5 ms·

My proposal for scaling open source: don’t

123 points by mpweiher 5y ago

celeritascelery 5y ago
I consider this whenever I see someone saying that “such and such” text editor needs to “be more like vscode”. Trying to out do Microsoft’s paid development team is a losing game. You will always be a second place player. And no one wants a crappy vscode clone. But as the article says, you don’t have to focus on the lowest common denominator, but can instead build an incredible experience for the small set of people who want it.
- pixl97 5y ago
  Which small set of users do you choose to optimise for?
  - legobmw99 5y ago
    My guess is the answer is something like “the group most similar to yourself”
  - void_mint 5y ago
    Some intersection of "The users that gave the most interesting or thoughtful feedback" and "The users who are closest to the original intent of the tool."
  - krisoft 5y ago
    It is your free time, you build it to whoever you want it to. Most often this will be you and people similar to you. Less often you build it for a friend and people like them, or an organization you care about and organizations like them.
  - Scarblac 5y ago
    You.
  - Torwald 5y ago
    Depends who do you want to serve. Here is an historic example: the GIMP. AFAIK the original purpose the software was intended to serve was to create merchandise, I think post cards, from fotos to sell at a stand. Compare that to Photoshop whose original intend was to make images ready for print production, you see two very different set of users. Photoshop filters got their name from studio lightning filters, a technical term completely logical to it's intended audience. Everybody who did movies/fotos in the pre-digital era new what a filter was. This naming made the Photoshop menu and concept intuitive, thus raising it's usability.
  - jlokier 5y ago
    I think that's a tough question. The replies that suggest just yourself, people like you, friends etc don't provide satisfaction to those people who gain the most satisfaction from building things large numbers of other people want to use and ideally enjoy using. I'm happy to write/build little things for myself, friends and other small groups up to a point. But the motivation is limited. I don't see myself putting in 1000 hours polishing something like that, as it feels like a waste of limited lifespan. It's just not fun. Whereas I'll happily put in 5000 hours to a project that I feel is, or could be, making a noticable, useful and liked difference to the lives of millions of people. It some level of psychology that's satisfying, a kind of deeper fun. Don't ask me why it makes a difference how many people are affected. I don't have a good answer. I know there are many people who don't feel this way, because they say so. But it seems to be how some people are wired. But there's that pesky scaling problem. So which people and which projects to focus on? I think it's a dilemma that a lot of people struggle with in times of creative freedom who are wired that way, and I have yet to recognise any really good articles or guidance on resolving it.
- osbzvyzaa 5y ago
  I dont contribute to open source these days cause of the amount of drama. But what I do is fork and add stuff that I find useful or want. Eventually someone or the other finds it and if they care enough pulls it in. I find this much better than convincing open source gatekeepers, who already have their brains overloaded, to accept pull requests.
  - chii 5y ago
    > fork and add stuff that I find useful or want this is the whole premise of open-source originally. If the "fork" is good, either the up-stream takes the changes, or the fork becomes the official one. This will be determined by the users of the program.
    - kiba 5y ago
      I feel free and open source software's promises is still yet unrealized since I don't do these sort of things, and probably neither do most programmers. I know that Free software is about freedom first and foremost not whether we take advantage of these freedoms, and that it has many ancillary effects on our privacy right and keeping old hardware running and all...
  - tomxor 5y ago
    > I dont contribute to open source these days cause of the amount of drama. I've seen the drama (from a distance) in some large FOSS projects, but there is a sea of small one-person or small group projects which are usually a pleasure to contribute to - they usually entail nice simple, small scale, person-to-person interactions, not massive discussions and red taping - it's fun :)
pizza234 5y ago
The same concept was advocated by 37signals/Fried, in Rework: > Let your customers outgrow you [...] When you let customers outgrow you, you'll most likely wind up with a product that's basic--and that's fine. Small, simple, basic needs are constant. There's an endless supply of customers who need exactly that.
cblconfederate 5y ago
Thats why it bugs me to see e.g. linux desktop inanely adopting mainstream interfaces instead of focusing on steeper learning curves that come with high rewards. Dumbing down technology also leads to dumbed down users , and now we have already a generation that is struggling to use any tool that is not a touchscreen. Thankfully some technologists are starting to wake up to the fact that dumber isnt better (or maybe they ve milked the dumbdown cow to the end) and that we need more capable tools even if they re hard to use (e.g. this interview of Palmer luckey https://www.youtube.com/watch?v=epy8UVZilmA&t=990s https://www.youtube.com/watch?v=epy8UVZilmA&t=990s)
