5 ms·

> the absolutely batshit advice and business building strategy that comes out of the world of venture capital and tech accelerators Part of this includes the w

> the absolutely batshit advice and business building strategy that comes out of the world of venture capital and tech accelerators Part of this includes the whole concept of rushing a 'knocked together' MVP out there. It all depends on the product. Yeah, if you have an app to share cat pictures on a social network, then who cares if you have a flakey cobbled together app that doesn't upload half the time. The consequences are minimal. If you have an app that is responsible for administering a life saving drug on a periodic basis, then you had better be sure the thing is polished, tested and reliable before even beginning to test it in the market. The whole 'lean' system sausage factory is just that - creating a mass production line of ideas where only less than 1% float to the top. Very often only based on the fact that the creators shouted louder than others who may have had better (or more boring) ideas.

asdguiuiobuibui 8y ago
Or if you're, say, building a system for self-driving cars. This Silicon Valley arrogance and recklessness has killed people. It's not just academic.
- jimbofisher1 8y ago
  Don't understand the hang up about this. Self Driving cars were always going to kill people. It's not possible to be 100% safe but even a minute improvement over human drivers would save thousands of lives a year.
  - TheCoelacanth 8y ago
    Uber has had a fatal accident with only a few million miles driven when an average human will go more like a hundred million miles per fatality. I don't understand why people insist on believing that they are already more safe than human drivers when the evidence suggests that they are currently much, much less safe.
    - jimbofisher1 8y ago
      I'm sorry where did I imply they were already safer? I think it's obvious they will continue to improve and surpassing human level is not far away.
      - TheCoelacanth 8y ago
        Nonsense. Uber's self-driving cars currently have something like 30x the fatality rate of an average driver and that is in cherry-picked locations that they chose for testing and with a human driver constantly watching it to take over when it does something wrong. There is no evidence that they are anywhere close to surpassing human drivers.
        jimbofisher1 8y ago
        Given a sample size of 1 I don't think there's much to be discerned form the data at all either way. > There is no evidence that they are anywhere close to surpassing human... Famous last words.
  - im3w1l 8y ago
    And even a slight worsening could be a worthwhile tradeoff for the economic gains it would give us.
Felz 8y ago
What's the alternative? Overfunded waterfall-style development on fewer things nobody wants? The lean startup model seems like it works, even outside of tech. Even yes, in medicine that can kill as a failure mode. Drug development, for example, consists of trying a bunch of plausible things. (And worked until the effective, simple chemical space was exhausted.)
- lainga 8y ago
  ...and moving a very small fraction of those drugs into long FDA trials.
  - 6cd6beb 8y ago
    I've been in a couple of those trials. They make it very clear that the drug has already been tested extensively and also do fairly extensive testing on you to make sure your vitals are within the "normal" range. I got a very clear impression that they're not throwing pills at the wall to see which one sticks. I'm also working at a startup that's had three names and just as many products, funded by a VC who's just going to start another one with most of the same people whether this one works or not. It doesn't matter one bit if this startup fails. The drug was pretty clearly indicated to work by the time a person swallowed a pill. Different worlds.
- filleduchaos 8y ago
  Drug development certainly does not consist of trying a bunch of plausible things on the public (or even on small carefully picked groups of human trial subjects, not until after extensive tests and trials have been exhausted). And thank fuck it doesn't. Besides, it's rather disingenuous to act as though development falls on a binary between "overfunded waterfall-style development" and the current mess at many shops.
  - Felz 8y ago
    My apologies if my comment came off as disingenuous or overly binary. I'd agree you could take minimal MVPs too far. (Also, the phrase "minimal MVP" is RAS syndrome but how would I say it otherwise?) What I haven't seen is "the current mess at many shops", and I'd be interested to hear about your experiences. I've only ever experienced failure by overplanning personally, and the stark divide between progress in things that can be cheaply iterated on and things that are expensive, preplanned disasters seems important to me. This leads me to naturally think that if I wanted to make progress in e.g. medicine I should start by finding out how to make it cheap and iterable, rather than come up with comprehensive, well-funded plans.

asdguiuiobuibui 8y ago

Or if you're, say, building a system for self-driving cars. This Silicon Valley arrogance and recklessness has killed people. It's not just academic.

jimbofisher1 8y ago
Don't understand the hang up about this. Self Driving cars were always going to kill people. It's not possible to be 100% safe but even a minute improvement over human drivers would save thousands of lives a year.
- TheCoelacanth 8y ago
  Uber has had a fatal accident with only a few million miles driven when an average human will go more like a hundred million miles per fatality. I don't understand why people insist on believing that they are already more safe than human drivers when the evidence suggests that they are currently much, much less safe.
  - jimbofisher1 8y ago
    I'm sorry where did I imply they were already safer? I think it's obvious they will continue to improve and surpassing human level is not far away.
    - TheCoelacanth 8y ago
      Nonsense. Uber's self-driving cars currently have something like 30x the fatality rate of an average driver and that is in cherry-picked locations that they chose for testing and with a human driver constantly watching it to take over when it does something wrong. There is no evidence that they are anywhere close to surpassing human drivers.
      - jimbofisher1 8y ago
        Given a sample size of 1 I don't think there's much to be discerned form the data at all either way. > There is no evidence that they are anywhere close to surpassing human... Famous last words.
- im3w1l 8y ago
  And even a slight worsening could be a worthwhile tradeoff for the economic gains it would give us.

Felz 8y ago

What's the alternative? Overfunded waterfall-style development on fewer things nobody wants? The lean startup model seems like it works, even outside of tech. Even yes, in medicine that can kill as a failure mode. Drug development, for example, consists of trying a bunch of plausible things. (And worked until the effective, simple chemical space was exhausted.)

lainga 8y ago
...and moving a very small fraction of those drugs into long FDA trials.
- 6cd6beb 8y ago
  I've been in a couple of those trials. They make it very clear that the drug has already been tested extensively and also do fairly extensive testing on you to make sure your vitals are within the "normal" range. I got a very clear impression that they're not throwing pills at the wall to see which one sticks. I'm also working at a startup that's had three names and just as many products, funded by a VC who's just going to start another one with most of the same people whether this one works or not. It doesn't matter one bit if this startup fails. The drug was pretty clearly indicated to work by the time a person swallowed a pill. Different worlds.
filleduchaos 8y ago
Drug development certainly does not consist of trying a bunch of plausible things on the public (or even on small carefully picked groups of human trial subjects, not until after extensive tests and trials have been exhausted). And thank fuck it doesn't. Besides, it's rather disingenuous to act as though development falls on a binary between "overfunded waterfall-style development" and the current mess at many shops.
- Felz 8y ago
  My apologies if my comment came off as disingenuous or overly binary. I'd agree you could take minimal MVPs too far. (Also, the phrase "minimal MVP" is RAS syndrome but how would I say it otherwise?) What I haven't seen is "the current mess at many shops", and I'd be interested to hear about your experiences. I've only ever experienced failure by overplanning personally, and the stark divide between progress in things that can be cheaply iterated on and things that are expensive, preplanned disasters seems important to me. This leads me to naturally think that if I wanted to make progress in e.g. medicine I should start by finding out how to make it cheap and iterable, rather than come up with comprehensive, well-funded plans.