Home Assistant for the whole household #40
Replies: 4 comments 4 replies
|
We need to organize them around who needs to see what. Yesss I also mentioned in person that these questions make for great Acceptance Criteria from a Usability point of view:
An open question/thought i have that i'm noting but don't expect to be addressed here
I think a lot about this book: https://yalebooks.co.uk/book/9780300229622/the-private-is-political/ which is more about social media but some of the insights and case studies could apply to a home situation. |
I really like that idea!
Interesting question, as it has relevance with the problem described above. If the admin controls the system, any data or logs inside HA are effectively controlled by the person the resident may need to build a case against. An extra reason to give residents access to their own data. |
|
I opened Discussion #53 which is an extension of the second point raised,
Which is the first push to begin the establishment of major UX writing tasks, which will in theory help to resolve this issue in some way! |
|
Hey @matthiasdebaat, thanks a lot for bringing this up publicly! Indeed, our research involving passenger users in HA-driven homes also shows that maintainer-centric interface architectures create somewhat odd barriers, even when residents are eager to increase their involvement. A few comments on the issues mentioned:
I'd like to challenge this one - exposing affordances while being sparse with the "whys" and "hows" is indeed efficient, but only as long as the system operates as desired. However, once the system's behaviour becomes confusing/discomforting, Residents will look for some explanation of the odd behaviour, and this is where exposure of the devices and related automations plays a major role. Even if the resolution would require the intervention of the Maintainer, the ability to locate the issue (and possibly isolate the affected part of the ecosystem) is a major component of the perceived sense of control. Our research indicates that this prolonged confusion of "sth is wrong but idk why and what else is affected" can easily feed all types of withdrawal reactions, as a result, increasing dependency on the maintainers. Really happy to engage with this vision, and I'm looking forward to seeing it evolve. Fingers crossed. |
|
Awesome that you joined this conversation @mpwozniak! Thanks for your research and insights. Good that you point out the need for residents to understand how Home Assistant works to have a sense of control. Last thing I want is to have the withdrawal reactions from residents. As Home Assistant is often described as a system with a steep learning curve, it is a design challenge to find the balance between providing the right information at the right time. And also gain the confidence of maintainers, when I speak to them I often hear that they find that their residents don't feel the need to learn how the system operates. What would your advice be for designers? |
|
I think many maintainers say that because they involved their families in the design and planning stages of the rollout of their smart home and their spouses legitimately are not interested in learning anything more. Would this change benefit them? Probably not, but at the same time it does have benefits for those that want more knowledge of the inner workings of Home Assistant. |
|
Apologies for coming to this conversation so late. I wanted to add one related thread that may be useful here. I think the Privacy Preference Declaration architecture could be relevant to the agency and awareness problems this discussion is raising. PPD is not a UI pattern by itself, but it creates a structured way for devices and services to express what data they collect, why they use it, where it goes, and what household privacy preferences apply. That kind of machine-readable layer could help Home Assistant distinguish between “what the maintainer needs to configure” and “what residents need to understand, consent to, or override.” One thing my research, and privacy/usability/transparency research in IoT more broadly, has consistently reinforced is that privacy management and detailed configuration are secondary tasks. People mainly want to use things. Power users can be different, but even then, configuration management is usually in service of the primary goal rather than the goal itself. A technique from my paper, The Best of Both Worlds, might be useful here. I wrote this a long time ago, and it was centered around Android permissions, but the techniques we employed there could generalize here too. In that work, we looked at reducing privacy configuration burden by crowdsourcing common preferences, clustering people into representative profiles, and using decision trees to ask a very small number of questions before recommending settings. A similar approach could potentially help Home Assistant offer preconfigured recommendations for residents and maintainers: dashboard layouts, shared-vs-personal visibility, automation transparency levels, and potentially more detailed privacy options (and really, options more generally, they need not be privacy specific) when they are well-defined and enough preference data can be collected on them. The important point is that this can work for novices and power users alike. Novices get usable defaults and a small number of meaningful questions. Power users still get the ability to inspect and change the details. That seems aligned with the direction here: more agency and awareness for residents, without turning the smart home into another configuration project they never asked for. |
|
Thanks for this, useful addition. Good point about privacy management being a secondary task. I like the profile idea of easily setting privacy defaults, and it adds an axis we do not have yet. We only describe roles, which are about what you are allowed to do and are assigned by someone else. Profiles give the resident easy control over their own privacy. Instead of, what we have documented now, the phone app permissions pattern, choosing once, always or never. Would love to ideate on PPD and what we can do in Home Assistant. For example, we know what the device can do inside Home Assistant, and we could let the instance say what it allows, and show this to the user. A limitation would be devices that have a local connection to Home Assistant, yet speak to their own cloud. Today we can't see that, though sitting inside the home means we could. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
TL;DR
Two roles: maintainers (who build and configure the system) and residents (who just live with it). Right now, residents either can't see what the system is doing, or they can see things they shouldn't. Their personal data like location defaults to being controlled by the maintainer, often without their knowledge.
