# Sustaining and fortifying OSM

## *2 most urgent* operational and infrastructure needs (include verbs :)

> [name=Allan]
> 1. Need to meet increased demand for OSM data, including enforcement of tile use policy.
> 2. Bolster sysadmins and OWG to ensure infrastructure doesn't fail.

> [name=Guillaume]
> 1. Tile servers are slow and overloaded, and maintaining them is taking lots of sysadmin tiles. Do we want to continue offering a tile service as usual? If we cut people off, on what criteria? Some people are donating tile servers based on the assumption that their local community will be allowed to use tiles.
> 2. Unfreezing OWG, helping it get stuff done again

> [name=Joost]
> 1. Tile serving - we need to have a policy where free as in free beer serving does not break the mapper experience
> 2. Expanding the capactity of sysadmins/OWG in more general terms

> [name=Mikel]
> * Tiles services policy action in order to reduce over-demand on tile services
> * Help sysadmin

> [name=Paul]
> * How to get API keys on secondary services like tiles & nominatim
> * Who is in charge of iD and what software is on osm.org

> [name=Rory]
> * Tile servers are slow, (when you go to osm.org, the slippy map tiles take a long time to load. They should appear faster) talk/decide about how to address it (paid staff? pay for server resources? Ask for resources? block access?)
> * iD & what software is on osm.org. Decide what software should be on osm.org.

> [name=Tobias]
> * Build a non-hostile relationship between iD and the community
> * Find & promode routing services that support established OSM tagging

DISCUSSION

## Tile services/servers

### What is the problem

* A secondary service has more operational load on admins than primary services (planet.osm, editing parts of site) https://operations.osmfoundation.org/policies/classification/

* We don't know who is using services and to what extent
* Can't reliably contact heavy users

* There is no technical way to regulate access to services in a substantial way. Any restriction would be trivial to moderately hard to bypass.

* Nonenforcement of tile service policy.

* Sysadmins don't want to cut off people without a mandate - they would be between a rock and a hard place

### Notes

* Operational load - Investigate solutions, e.g.
* Define associated policies and thresholds,
* Improve enforcement of responsibilities of heavy users
* Consider/scope API keys

* Tile policy
* Need better technical means, industry standards, libraries that use tiles support API keys
* Could volunteers help wth monitoring heavy users?
* Sysadmin dedicated to tiles
* Tile CDN costing would be $5M-$10M USD at market rates, based on preliminary research
* Better document setting up tile server/building a business on same?
* Don't compete with others offering services
* Better policy, data analysis, and organizing enforcement might avoid radical changes
* Consider 2 tiers of tile serving? Paul: makes system vastly more complicated

### Decisions/action items

* Talk to sysadmins
* update action items from there

## iD

### What is the problem

* Many stakeholders are ignored by iD development, and are unhappy.
* iD (like most OSM software) has no formal governance
* We have control over what is deployed on the website, unlike JOSM
* Some of iD's tagging recommendations are poorly thought out and/or at odds with recommendations in the wiki and other stakeholders.
* Bryan's style (and the style of the rest of the OSM community) have often collided. There's even a wiki page: https://wiki.openstreetmap.org/wiki/ID/Controversial\_Decisions
* Quincy (the new maintainer) is upset because his friend and mentor got kicked out. He feels like he's being bossed around by the OSMF board, and probably Guillaume in particular.
* Tagging is a mess, and iD is forced to manage and take decisions
* iD has been a target of criticism since its creation
* An editor inevitably attracts flak, the 'default' even more, especially aimed at new users.
* iD devs pick between supporting standard tagging and what they want case-by-case
* iD is promoting community platforms based (originally) on the developers' preferences.
* iD prioritizes good usability and that's not a shared value among community
* There's a conspiracy that iD enforces things Mapbox wants

### Notes

* Look at this in a software governance, do we need structure, e.g. "liquid democracy", is this right for a test of that?
* concern about limiting or constraining developers
* OSMF decisions not necessarily considered binding
* Doesn't need to apply across all software; need to leave creativity and innovation in place
* Could address governance issues via forking
* What are expectations for software that touches core of projec
* What should role of EWG be in all this?
* Maintainers are in tough spot, have a need for some governance

### Decisions/action items

* Directly engage maintainer (Allan, then assign next action item)
* Convene discussion among stakeholders in ID
* TBD: Define "stakeholders"

## Supporting sysadmins

### What is the problem

* Too few people +1
* We have no direct communication (+2)
* Distrust between sysadmins and Board
* No consistent picture of what sysadmins work on and how much
* No idea what sysadmins want to happen. Sysadmins enjoy working on the systems more than managing communication with the board
* Focus on "just" keeping things running, when the community wants so much more. Fundamental conflict between progress and stability.
* Misunderstanding on what sysadmins do
* The current sysadmin team is a bottleneck/gatekeeper for stuff to happen. They are essential for some things, but they could probably delegate some of the boring stuff to someone else
* We talk about "sysadmins", but this is really about 2 individuals: Grant and Tom

### Notes/decisions/action items

*

### What do we need to discuss with sysadmins

* What would be useful outcomes of this conversation for you?
* How would you ideally want to spend your time on OSM sysadmin stuff? What stuff do you currently do that you don't enjoy doing or find boring?
* Very roughly, how time do you spend on sysadmin stuff every quarter, and on what? What other things do you do in OSM outside of sysadmin?

* How do we recruit more sysadmins to share the workload? Do we need to hire staff you would supervise? Would you want to supervise people at all? Should we find a project manager/assistant? Move to the cloud (be specific in the question, that's a wide term)? Another solution?
* Are you for or against the idea of paid sysadmin help and why? Permanently, or consulting/project? If yes, how would you want to relate to them? Would someome getting paid for that work, cause you to stop/quit?
* Would you be interested in being a full-time (or part time?) paid sysadmin?
* What can the Board do more of to build your trust, and keep an open line of communication?
* Do you need hardware/software/consulting resources beyond what the OWG asked for in its budget? Would you like the board to look into that?
* What does OSM infrastructure ideally look like and what kinds of things would it do?
* Would you like us to support efforts on CI/CD/testing? How?
* How much time on tiles relative to other stuff? Would a (different, wider) mandate help?

* Set expectations: Tom isn't someone who talks a lot :). He's afraid he'll be bored in our meeting, because meetings are boring. *(If only we could have this conversation in a pub over a pint!)*

https://hackmd.io/mpbmMgTlQTqVRNrS7XxLhQ

## Disposition on paid staffing

### Notes

* Paid staffing comes with risk, where does money come from?
* Companies are willing to invest substantial monies
* Concern of big donors expecting to have substantial control
* In terms of influence, we need to discuss and we need to tap and learn from other FLOSS communities
* Pursue diversified strategy, pursue other sources of money
* Concern about moving on this before talking to sysadmins
* Need a structured sense of OSM in 2-5 years from now
* Not just sysadmins, also seeing strain on finance side, banking, might have to move Foundation to another country, it's a lot of work, conversation should be broader

### Action items

* Scope out possible scenarios for possible paid staff, including risks
* Personnel committee:
* structured way of working with Dorothea
* Concern about confidentiality /HR issues
* Scoping out boundaries on committee, safety measures, how to mitigate effects on volunteers, e.g. de-motivation
* Who to take this forward: Allan, Mikel (ACTION OWNER), Joost, Tobias

* Scope out fundraising structure
* Committee? wg?
* Need to figure out to manage and spend money
* Need to do research on strategies for increasing revenue
* Guillaume and Allan to move forward on proposal