- rgoulter 5y ago
  > Dumbing down technology also leads to dumbed down users It seems more likely to me in general that 'making technology more accessible increases the number of people who will be able to access that technology' than 'simpler interfaces make people dumber'.
  - cblconfederate 5y ago
    OTOH i think dumbing down is an active process that leaves end users untrained to complex interfaces. I can compare young students who struggle with a complex interface like blender , yet i remember my uncle (a lawyer) who had learned sql just to use DbaseIV in the '80s
    - yarcob 5y ago
      The difference being that SQL is a standard that's used across many products, has a very flat learning curve, and will be familiar to anyone who uses databases. Whereas Blenders UI is unlike anything else and has a very steep learning curve even for people familiar with other 3D modelling tools.
    - II2II 5y ago
      Your uncle was special, even for the 1980's. While many people did take the time to learn how to use those more complex interfaces, many people avoided computers altogether. Contrast that to today. People are expected to use various forms of computers, even if there is no compelling reason to do so. The thing that I find bothersome is the expectation of simplified interfaces and feature sets outside of professional domains. It is an approach that works well if the software does exactly what you want of it, yet fails miserably the moment a user's expectations deviates from the developer's expectations. There is little room for creative thought or problem solving since the software tries to do everything for the user rather than providing a set of tools to them.
smitty1e 5y ago
The truth that all human organizations tend toward Towers of Babel has never deterred tower construction.
greatgib 5y ago
The only issue with this concept is that when everyone use a proprietary solution and you are a marginal, it is very hard for you to have support. If everyone you know is using zoom and WhatsApp, they will not use signal and jitsi just for you, so you will be forced to use the proprietary tools. Think about remote school and trainings for example ,here you have zero choice. And in the past, we saw the case with Linux where it was very hard to have support for hardware as manufacturer did not care at all about Linux. For such a case, scaling is an important factor to be able to use your open source software. Sadly.
pornel 5y ago
To me a lot of OSS already has a disdain for the "lowest common denominator" type of users, and this article has this vibe. You don't have to "growth hack" and beat $1T companies, but don't use that as an excuse for ignoring usability. In OSS "simple" more often means "works for the maintainer" rather than something that's actually intuitive and usable for the majority of people. If you only have a handful of hardcore users, because everyone else failed to run the software, and was told to read the (barely existing) manual, that's a problem with the software's design, not scale.
- spinningslate 5y ago
  Guardedly agree, though there's also entitlement risk here. If an author scratches their own itch, and it does what they wanted, and they decide to open source it, then good on them. Choosing to do so doesn't create any requirement to make it useful to anyone else. This is the root of a chunk of angst today: consumer X finds open source project Y, discovers it doesn't do Z, and throws a twitter/issue hissy fit. There's *zero* requirement on the author to fix the perceived deficiency, nor should there be any expectation. If X wants it addressed, then - as noted elsewhere in this topic - a founding principle of open source is that X is free to fork and fix. Now, of course, some authors create open source and actively solicit use by others. In that case, they do need to consider feedback from their users. But again, there's no divine right conferred on users that the author will address their every whim. Fork and fix should be the first option - with no expectation it'll be upstreamed.
fouc 5y ago
I think this is telling us to drop X windows / Wayland, it's time to acknowledge that's a failed experiment. We should go back to just terminals and load firefox in framebuffer when we need that. ;)
arp242 5y ago
> We already have various tools for enabling growth: the freedom to use the software for any purpose being one of the most powerful. This has been tried for the last 30 years; with some success, but very limited, and probably not really primarily attributable to "Software Freedom" either. Essentially this article is "keep doing what we've been doing for the last 30 years", which doesn't strike me as a very good strategy. I do think we should not just "copy" whatever others are doing. Actually, I call this the "new coke fallacy": 1. Project X is more popular than us. 2. Project X does Y. 3. Therefore, we must do Y. There are all sorts of problems with this: you may misidentify Y, X probably does a good job solving Y already (not always though!), and exactly because it's not X and does Z instead of Y is why you have (part of) your user/customer base in the first place. It didn't work out very well for Coca Cola, and unless you're really sure what you're doing it's probably also not going to work for you (but if you can do Y substantially better, it might be a good idea).

