Ashley's Forum Posts

  • I just checked, and... I was wrong, Sprite(N).X does index within IID ignoring all the picked instances. Oops.

    I think that is the wrong decision and the feature should not work like that, but it's probably too late to change now for backwards compatibility reasons. This is kind of a mess. I'd rather not touch it, and rather not add any more features that extend the mistake any further.

  • This is not to do with the signalling server, it's to do with the peer-to-peer connections between devices. See the section on connectivity in the multiplayer tutorial.

  • We have to work in the constraints of the platform we use, and they don't let us limit the number of submissions someone can make. So I thought we would give people 10 votes and state an official policy that ideas with zero votes will not be considered, in order to discourage any one user from inundating us with loads of suggestions.

    I do want to make people choose with a limited vote count. If you have 50 votes, you can vote on 40 things that aren't really that important but you kinda like, and 10 things that really matter. Given our limited resources and how many suggestions we face, I want to force people to think carefully and only choose the top 10 things that really matter to them. So then every vote counts and we get a better impression of what people really really want, rather than a bunch of popular nice-to-haves. Features that make something impossible become possible are usually more worthwhile than features that make something already possible a bit easier, so I think this would help focus attention on the former.

    I went through the minor suggestions category and most of them weren't minor. They were either actually more significant projects, or were unclear enough that it would take at least a fair bit of discussion about the idea, the motives, the alternatives, and how it would really work, that the discussion alone means it's not a quick thing to do. This pretty much confirms in my mind that it's not really useful for anyone to categorise something as a "minor suggestion" - it's rare that it really is quick, and most of the time it's much more complicated than anyone imagines. This is another reason I think we should limit votes, since the idea you can vote on a big pile of minor suggestions that are quick and easy is wrong, since chances are all those votes are actually going on big projects, complicated or unclear ideas, or things that will turn up unexpected complications during development.

    I think I'm going to have to just make a decision and go with it, so some time in the next few weeks I'll look in to setting up a brand-new suggestions platform to run for 6 months from scratch, probably from June to the end of the year, with limited votes. I'll try to make sure the old one is still accessible but read-only, since that seems better than deleting loads of ideas and discussion, and we might want to be able to reference back to it even after starting over. I know that maybe not everyone will like it, but I think there's a chance it could actually work much better overall. Let's treat it as a bit of an experiment and see how it goes. If everyone really hates it at the end of the 6 month period we can always try something else, we don't have to stick to one approach forever.

  • I think reddit or Trello will meet the same fate of being inundated with an impossible amount of work, if there's no limit on the number of suggestions you can post. To avoid the same fate there has to be some kind of restriction on how many posts people can make or how long they remain for.

    We don't really do any formal long-term planning ourselves, so I don't want to share a roadmap that either doesn't specify anything in particular, or commit to ideas that we later delay or cancel and then have to deal with all the backlash from users who were expecting it. As it is we have a lot of flexibility which is good for us as a team, especially since we can easily respond to changing circumstances, and it avoids risking broken promises to our community.

    I just went through a bunch of suggestions on the existing platform. I really struggle to know what to respond to some suggestions. A lot fall in to a kind of "I guess we could do that, I don't really know how hard it'd really be, and there's a lot of other ideas with a similar number of votes and I'm not sure what the rationale for picking this over all the others would be". I don't think it would be useful to say that. A lot of responses are also not simply a statement but a question clarifying some aspect or asking for more details, and often become fairly long and involved discussions with complicated trade-offs that need careful consideration. It also took ages and I got through only a fraction of the total number of ideas on there, and I'd need to repeat much of that, going over the same ideas, to examine answers to questions and continue discussions. I really think we should make it clear that we can't be expected to provide a response to every idea. I think we have to let the votes speak for themselves and leave the responsibility of making clear and detailed proposals up to the user who submits it. All too often it takes a lot of questioning and clarification to be able to conclude anything.

    Also I wish there was some way we could encourage people to vote on small and quick ideas first, since it would be much easier to get those done and more rewarding for people to see their votes turn in to action. But as I said, it's hard for anyone to know what the easy ideas really are, and I suspect most votes are going in to the major features anyway - I guess they're more exciting.

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads

  • It shouldn't matter to you if the top post has 20 votes or 20000 votes, if it's on top. But it matters to us!

