Nicholas Piano:
Off topic: medical students must dissect a whole cadaver to get their licence. What would be similar for programmer licence?
StJohn Piano:
Implementing an API on a server? Complete with automated deployment.
Nicholas Piano:
It needs to be something that proves someone's worth beyond a doubt. Imagine we treated programmer failure like the death of a patient in surgery.
Think of a system tested so completely that every possible use case has been accounted for.
Perhaps an air traffic control simulation or a bookkeeping task
Like the Kobayashi Maru
StJohn Piano:
Yeah
Hm. Don't know.
Can't do it perfectly. We don't understand the body perfectly.
But could do it very well indeed, yes.
Can't have "beyond doubt" but could have "beyond even very unreasonable doubt".
Nicholas Piano:
Consider: medical students start for two reasons; high pay and desire to help people. These are both very visceral motivations. You cannot have programmer licensing without a drive as strong as this.
People became airborne rangers in the US army during WW2 because of slightly higher pay
StJohn Piano:
yes
didn't know that about rangers. ha.
hm
ok, how about this:
q: why are medical licences helpful?
a: because patients, hospitals, insurers etc want to see them, or at least know that they exist and can be verified.
because people are quite sensitive about their survival
up to now, software (outside of e.g. air traffic control, trains) has not been critical to survival
sony etc lose customer data all the time. no one cares. no one stops using them.
speed of deployment and time to market has mattered most for the last generation (or more) in software [0]
and then growth, until ideally you crack the world with your weight (e.g. microsoft)
this approach necessarily involves a lot of leakage, both of money and of personal information, and of control of user accounts.
(convenience necessarily trades off against security)
now, post-bitcoin, something changed.
money can be scarce [1]
mistakes with code that handles user account access (i.e. identity) and with code that handles money can hurt.
really hurt.
because the damage cannot be printed away
and pain makes people sensitive to risk.
so I think we'll get programming licences because those who purchase programming time now have a reason to care about programmer quality much more than they used to. [2]
this applies even if bitcoin dies, etc etc, and is replaced by 5 variants. the pain will still exist, perhaps in a more diluted form, but it will still be present, more so than in pounds or dollars.
obviously this will take a while. it'll happen first wherever software systems are interfacing with bitcoin. or are under pressure from ransomware. wherever it hurts the most.
StJohn Piano:
^ mind if I publish this convo?
starting from "medical students must dissect a whole cadaver to get their licence."
Nicholas Piano:
Yes, no problem
StJohn Piano:
Want to be anonymised, or happy to have your name in print?
Nicholas Piano:
Name is fine
StJohn Piano:
Cool
Off topic: medical students must dissect a whole cadaver to get their licence. What would be similar for programmer licence?
StJohn Piano:
Implementing an API on a server? Complete with automated deployment.
Nicholas Piano:
It needs to be something that proves someone's worth beyond a doubt. Imagine we treated programmer failure like the death of a patient in surgery.
Think of a system tested so completely that every possible use case has been accounted for.
Perhaps an air traffic control simulation or a bookkeeping task
Like the Kobayashi Maru
StJohn Piano:
Yeah
Hm. Don't know.
Can't do it perfectly. We don't understand the body perfectly.
But could do it very well indeed, yes.
Can't have "beyond doubt" but could have "beyond even very unreasonable doubt".
Nicholas Piano:
Consider: medical students start for two reasons; high pay and desire to help people. These are both very visceral motivations. You cannot have programmer licensing without a drive as strong as this.
People became airborne rangers in the US army during WW2 because of slightly higher pay
StJohn Piano:
yes
didn't know that about rangers. ha.
hm
ok, how about this:
q: why are medical licences helpful?
a: because patients, hospitals, insurers etc want to see them, or at least know that they exist and can be verified.
because people are quite sensitive about their survival
up to now, software (outside of e.g. air traffic control, trains) has not been critical to survival
sony etc lose customer data all the time. no one cares. no one stops using them.
speed of deployment and time to market has mattered most for the last generation (or more) in software [0]
and then growth, until ideally you crack the world with your weight (e.g. microsoft)
this approach necessarily involves a lot of leakage, both of money and of personal information, and of control of user accounts.
(convenience necessarily trades off against security)
now, post-bitcoin, something changed.
money can be scarce [1]
mistakes with code that handles user account access (i.e. identity) and with code that handles money can hurt.
really hurt.
because the damage cannot be printed away
and pain makes people sensitive to risk.
so I think we'll get programming licences because those who purchase programming time now have a reason to care about programmer quality much more than they used to. [2]
this applies even if bitcoin dies, etc etc, and is replaced by 5 variants. the pain will still exist, perhaps in a more diluted form, but it will still be present, more so than in pounds or dollars.
obviously this will take a while. it'll happen first wherever software systems are interfacing with bitcoin. or are under pressure from ransomware. wherever it hurts the most.
StJohn Piano:
^ mind if I publish this convo?
starting from "medical students must dissect a whole cadaver to get their licence."
Nicholas Piano:
Yes, no problem
StJohn Piano:
Want to be anonymised, or happy to have your name in print?
Nicholas Piano:
Name is fine
StJohn Piano:
Cool
[start of notes]
Chat transcript edited slightly for readability.
[end of notes]
[start of footnotes]
[0]
"generation" here refers to a human generation. I am using "speed of deployment" to refer to "speed of deployment of new features", whereas "time to market" is "time taken to get minimum viable product to market". These should probably be the other way around. I should also add "compatibility with existing software and hardware" - this has consistently been a winning characteristic for 40+ years, I think.
[return to main text]
[1]
This should be "scarce again". It's been scarce before. Consider the effort people used to go to in order to move gold around:
John T. Flynn: How the great Rothschild family of bankers got started
John T. Flynn: How the great Rothschild family of bankers got started
[return to main text]
[2]
Not just quality, but also a reputation for trustworthiness.
[return to main text]
[end of footnotes]