123 points by mpweiher 5y ago

I consider this whenever I see someone saying that “such and such” text editor needs to “be more like vscode”. Trying to out do Microsoft’s paid development team is a losing game. You will always be a second place player. And no one wants a crappy vscode clone. But as the article says, you don’t have to focus on the lowest common denominator, but can instead build an incredible experience for the small set of people who want it.

pixl97 5y ago
Which small set of users do you choose to optimise for?
- legobmw99 5y ago
  My guess is the answer is something like “the group most similar to yourself”
- void_mint 5y ago
  Some intersection of "The users that gave the most interesting or thoughtful feedback" and "The users who are closest to the original intent of the tool."
- krisoft 5y ago
  It is your free time, you build it to whoever you want it to. Most often this will be you and people similar to you. Less often you build it for a friend and people like them, or an organization you care about and organizations like them.
- Scarblac 5y ago
  You.
- Torwald 5y ago
  Depends who do you want to serve. Here is an historic example: the GIMP. AFAIK the original purpose the software was intended to serve was to create merchandise, I think post cards, from fotos to sell at a stand. Compare that to Photoshop whose original intend was to make images ready for print production, you see two very different set of users. Photoshop filters got their name from studio lightning filters, a technical term completely logical to it's intended audience. Everybody who did movies/fotos in the pre-digital era new what a filter was. This naming made the Photoshop menu and concept intuitive, thus raising it's usability.
- jlokier 5y ago
  I think that's a tough question. The replies that suggest just yourself, people like you, friends etc don't provide satisfaction to those people who gain the most satisfaction from building things large numbers of other people want to use and ideally enjoy using. I'm happy to write/build little things for myself, friends and other small groups up to a point. But the motivation is limited. I don't see myself putting in 1000 hours polishing something like that, as it feels like a waste of limited lifespan. It's just not fun. Whereas I'll happily put in 5000 hours to a project that I feel is, or could be, making a noticable, useful and liked difference to the lives of millions of people. It some level of psychology that's satisfying, a kind of deeper fun. Don't ask me why it makes a difference how many people are affected. I don't have a good answer. I know there are many people who don't feel this way, because they say so. But it seems to be how some people are wired. But there's that pesky scaling problem. So which people and which projects to focus on? I think it's a dilemma that a lot of people struggle with in times of creative freedom who are wired that way, and I have yet to recognise any really good articles or guidance on resolving it.
osbzvyzaa 5y ago
I dont contribute to open source these days cause of the amount of drama. But what I do is fork and add stuff that I find useful or want. Eventually someone or the other finds it and if they care enough pulls it in. I find this much better than convincing open source gatekeepers, who already have their brains overloaded, to accept pull requests.
- chii 5y ago
  > fork and add stuff that I find useful or want this is the whole premise of open-source originally. If the "fork" is good, either the up-stream takes the changes, or the fork becomes the official one. This will be determined by the users of the program.
  - kiba 5y ago
    I feel free and open source software's promises is still yet unrealized since I don't do these sort of things, and probably neither do most programmers. I know that Free software is about freedom first and foremost not whether we take advantage of these freedoms, and that it has many ancillary effects on our privacy right and keeping old hardware running and all...
- tomxor 5y ago
  > I dont contribute to open source these days cause of the amount of drama. I've seen the drama (from a distance) in some large FOSS projects, but there is a sea of small one-person or small group projects which are usually a pleasure to contribute to - they usually entail nice simple, small scale, person-to-person interactions, not massive discussions and red taping - it's fun :)