    I think it's better to force people to choose. It gives us a smaller set of things to focus on, which is what we need to do to prevent the problem of being inundated with suggestions.

    I mean that "minor suggestions" category should be for small things that make our experience with C3 easier.

    I don't know what you expected, but in my mind the category was originally created for things that were meant to be quick for us to implement. But as I said, this is not really possible to tell in practice.

    After 2 months you quickly review the ideas (maybe just the titles) to discard those that already exist or not possible. Then you open voting for one week.

    I think we're going in circles again... reviewing a whole set of ideas is a lot of admin work, and we often can't actually usefully comment on them without at least partially implementing them. This kind of thing significantly increases the workload on us, and the goal here is to reduce the workload on us. So I'd rather come up with a system that we weren't required to comment on everything. Maybe just the top few ideas or something.

    I'm not sure a limited voting period is that useful either - I suspect lots of users come and go, and having just say one or two weeks to accept votes means a lot of passer-by type users won't be counted. I'd rather leave the voting open for the whole period, so people have plenty of time to consider their choices and make votes when it suits them.

  • I think the discussion is starting to go in circles. For example people are asking again why we can't comment on every suggestion - I've already explained how much work that is and how often we're simply wrong about our guesses.

    If everything is a priority, nothing is a priority. So I don't think unlimited votes is a good idea, since you don't actually have to choose what's most important to you, you can just throw votes to everything that might be even slightly useful. I know it might feel like an impossible choice, but I want to make people choose their top requests, so we really focus on what is the absolutely most wanted things. Remember that the votes will be accumulated across hundreds or thousands of users, so one person's votes probably aren't going to influence a decision alone anyway. It's also a measure to prevent inundating us with work. If you can only submit at most 10 ideas, then you can't post a mountain of work that's impossible for us to do.

    We'll have a clear notice on the suggestion tracker in the couple of weeks before the reset so everyone knows it's coming. If an idea is lost or submitted late or not voted on much, there's always the next phase.

    I'm sceptical about stating a specific vote count at which we guarantee a response. Ballot stuffing already happens to some extent on the voting platform, and I think if we said "we will respond to all suggestions with 20 votes", people will regularly ballot-stuff their ideas to force us to respond. I think a better way would be to say we will try to respond to the top 20 ideas or so. Although as I've said before, it's very hard to give a better response than "maybe easy" or "maybe hard".

    I believe "Minor suggestions" category was misunderstood from the start. It's not about ideas which are easy for you to implement (we can't possibly know that), it should be for ideas which are minor for us.

    Users don't know what will be quick to do. What I've kept saying in this thread, is neither do we, even though we're experienced programmers looking at our own codebase. Since nobody can tell in advance whether something will take 15 minutes, or descend in to a weeks-long nightmare of breakage and problems, then I don't think it's worth pretending that we know what is or isn't a major or minor idea.

    I almost regret starting this thread. If as a result of it you cut the votes limit from 25 to 10 and remove "minor suggestions", this will be a huge step back to what we had 3-4 years ago.

    Please try to see things from our point of view. Sitting here, the problem is that people have absolutely inundated us with an impossible mountain of work. If we are to prevent that happening again, we need some kind of restrictions or limitations. If we don't have any such restrictions, I would fully expect things to end in exactly the same way - with an impossible mountain of work.

  • This is incredibly inefficient and wastes loads of system resources. If every app did it, your system would slow to a crawl. Why not just use the Date plugin to check how much time passed?

  • I missed a few posts while writing that, so I'd add: the problem with polls is they are only things we think of. Going back to the scene graph suggestion again, I'm not sure we'd have ever added that to a poll. It was user submitted and loads of other people voted on it, and we spotted that, realised the importance, and started the work. If we only do polls with pre-approved features, we might miss that kind of thing, even if lots of people want it. How would everyone show that an idea we're not considering is important to them? Voting on the suggestions platform is meant to be the answer to that.

  • I think if we do a 6-monthly reset, we could take the top 10 or so ideas at the end of that period, and permanently archive them somewhere, so the fact they got a lot of votes is recorded and isn't lost right away.

    The whole point of this discussion is to figure out how to avoid people inundating us with a mountain of work that is literally impossible to implement. Then people get disappointed that what they suggested isn't being worked on. If people can post unlimited suggestions, we'll be inundated. If people can have unlimited votes, then one person can vote on an impossible amount of work for us, and they are guaranteed to be disappointed. If ideas are left behind forever, then gradually it builds up to an impossible amount of work.

