| 1 |
On Fri, Apr 26, 2019 at 10:56 AM Michał Górny <mgorny@g.o> wrote: |
| 2 |
|
| 3 |
> On Fri, 2019-04-26 at 16:29 +0200, Alexis Ballier wrote: |
| 4 |
> > On Fri, 12 Apr 2019 18:25:57 +0200 |
| 5 |
> > Michał Górny <mgorny@g.o> wrote: |
| 6 |
> > |
| 7 |
> > > Why do you |
| 8 |
> > > presume that ComRel will never abuse its power, and at the same time |
| 9 |
> > > presume QA will kick people 'on a whim'? |
| 10 |
> > |
| 11 |
> > comrel does not create any rule. QA does. That's called separation of |
| 12 |
> > powers. |
| 13 |
> |
| 14 |
> If you follow that logic, we end up with ComRel deciding to punish |
| 15 |
> people based on the private opinions of its members rather than |
| 16 |
> established rules. |
| 17 |
> |
| 18 |
|
| 19 |
So when the QA team votes to punish someone, how is that not "based on the |
| 20 |
private opinions of QA team members" and as opposed to "a set of |
| 21 |
established rules?" |
| 22 |
|
| 23 |
This is an important distinction. There is a recent article[0] about how |
| 24 |
Amazon has automated the firing of employees in their warehouses because |
| 25 |
they have a minimum quota (work / time) for workers and if you miss your |
| 26 |
quota too often, the computer terminates you. Many people don't like this |
| 27 |
because it removes a critical factor of human *judgement*; that the rules |
| 28 |
are not sufficient to judge every situation. There is context around each |
| 29 |
situation (worker was ill, had person issues that impacted their speed, has |
| 30 |
problems with other co-workers, etc) and so just "well bob didn't work fast |
| 31 |
enough" is not sufficient for termination. |
| 32 |
|
| 33 |
I'm suggesting that someone has to have this judgement. In Gentoo I think |
| 34 |
there are four judgments to be made: |
| 35 |
(0) Create the list of QA rules. This is firmly in the QA team's purview. |
| 36 |
(1) Was a rule broken? Gentoo work is more complex than Amazon's warehouse |
| 37 |
work, so this question is not as simple as "bob missed his picking quota by |
| 38 |
10% last month" because many QA violations have associated context and |
| 39 |
engineering tradeoffs (which is why humans are doing them.) So we cannot |
| 40 |
just automate this step. |
| 41 |
(2) Why was the rule broken? This is a question Amazon seems not to be |
| 42 |
asking according to the article (they just fire anyone who doesn't meet |
| 43 |
quota), but I suggest we do ask it because it helps us improve the rules |
| 44 |
and developer process. Understanding why violations happen help us prevent |
| 45 |
and mitigate them. |
| 46 |
(3) If a developer makes deliberate repeated mistakes over a time period, |
| 47 |
what do we do about it? |
| 48 |
|
| 49 |
I think 0-2 are clearly in the realm of the QA team to judge. I'm less |
| 50 |
convinced of (3) and this is the judgement I expect Comrel to be making |
| 51 |
because I believe this is not a technical problem. Ultimately if people |
| 52 |
refuse to follow the policies of the organization they should either get |
| 53 |
the policies modified, or leave. |
| 54 |
|
| 55 |
[0] |
| 56 |
https://www.businessinsider.com/amazon-system-automatically-fires-warehouse-workers-time-off-task-2019-4 |
| 57 |
|
| 58 |
|
| 59 |
> -- |
| 60 |
> Best regards, |
| 61 |
> Michał Górny |
| 62 |
> |
| 63 |
> |
Archives