Three concrete changes follow from this: resident data should default to private, devices and automations should move into the shared space so residents can see what the home is doing, and residents should be able to create their own dashboards.
Open questions include override controls for residents, feedback tools, calibrated access per person, and what to do when someone doesn't have the app.
The framing is evolutionary, not a rewrite. Nothing existing breaks.
The problem
Home Assistant started as a tool for technical enthusiasts. It is growing into a platform for the whole household. That shift means we can no longer organize things by how they work under the hood. We need to organize them around who needs to see what.
Smart homes are social systems before they are technical systems. Organizing them around how they work under the hood ignores the people who actually live with them. The real question is who needs to see, and control, what. The goal is to give residents a clear, honest, and reliable view of their smart home, while keeping the full configurability that maintainers need. This is not about simplifying Home Assistant or hiding what makes it powerful. Nothing becomes less configurable.
There is a name for what happens when a smart home becomes psychologically opaque: haunted house syndrome. Lights change without explanation. The climate adjusts mysteriously. Automations interact in ways nobody planned. The maintainer understands the system, but everyone else experiences it as unpredictable magic. Once a home starts feeling that way, trust collapses fast. Adding to the problem is the fact that in many households, the people who are not maintaining it simply never interact with Home Assistant at all.
Even a mature Home Assistant installation can turn into a haunted house over time. Abandoned automations that still fire. Entities with names that made sense two years ago. Dashboards nobody maintains but everyone sees. Helpers that were temporary fixes. The system works, mostly, but it has become hard to understand or update its logic. Transparency alone does not fix that. A chaotic system that adds visibility, is still chaotic.
Right now, Home Assistant is both too hidden and too exposed at the same time. That's a strange combination, but it creates three friction points that the platform has not yet fully solved:
The sweet spot
A completely invisible system leaves residents in the dark about what's happening and why. Full transparency all at once is overwhelming. People who live in a smart home just want to feel in control of it. Ideally they do not even have to think about what is happening in the background. The goal is not pursuing transparency so dogmatically that we overwhelm residents. Rather, it is to fine-tune how information is presented so that it is transparent and usable.
In this way, a home won't have to constantly explain itself. In fact, the home itself won't break into a resident's conscious experience except when absolutely necessary. Nobody thinks about what happens in the background when they turn on a light. Living in a smart home should feel the same way.
The sweet spot sits between those extremes, one where maintainers see what they need to configure the system, and residents what they need to live in and interact with it comfortably. We aim for:
What transparency means
When we talk about transparency here, we mean two things.
Both matter. Knowing the system exists but not understanding how it works and what it is doing might leave residents feeling helpless. Not knowing it exists at all means residents never think to question it, shape it, or make it their own.
Two roles, four spaces
This reorganization of how Home Assistant is structured is built around two core roles.
Maintainer
The person who sets up Home Assistant, manages it, or takes on both responsibilities. This includes the initial installation as well as ongoing configuration such as adding integrations, creating automations, and managing the overall smart home system. This role covers both the hands-on installer, who may be a hired professional setting up the system, and the household member who takes on ongoing admin responsibility afterward. In most households these are the same person, and a household can have more than one maintainer. Maintainers are often doing invisible labor for the household, setting up automations, managing integrations, keeping everything running.
Resident
Anyone who lives in or uses the smart home. A partner. A child. A roommate. An aging parent.
It can be someone who never asked for a smart home but now lives inside one. Or someone who shapes it every day and is the maintainer of the home. Research confirms these two roles emerge naturally in households.
What about non-residents?
Not everyone who enters a smart home lives there. A guest, a visiting family member, a repair worker. These are non-residents. They interact with the home in one of three ways.
Both cases need deliberate design choices from the maintainer.
The four spaces
The four spaces give us a way to talk about something that is hard to talk about. Shared, Admin, Personal, Unresolved. In practice the lines blur. Some automations are both personal and shared. Some data is individual in origin but affects the whole household in practice. For example, a person's location triggering the heating. Or a sleep state affecting the morning routine. The names are a starting point, not a strict classification.
The Unresolved space is the exception. It is not a design target, it is what happens when the maintainer is no longer available and no one has taken over. It exists to name the problem honestly. Designing to minimize it is part of the framework too.
The shifts
Three things need to change in order to make the home more equitable and intuitive to live in.
1. Data autonomy
Right now, sensitive data flows to the maintainer by default. Many residents aren't aware of this, and others have only limited ways to change the status quo anyway.
That should flip. Resident data starts in the Personal space by default. The maintainer cannot access or use it without the resident's consent, similar to how you grant permissions to apps on your phone, choosing once, always, or never.
Some data is clearly personal. A phone and its location. A sleep pattern. A presence sensor tied to one person. Even a room can be a personal space, with all the devices in it belonging to whoever lives there. The boundaries are not always obvious, which is why good defaults matter. A well-designed starting point means residents are protected without having to think about it, and maintainers can still do what they need to do. A maintainer can know that someone is home to trigger an automation, without knowing which resident it is or where the others are. If more detail is needed, the maintainer can request access from the resident, with a clear explanation of why.
