We really need more DOTS roadmap information

I hope Unity is preparing to announce more long-term plans for DOTS in 2021.

We’re going on four years since DOTS was first announced, and tremendous progress had been made. No one has ever attempted this before at this scale, and the potential could change the industry.

But we don’t have a good idea of when ECS 1.0 might ship. Or when DOTS-based, skinned animation will be possible in a performant way.

Lately, we’ve been trying to make our best guesses about when those things might happen, to plan our next steps as a company. Our hopeful estimate is ~5 years from now, before we might expect to start a new DOTS project, and make a shippable, single player game over a 2-3 year dev cycle, with animated elements, and without major rewrites to keep up with major deprecations.

Without that information, it’s hard to know when our financial and time investment in DOTS (vs other engines or technologies) might pay off. 2.5 vs 5 years off is a huge difference for devs who need to plan how to keep the lights on. If we knew it was a long way off - that wouldn’t be bad. In that case we could pivot to shipping our next few titles with the traditional GameObject tech, and stay afloat until DOTS is more ready. But what we can’t afford (and I imagine lots of other devs can’t afford), is complete uncertainty. In this case, not knowing is the same as knowing that it’s unreliable.

So If that stable development reality is 5 years from today? If the needs have gotten more complex, and 1.0 is just taking more time? That’s fine. It will still be groundbreaking tech that no other engine has. But since it’s been nearly four years, Unity should give us more of an idea. When you get on stage and say “This tech we’ve been working on with revolutionize game development”, your customers don’t expect the end of that sentence to be “...in ten years!”.

So, if you realize one day that it’s going to be that long, ok. We trust your reasons. But you’ve got to let us know. There are mouths to feed.

With what Unity is working on to show off later this year, I hope they will also announce a clearer DOTS timeline.

 

Hence my reaction earlier. It seems they're getting somewhere "stable" (I hope), but it is a little frustrating to not get enough details on the development here.

I am extremely on the fence right now on whether to adopt the full ECS-approach as much as possible for the project I will be starting from next month, or if I should just stick with the Job System + Burst for now as it currently works really well with the rather unfortunate legacy code I have that is simpler to work with due to familiarity. It's not directly gamedev-related, but it will tie directly into gamedev.

 

Unity has gone out to say (summarizing) that scope has changed, their original roadmaps are no longer accurate, and that DOTS 1.0 will be ready when its ready. Unity has been consistently bad with roadmaps and estimates that are reliable, and I think they learned their lesson and now are no longer providing them. If Unity were to make such a roadmap, making decisions based on it would be irresponsible. You should be making decisions based on the current state and the currently perceived velocity in the last few releases.

What I do find frustrating is not knowing what the developers are actively working on that is slated for the next release or two. If I need 5 features in the next two years, and one of those features may be provided by Unity in the next release or two, I would love to deprioritize that feature until Unity releases their version then evaluate it. If the feature ends up being good enough, I'm now four months ahead of schedule! If not, I haven't lost anything. I was extremely paranoid during the three months I was developing my audio solution that Unity was going to release their own during that time frame.

 

I’m not angry at Unity - I’m grateful that they are taking on this difficult work. What is described below is unfortunately just a professional reality.

After asking people to pay for your platform on the notion that this tech is coming relatively soon, after 4 years, an appropriate response to this situation would be to provide a more accurate roadmap. Along with a good reason for customers to trust the new roadmap.

Saying the previous roadmaps are unreliable, and then not releasing anything doesn’t help the customers who invested in Unity’s ecosystem four years ago on the promise of this advertised tech. This is a business.

The consequence of releasing no new roadmap data within the new few months will be losing customers, and customer trust. Promising, new tech is truly amazing. But customer trust is even more important. Unfortunately, when there are children to feed, waiting with no forecast cannot be a choice.

This is not an angry exit interview. I sincerely believe in the promise of Unity, and don’t want to go anywhere. My hope is that later this year, Unity will release a more reliable roadmap for DOTS, regardless of how far out it reaches.

 

You say the following:

If there are children to feed that are not getting fed, wouldn't an appropriate response be to create a technology that could automatically create food for them?

It sounds ridiculous, right?

But you might say, "Well these other companies have successful long-term roadmaps." But yet that ridiculous technology that automatically creates food is something that is coming closer to reality. How do I know? Because that's the industry I work in and I have a direct impact on making that become reality. But does that mean your company could create the technology? Absolutely not! We have decades of experience and expertise in the industry that have allowed us to push the technology this far.

The reality is Unity does not have the experience to provide accurate long-term roadmaps for bleeding-edge features for a game engine used by thousands of professionals. Therefore the appropriate response is not to "try again" and risk misleading clients even more, but instead not make statements (that people tend to treat as promises despite disclaimers stating otherwise) that they may not be able to adhere to and instead provide information that is reliable about the current state of the product today and its longevity. In addition, keeping quiet about the roadmap allows them to listen to users and adjust on the fly, allowing them to more effectively meet our use cases and concerns. And they are doing this. I am 100% confident of that based on real evidence I have observed. And for me, that is way more valuable than a 1-2 year roadmap of features that probably won't work for me without user feedback anyways.