    So I do think that if you want to avoid disappointment, then we need limited idea submission and limited votes. If you are OK with being disappointed, then the current system is working fine. But the general view here is it's not working fine, which is why we're having this discussion. We probably need to make a significant change. If you want to post 100 suggestions, I can see how a system that resets every 6 months is annoying. But it's more realistic. It's probably far too much work to do all 100 ideas that you want to submit, especially taking in to account all the other ideas other users are submitting. So resetting the system acts as a filter. If you are willing to keep reposting your ideas, then it's another signal it's something you really want, rather than a throwaway idea someone made a couple of years ago but nobody really cares about it. If you know the ideas will all be deleted in 6 months and so you post your top 10 ideas instead of 100, that is actually much more useful to us.

    I get that Scirra can't make accurate judgements on the feasibility of a new idea, but if that's the case, then ANY method of handling suggestions is fundamentally pointless

    No, that's not the case. When we take on a big feature, we both investigate the feasibility and figure out what it will involve, and plan assuming there will be unexpected complications, follow-on feature requests, and a long tail of associated bug reports. The problem arises when we think something will take say 2 days, and it ends up taking 2 months when we never planned for that. We'll have full-time work already lined up for those 2 months. So then we end up with the stress of unexpectedly having more work than we intended, and other projects get delayed or postponed as we scramble to sort out the things that have gone wrong.

    On a positive note, there are lots of big improvements that came from the suggestions platform. For example, I doubt we'd have done the scene graph feature by now, without having a user submit the suggestion and lots of other people vote it up in to one of the most popular ideas. Taking that from initial design to completion was about 6 months though, and when starting it, I expected it to take about that long. So naturally we can't exactly do that every day. But the fact lots of people voted for it was a useful signal to us. That's the main purpose of the system: it's a way to give us feedback on what most people want.

  • I think anything that involves curation or triage of hundreds of suggestions is not going to work. We want to spend our time improving Construct 3, not researching the feasibility of things only one person wants. Only us developers know the codebase well enough to really evaluate that, and even then, as my past blog post describes, it's often impossible to know what the work will really entail once we start it. Small ideas can easily snowball in to lots of work, so even an initial evaluation that says "should be easy" could simply be wrong.

    For this reason I think it would be best to remove the "minor suggestions" category, so we stop pretending we know what is and isn't easy, because in truth we don't know. I think we should also aim to design something where we are not expected to reply to every single suggestion. It's just too much work, and often our conclusions about feasibility are completely wrong anyway. I think the things people are willing to actually commit their votes to is the best signal to rely on, in the context of the suggestions platform (as the guidelines stress, there are other things we take in to account too, outside of the suggestions platform).

    I quite like the idea of basically deleting everything and starting from scratch every 6 months. It sounds like it would solve a lot of the problems. Instead of accumulating a mountain of old suggestions and votes, it keeps things fresh. It would be focused on what people want now, rather than what lots of votes went to in previous years, perhaps even by users who are no longer active.

    I still think everyone should have a limited number of votes, probably just 10. Once you've voted on 10 things, chances are beyond that there is very little chance it'll really get done. What's the point in voting on 50 things if we only have time to do maybe 20 across everyone's suggestions? I think it's better if every six months you pick your top ten suggestions and that's it. You can always change your votes and vote on other things after the next reset. I also don't think there's any point in submitting suggestions with zero votes. If nobody is willing to put a single vote on it - not even the original author - it seems easy to conclude it's not a priority, given the vast amount of other work that could be done. In fact maybe we should even state our policy is to ignore suggestions with zero votes, to discourage anyone posting them. There's no point inundating us with hundreds of suggestions, so this seems like it would help mitigate that.

    So: how does it sound to have a complete reset of the suggestions platform every 6 months, and everyone gets 10 votes? Any particular objections to that? I'm inclined to move towards doing that in the next few weeks, unless anyone can persuade me of a better way to do it... (and note if your proposal involves us doing lots of regular work or users being able to inundate us with ideas, I probably won't be keen!)

  • Transparency doesn't work well with depth buffers which are required for 3D rendering. This is a known limitation and is probably going to stay.

  • Well, if you're not willing to move your existing votes to newly submitted ideas, then presumably the ideas you've already voted on are more important to you, right? So sorting by highest voted first seems to be the most fair approach.

