturer. But somehow we've collectively managed to convince our customers that this is the best we can do and that there is no honor in going slower but safer. This is all fine until software becomes critical to our survival, and we're definitely passing some threshold in that respect and as an industry I don't think we are ready to accept that responsibility without making some major changes in how we go about our daily work.
    - coredog64 10y ago
      What if they are crap, but just by a metric that isn't "falls apart immediately"? I watched my state government build an overpass that wasn't needed and didn't even have an _exit_ for 3 years. It required multiple road closures and for several months made traffic awful and my drive unsafe. I call that "crap".

by croon 10y ago

jacquesm 10y ago

Evolution is nothing like engineering. Evolution by its very nature makes forward and backwards movements relative to what an outside observer would consider to be an optimum, and selection pressure is what eventually gives rise to a perceived movement in the 'right' direction. Software is - in principle - an engineering domain, but we tend to be much more liberal in applying engineering norms to software production schedules and reliability. This need not be the case. If on the other hand you wish to argue that software is produced in a random fashion and selection pressure is what decides what survives then I rest my case, that view definitely does explain a lot of what I'm seeing.

croon 10y ago
Sure, your points on their differences are correct, I agree with that. But I still think my comparison holds. Basically the comparison was made to convery Sturgeon's Law; that 90% of everything made is crap. In that sense it is like evolution: A numbers game. While not explicitly random, distributing risk among groups going in different direction could be an efficient way forward, when counting for what would be lost with some central governance. You then select the optimal direction by results.
- jacquesm 10y ago
  90% of all bridges and buildings, powerstations, cranes and heavy machines are not crap. Engineering is a responsible job, it's nothing at all like making clay vases or plates where if you throw away 9 out of 10 of the things you make isn't going to cause anybody to suffer other than yourself.
  - croon 10y ago
    Sure, but that's because of safety necessity, not because of optimal solution. If building bridges, powerstations, cranes, etc etc were essentially free outside of drawing/planning them, and could be deployed outside of "production" infrastructure, I'm sure that industry would be lightyears ahead of where it is.
    - jacquesm 10y ago
      Software is rapidly becoming just as critical, and at some point may be more critical. The ability to copy a product is a thing that works to your advantage after you've created it, the fact that each bridge and each building is engineered with slowly changing techniques and standards is what makes engineering a solid profession. I think that we probably have had too much change in too short a time thrown at us in the software related professions, if our hardware and other engineered products would perform at the same reliability level and with the same degree of resource consumption that our software routinely gets away with - not even looking at the risks associated with deploying crappy software - we would have returned the product to the manufacturer. But somehow we've collectively managed to convince our customers that this is the best we can do and that there is no honor in going slower but safer. This is all fine until software becomes critical to our survival, and we're definitely passing some threshold in that respect and as an industry I don't think we are ready to accept that responsibility without making some major changes in how we go about our daily work.
  - coredog64 10y ago
    What if they are crap, but just by a metric that isn't "falls apart immediately"? I watched my state government build an overpass that wasn't needed and didn't even have an _exit_ for 3 years. It required multiple road closures and for several months made traffic awful and my drive unsafe. I call that "crap".