In practice, a resident can see exactly what data exists about them and what the system does with it. They now have visibility into how their personal information is accessed and real control over whether to grant that access. Only the resident can move data to Shared, making it visible to everyone in the household. Clear indicators show what is being shared and what is not. Granular sharing between individual residents is a future consideration.
2. Transparency of system logic
Right now, devices, automations, and floors/areas live in Settings. If you are not the maintainer, you have no idea what the smart home contains or what it is doing.
That changes. Devices and automations move into the Shared space, so residents can see what the smart home contains and what it is doing. For example:
Nothing is taken away from the maintainer. They still build and configure everything. What changes is that everyone else in the household can finally see what has been built.
3. Dashboards
Right now, the maintainer decides what dashboards exist and who sees them. Residents have little say.
That changes too. There's a Shared directory of all available dashboards. Residents can pin or create their own Personal dashboards from there. As both spaces grow, the current sidebar won't scale. The community already knows this. A popular custom sidebar exists precisely because people want more control over what they see and how it is organised. A navigation model where pinned items stay visible and everything else collapses behind a More button builds on those learnings. Shared and Personal items can coexist without either overwhelming the other.
Future considerations
This framework raises questions we don't have answers to yet. They're worth naming.
Manual override automations
Residents should always be able to step in and override an automation, even if they can't edit it. People are much more comfortable with automations if they know they can take back control at any moment. What that looks like for residents without Admin access is something we still need to figure out. One direction worth exploring: a simple way for residents to flag an automation as confusing or broken, temporarily disabling it until the maintainer reviews it.
But the questions go further than a simple pause button. Can a resident ask why something happened? Can they see automation history? Can they distinguish a manual change from an automated one? Can they report confusing behavior to the maintainer?
Home Assistant, with its depth and community, is better placed than anyone to get this right. Apple Home and Google Home have not gone there yet.
Reporting and feedback
The conversation between residents and maintainers already happens. A partner mentions the heating felt off last night. A child asks why the lights keep turning on. A parent calls to say something is not working the way they expect. But the tools to support that conversation inside Home Assistant don't exist.
Think of a maintainer managing a system remotely for an aging parent. When something behaves unexpectedly, the resident has to remember to report it from memory. The maintainer has to dig through logs and traces to piece together what happened. A simple way to flag something in the moment, a button that says something felt wrong, would close that gap. The conversation is already happening. Home Assistant could make it much easier.
Calibrated access
Not every household needs the same level of transparency, and that can change over time. A maintainer should be able to adjust how much of the Shared space is visible to different residents and non-residents. In some cases, a specific resident may need limited access for safety reasons or because of their capacity, for example a young child, someone with dementia, or a person with a cognitive condition. That resident is still a resident: they have a Personal space, they can see the Shared space, but some controls aren't available to them. Crucially, they always know what they can and can't access, to the degree that is possible for them. Those edge cases require extra care in design and are part of the open questions around calibrated access.
Needs also change over time. A teenager gets more access as they grow up. A partner who was never interested starts wanting to understand the system after a few years. And when the primary maintainer becomes unavailable through separation, illness, or just moving out, whoever is left behind should not be locked out of their own home.
This also raises the question of granular sharing between residents, for example whether someone can share their location with one person but not the whole household. That too is an open question.
Reaching beyond the app
This framework assumes people have the app. But many don't. A resident who never installed it, or a non-resident encountering the home for the first time, can't rely on a screen. For residents, that means better physical touchpoints and better ways to engage with the home without needing a phone in hand. For non-residents, it means presence awareness: the home should signal that it's smart in a simple, clear way, without requiring any prior knowledge.
Mapping the four spaces
Every feature, setting, and data point in Home Assistant sits somewhere. The question is whether it sits in the right place.
We've mapped where everything should go using these four spaces. Every screen, every setting, every piece of data, placed into one of the four quadrants. The map shows what a reorganised Home Assistant could look like, and surfaces the conversations we need to have about what moves, what stays, and what gets rethought entirely.
Link to Figjam
The map breaks down into these groups:
Getting there
This is not a reset of Home Assistant. It is a layering of new household-facing structures on top of the flexibility that already exists. Existing setups remain valid. Advanced configurations stay first-class. Nothing breaks.
The path is evolutionary, not revolutionary. New visibility layers can be generated from existing automation structures without requiring manual effort. Residents who never install the Companion App still benefit from a maintainer who organizes the system with them in mind. Households adopt at their own pace.
The hardest part is not the architecture. It is the hybrid states along the way: old dashboards sitting next to new visibility models, legacy automations without descriptions, partial adoption across a household. Those in-between states may persist for years. That is fine. The goal is not a perfect migration. It is a direction everyone can move toward without leaving their setup behind.
For contributors working on features, settings, or how Home Assistant is organized, these four questions can help evaluate where something belongs in the four spaces.
Further reading
The ideas in this document draw on research into smart home use, privacy, and interface design. These sources are listed here for anyone who wants to explore the underlying work.
Show table
All reactions