Fork me on GitHub
#clojure-uk
<
2019-06-10
>
dominicm07:06:32

@lady3janepl I've started "deep work" this morning. On reflection I think that non distracted hours are non-negotiable. So I think the real conversation point is counter balance tools. Over the weekend I played with generating a who knows what knowledge graph, the idea being to direct conversations to the right people when you need help. But perhaps the tool should instead be focused on getting people to contribute their learnings so that company knowledge can be accumulated and reviewed?

👍 8
dominicm07:06:38

Also morning :)

Ben Hammond07:06:44

start crashing a bus into nodes of your knowledge graph

3Jane08:06:50

I love this 😄

Ben Hammond07:06:09

How quickly do we go back to the Dark Ages?

😂 4
Ben Hammond07:06:32

also 👋

yogidevbear08:06:48

Yeah, I think putting knowledge down in a central/shared location is very important. Main hurdle here though is reviewing everything on a fairly regular basis to make sure obsolete knowledge is removed/updated and emerging gaps are found/covered to avoid confusion over time. Documentation is hard 🤷

otfrom08:06:22

what is on wiki's mind?

dominicm08:06:20

I was actually thinking that the wiki format is perhaps broken

dominicm08:06:30

I was thinking something more like a log

otfrom08:06:35

I think they have trade offs

otfrom08:06:54

logs are good for recording your last thoughts. Wikis are good for recording the current consensus

otfrom08:06:16

wikis require editing. logs only require writing

otfrom08:06:50

we currently use an internal blog for a lot of our stuff, but I do miss having questions answered and things resolved and a clear statement of our current understanding

otfrom08:06:15

a log can also hide when there are major differences in opinion

otfrom08:06:31

tho having a wiki point at an "immutable log" might be a good half way house

dominicm08:06:21

Conflicting opinions seem like they should both be in logs and readers can decide based on that.

otfrom08:06:17

I'm saying that the logs can hide conflicting opinions and a lack of consensus and that there is no way for someone to decide from that (or even decide if there is still a lack of consensus)

3Jane08:06:06

incl. who creates the log (power over what goes into it, if it's not automatic), and who reads the log (reader bias for most recent records etc)

otfrom08:06:17

yep, power and resistance are inescapable. Generally, when I've started a wiki at a company it was an act of resistance. 😉

yogidevbear08:06:27

re wikis and current consensus

otfrom08:06:42

this is true in any documentation tho. Including logs.

yogidevbear08:06:15

100% agree 🙂

yogidevbear08:06:51

>Documentation is hard 🤷

3Jane08:06:20

wiki isn't my favourite, I miss something wiki-like with a graph navigation

3Jane08:06:17

word clouds reflecting word similarity are perhaps the closest

yogidevbear08:06:02

I suppose a lot of this also depends on the area/domain being documented

3Jane08:06:19

with people, you're always dealing with a graph though

3Jane08:06:22

having a graph of who knows what would allow you to see: 1) orphaned areas (and managing soon to be orphaned areas when someone is leaving) 2) people who are overloaded, people who have not been given enough responsibility 3) people who tend to overlap

dominicm08:06:32

I'm feeling like text analysis is an important property.

3Jane08:06:48

and then internally there are also networks of who counts on whose support, which is completely informal and cuts across the normal company structures

3Jane08:06:23

and also the last time I asked an HR person about this, she told me she: 1) studied it at uni 2) would never dream of attempting to map it inside a company because it's horribly invasive.

otfrom08:06:58

Quantum Black have been doing that for F1. They analyse email msgs

3Jane08:06:20

F1 ... formula 1? 😄

3Jane08:06:41

also: personally I find it incredibly cool! what are they doing with it?

otfrom08:06:02

yes, Formula 1. they have a case study. I've never been sure how much of this was just marketing and how much was actually useful. Got them a good acquisition price tho. https://quantumblack.com/work/auto-development/

alexlynham08:06:07

I did a research project that did something lightweight and similar

alexlynham08:06:21

it was… well, interesting in terms of how social groupings fell out

alexlynham08:06:29

the text mining part was past our remit though

alexlynham08:06:37

and the whole thing was super invasive regardless

otfrom08:06:50

oh, it is massively invasive and deeply problematic. It might be OK if you state up front and repeatedly that emails might be used for this. Then people could keep other private conversations out of their email.

alexlynham09:06:33

yeah I mean we had to get ICO oversight and all sorts, and it was a university research project, but I was super freaked out by it tbh

3Jane08:06:45

text analysis, for documentation?

dominicm08:06:18

Yeah, to infer the important topics from the document, and relay that into the graph.

3Jane08:06:46

Huh, that would be interesting

3Jane08:06:34

also for automation maybe something like: specify which repos / areas of code an article is responsible for, and automatically list out-of-date articles to be updated by their respective authors once that code was modified later than last modification date of article

3Jane08:06:54

hey, let's revolutionise the world of organisational knowledge management, in Clojure 😄

dominicm08:06:35

There's quite a lot of automation value we can get from analysing code. You can analyse my projects and realise that I've worked on lots of code using yada, and automatically highlight me as someone to ask.

dominicm08:06:23

I haven't looked at what is out there yet, because I don't want to be too influenced by other solutions :)

dominicm08:06:00

Back references from code to documentation seem important

dominicm08:06:35

I wonder if there's anything in recording processes? E.g. Recording the terminal session every time someone does a deploy means that the log is reasonably up to date. (Although slow to parse)

rickmoynihan08:06:37

This is part of the push towards chat ops. Wire things up so your log / system of record is slack/IRC etc. And wire up all ops tasks to bots either watching the channel(s) for commands or pushing the commands run to the channel(s).