  • That would break things, which is why we haven't changed it.

  • Many of the 1000 ideas are outdated and no longer relevant. What we asking is to review at least recently posted ideas.

    This is still potentially a great deal of work. As I said, it can involve a fair bit of research and prototype to comment on it at all. And if we make an incorrect or misleading comment, people will get upset.

    I really hope you are not judging the ideas by the number of votes they receive.

    I am kind of mystified by this comment. It is the entire purpose of the system. We want to focus our work on the most popular ideas that lots of people want - not ideas that only one or two out of thousands of customers want. With such limited resources, it would be unfair to the majority of users, if someone managed to force us to spend our valuable time implementing a feature only they or a small number of people want.

    People will switch to posting their ideas on the forum, this will at least give them a lot more exposure.

    Ideas posted to the forum eventually fall in to the back pages and get lost. One of the reasons we set the system up was to solve this problem. If we get rid of it, I expect we will go back to ideas being lost and forgotten in the forum, which I'm not sure is better.

    I'll try to make some time next week to go through the suggestions platform and update the status on some of the ideas, starting with the ones with the highest votes (as those are presumably the ones most people care about). I guess one way to solve the insurmountable-amount-of-work problem is to be more proactive about rejecting ideas that are especially difficult or not something we envision as likely to be done. But that also can upset people and result in angry customers. At least ideas left alone can still collect votes if lots of people think they're important.

    I think the solution is to figure out how to have fewer ideas that most people care about. Having hundreds of suggestions covering years and years of potential work is indeed getting unmanageable. So, if anyone has any ideas on how to radically reduce the number of ideas, let me know... but I'm pretty sure things like letting people have more votes, and be able to submit more ideas, will just make this much worse.

  • As the suggestions platform guidelines describe, the intention is just to collect feedback, and we don't guarantee that any suggestions will be implemented. The reason I keep coming up with "we're a small company" is it's true, and it's by far the biggest limitation we face in our ability to get things done. Frankly given our size I'm proud of how much we actually do get done.

    Only 2 are marked as "shipped" this year.

    That's misleading - I figure you're looking at the "shipped" category, sorted by recent. In that case the dates are when the submission was submitted, not when we marked it shipped. So if a suggestion posted in 2020 was shipped this year, in that view you'd think we didn't do it this year, but we did. Given the suggestion platform has been up close to 4 years, and 200 ideas are shipped, that works out to about 50 a year, or one a week. And that includes ideas that were major projects that took months to complete, like the scene graph feature.

    It would be really helpful if you could mark ideas that you are considering to implement.

    The problem with the size of the suggestions tracker is with 1000 ideas, it's a huge amount of administrative work merely to review them. Providing a useful comment is often difficult unless we research the options and experiment with some prototyping. Doing this for every idea is just an absolute mountain of work and could tie us up for months, during which time we get nothing done at all. So, perversely, providing useful feedback on every idea means would mean we actually implement fewer features.

    I don't think the "minor suggestions" category works that well, because users often don't have the technical knowledge to know if what they are asking for is actually easy or not. It can still take a fair amount of research (e.g. examining thousands of lines of code to estimate how easy a change would be to make to our existing codebase) and prototyping work to evaluate what is ostensibly a minor suggestion, which ends up being a lot of work. In fact I wrote a whole blog about this a while ago, on the unexpected complications of minor features, which covers two cases which turned in to loads of work, for example.

    Every time I post a new idea, I have to revoke a vote from one of my older ideas.

    Well, I think this is kind of reasonable actually. With 1000 ideas, which I estimate is probably about 10 years worth of work, how useful is it really to throw another idea on the pile? I'd rather we had fewer popular ideas, rather than a mountain of things only one or two people want. That's what the limited voting is for. It's meant to force you to pick a few things you really care about, and not so you can inundate us with hundreds of ideas amounting to an infeasibly huge amount of work.

    In the feature guidelines I wrote back in 2017, I did write about some of my fears how this might end up working badly:

    In particular we are worried about people making statements like "Scirra don't listen to their customers - they're ignoring this feature even though everyone wants it". If that kind of thing happens a lot, we will probably just shut down the suggestion platform.

    So if the suggestions platform just ends up being a big source of disappointment and complaints, I guess we could just shut it down. But do you really think that would be better? At least people can vote on popular things as it is.