I hope they don't release anything greater than 3-6 months out.

I'm not saying the mistake Unity made 2-3 years ago with their roadmaps wasn't real and damaging. It most certainly was. And I empathize with the frustration. It is valid, and I don't judge you for it.

But to everyone reading this, please stop complaining that "DOTS isn't ready like they said it would three years ago" and asking "when will DOTS be ready?"

Instead, discuss the actual pain points and problems you are having. Such conversations are far more productive and have historically resulted in quicker resolutions. Quicker resolutions and more intelligent decision-making based on facts rather than speculation (a roadmap is just speculation since there is no guarantee there won't be problems) will result in less starving children, and I really don't want to see more children starve.

 

They can take all the time they want. They are bad with roadmaps, no more roadmaps <- That's fine. Use it when it is ready, not before, not count on it being ready before it is. Act as if it doesn't exist before it's ready.

All the above would be totally fine with me if in the meantime they hadn't semi-abandoned all other parts of Unity.

"This is slow" "Oh yeah, it will be faster in DOTS"
"This is kinda bad" "That feature is in maintenance mode until we have a replacement in DOTS"

/\ This is the part that makes it not okay.

PS: (also going with Unity's track record with all previously released features, DOTS will be released during some conference that's overhyping it, way before it's ready, it will be kinda okay / eh, and then will be instantly abandoned)

 

It might be worth saying, that arguments in this thread won’t have an impact on our business decisions. I didn’t write these posts to win an argument. I don’t want to prove anyone here wrong. It’s just our business reality. There’s not really much point to arguing about this. Please don’t do that here.

I wrote this post in hopes that someone at a unity will see it. Please know that we love working with your engine, but as a company, we adopted it based on what you advertised about DOTS. If DOTS wont be ready in the ways we need for another several years, please let us know. We don’t have the financial option to proceed with Unity, having no way to plan about when it might be ready.

But if we know when that date might be, then we can afford to stay with the engine and invest in the ecosystem in other ways, until it’s ready.

 

This is a bad plan. Even if someone from Unity gave you their honest best current estimate right now, it still will probably be completely and utterly wrong.

Also Unity has a bad track record with v1.0 releases.

Don’t make business plans around features that aren’t ready.

 

If Unity is the type of company that will advertise new features like they are coming soon, and then go dark when they realize they won’t be, then we may have to abandon it.

I don’t believe that is the case. I think Unity is made up of professionals who understand what kind of information their working customers need to make business decisions. Unity itself could not rely on a business partner who started talking about a new product 4 years ago, and then today had no information about when Unity might be able to use it. They would have to consider going elsewhere.

My hope is that they release a more conservative timeline sometime in the next several months - even if it’s only an internal target - so that their customers can make tentative plans around the products they have advertised for about four years.

 

I believe they are a much worse company than that, but that just may be me.

Having been in the audience of a Unite where they showed a bunch of shiny features, the sentiment in the audience was “yeah, none of these things they are showing us is going to be usable any time soon”. But again that could just be me and my immediate surroundings.

 

Care to share recent examples (recent as in the last 8 months)? If anything, this will provide the DOTS Team awareness of what is being said so that they can go scream at the non-DOTS devs telling clients BS.

Yes. And if that track record is sketchy, then consider going elsewhere. I'm in full agreement with you that they screwed up 2017-2019. But my argument is that they are no longer making that mistake today because they are no longer advertising DOTS whatsoever (well other than letting studios using it chat about it on live streams).

The forums would be a good place for this, stickied and with updates on progress and complications. Most of these internal targets are within that short-term timeframe anyways.

How would this info be helpful for planning? It sounds ridiculous every time you say it, but maybe we are overlooking something?

Not just you.

 

If estimates are a problem, we might get, at least, the prioritized feature list. Updating this one shouldn't be much of an issue

 

I don't expect any updates this year. They say that they are still planning on releasing ECS 1.0 in the future. Hoping for a update in 2022.x. The core part of ECS feels like it is not too far away from 1.0.

 

There has to be some sort of middle ground between an all out marketing blitz to a general radio silence(and reading the occasional forum post by a dev is like watching a tumbleweed roll by, its exciting but only because nothing else appears to be happening).

I would love to see the dots packages developed on github, much like how the renderpipelines have been developed in the open. This afaik has never been acknowledged when asked in the past and I don't understand why. Trade secrets? Too many moving parts? Could we get a reason why this is all done internally when the package source is all out in the open?

 

AFAIK they've been meaning to move a lot of the package stuff to be internal as well (and they almost did).

Not sure if it's still the plan.

 

I always just assumed they where switching to Plastic SCM after the buyout.

 

What happened to the math-library? It seems last update was 13 days ago per this post:
https://github.com/Unity-Technologies/Unity.Mathematics