otfrom08:06:17

I'm saying that the logs can hide conflicting opinions and a lack of consensus and that there is no way for someone to decide from that (or even decide if there is still a lack of consensus)

otfrom08:06:42
replied to a thread:re wikis and current consensus

this is true in any documentation tho. Including logs.

otfrom08:06:02

yes, Formula 1. they have a case study. I've never been sure how much of this was just marketing and how much was actually useful. Got them a good acquisition price tho. https://quantumblack.com/work/auto-development/

alexlynham08:06:39

my 2c - a lot of comms problems are intractable, hire good people, try and focus on comms skills aaaaand… it will all still fall over

alexlynham08:06:43

¯\(ツ)/¯

alexlynham08:06:48

that’s where good leadership comes in

otfrom08:06:30

@alex.lynham how do you see the good leadership as fixing the comms problems?

alexlynham08:06:49

trying to connect up conversations

alexlynham08:06:02

I mean, this is why I found doing that role unsustainable for my well-being

rickmoynihan08:06:05

Logs have the benefit that they happen automatically. However they can be impossibly hard to read and follow if you weren’t part of the conversation; (or sometimes even if you were). Real decisions and current thinking needs to be captured and curated. The logs are a fallback, and nice to have around… but they don’t really capture or communicate knowledge. Knowledge is best captured and communicated through good writing. However that’s expensive, which is why the logs are nice to have too… but they’re the cheapest and (usually) worst option for the reader 🙂

👍 4
3Jane08:06:42

Tragedy of the commons, unfortunately

alexlynham08:06:55

and @otfrom it’s worth saying that it means that you become a bus factor of 1 - not for knowledge but for behaviours if other people don’t see why you do as valuable, cc @lady3janepl and the ‘glue’ phenomena

👍 4
3Jane08:06:15

Knowledge sharing is something that benefits the entire team, but is not within most people's core responsibilities

3Jane08:06:25

(yeah, that, glue 🙂 )

alexlynham08:06:39

the number of times I stumbled into situations where two teams were diverging, and nobody above or below my level had noticed or cared was… insane

otfrom08:06:55

@lady3janepl don't get me started on the tragedy of the commons (and how it is a myth)

3Jane09:06:17

I'd be interested to hear more about how it is a myth

3Jane09:06:58

as it's opposite to my experience, I'd like to see what I missed 🙂

3Jane09:06:17

...but I need to work now >< I'll ping you in the evening 🙂

otfrom09:06:22

tl;dr the tragedy of the commons is just right wing propaganda to support privatisation and rent seeking

dharrigan08:06:04

A problem is that often people with the knowledge a. don't have time or b. can't be arsed, to inform others

alexlynham08:06:33

but that’s why for a good portion of my time at the old place I didn’t even get assigned a team, I just was left to use my judgement as to who needed help or who needed visibility… and a lot of what I was doing was just trying to help people find the right people to talk to, it was mad

otfrom09:06:37

@alex.lynham that sounds like a good way of frustrating someone

3Jane09:06:40

that's a valuable role, as long as you can get someone above you to recognise that it's a valuable role

otfrom09:06:51

that someone being whoever has to go around to all the teams and try to fix them

3Jane09:06:01

but ime large organisations simply fracture into vertical groups and there is not a sane way of bridging them. even smaller ones can do it.

otfrom09:06:10

did you at least get a good title like a coach so they knew they had to listen to you?

alexlynham09:06:47

all these things are also dependent on you wanting to do them, job titles aside 🙂

otfrom10:06:27

there is that. 🙂 At least if the title was "coach" then it would give an idea of what the job would be like and whether or not you'd like it.

alexlynham12:06:05

yeah I think it was one of those where nobody knows they need it until they need it

alexlynham12:06:18

if you’re that shape of person you end up doing it :man-shrugging:

otfrom13:06:54

been there. done that. 🙂

alexlynham14:06:54

I suspect ’tis my lot in life, in the long run

3Jane09:06:19

also need to run >< see y'all later o/

rickmoynihan09:06:30

I think all things being equal in a large organisation; organising into verticals tends to be better than organising into horizontals - it’ll lead to more duplication of effort; but less dependencies. Also some would argue the ideal is a balance between chaos and order. Dee Hock’s chaordic “organisation”.

otfrom10:06:39

@rickmoynihan ah, the return of silos. There are some real advantages to it. There are some real downsides too.

rickmoynihan10:06:44

oh absolutely — though the same can be said of being purely horizontal. I’m really only talking about a hypothetical case of pure vertical vs pure horizontal (if such a thing ever existed). My real preference as with most things is to think hard about it, and not to treat the differences like a religion 🙂.

seancorfield18:06:02

Macromedia was kind of interesting in that regard: it was mostly organized into silos but had three cross-cutting teams that work with all the siloes.

seancorfield18:06:07

Each product team was autonomous but there was an overall production strategy team that was also technically deep and it worked with all the teams on common architecture, infrastructure, shared components, etc. There was also an overall experience design team, that cross-cut over everything Macromedia exposed to the public. And then there was my architecture team that worked across everything outside of products (all the back office systems for the entire company) but also with the product teams whenever they needed to interact with IT/company infrastructure. Such as the "news & community* widget that was part of most products' UI back in the early Studio days.

👍 8