pizza234 5y ago

The same concept was advocated by 37signals/Fried, in Rework: > Let your customers outgrow you [...] When you let customers outgrow you, you'll most likely wind up with a product that's basic--and that's fine. Small, simple, basic needs are constant. There's an endless supply of customers who need exactly that.

cblconfederate 5y ago

Thats why it bugs me to see e.g. linux desktop inanely adopting mainstream interfaces instead of focusing on steeper learning curves that come with high rewards. Dumbing down technology also leads to dumbed down users , and now we have already a generation that is struggling to use any tool that is not a touchscreen. Thankfully some technologists are starting to wake up to the fact that dumber isnt better (or maybe they ve milked the dumbdown cow to the end) and that we need more capable tools even if they re hard to use (e.g. this interview of Palmer luckey https://www.youtube.com/watch?v=epy8UVZilmA&t=990s https://www.youtube.com/watch?v=epy8UVZilmA&t=990s)

rgoulter 5y ago
> Dumbing down technology also leads to dumbed down users It seems more likely to me in general that 'making technology more accessible increases the number of people who will be able to access that technology' than 'simpler interfaces make people dumber'.
- cblconfederate 5y ago
  OTOH i think dumbing down is an active process that leaves end users untrained to complex interfaces. I can compare young students who struggle with a complex interface like blender , yet i remember my uncle (a lawyer) who had learned sql just to use DbaseIV in the '80s
  - yarcob 5y ago
    The difference being that SQL is a standard that's used across many products, has a very flat learning curve, and will be familiar to anyone who uses databases. Whereas Blenders UI is unlike anything else and has a very steep learning curve even for people familiar with other 3D modelling tools.
  - II2II 5y ago
    Your uncle was special, even for the 1980's. While many people did take the time to learn how to use those more complex interfaces, many people avoided computers altogether. Contrast that to today. People are expected to use various forms of computers, even if there is no compelling reason to do so. The thing that I find bothersome is the expectation of simplified interfaces and feature sets outside of professional domains. It is an approach that works well if the software does exactly what you want of it, yet fails miserably the moment a user's expectations deviates from the developer's expectations. There is little room for creative thought or problem solving since the software tries to do everything for the user rather than providing a set of tools to them.

smitty1e 5y ago

The truth that all human organizations tend toward Towers of Babel has never deterred tower construction.

greatgib 5y ago

The only issue with this concept is that when everyone use a proprietary solution and you are a marginal, it is very hard for you to have support. If everyone you know is using zoom and WhatsApp, they will not use signal and jitsi just for you, so you will be forced to use the proprietary tools. Think about remote school and trainings for example ,here you have zero choice. And in the past, we saw the case with Linux where it was very hard to have support for hardware as manufacturer did not care at all about Linux. For such a case, scaling is an important factor to be able to use your open source software. Sadly.

pornel 5y ago

To me a lot of OSS already has a disdain for the "lowest common denominator" type of users, and this article has this vibe. You don't have to "growth hack" and beat $1T companies, but don't use that as an excuse for ignoring usability. In OSS "simple" more often means "works for the maintainer" rather than something that's actually intuitive and usable for the majority of people. If you only have a handful of hardcore users, because everyone else failed to run the software, and was told to read the (barely existing) manual, that's a problem with the software's design, not scale.

spinningslate 5y ago
Guardedly agree, though there's also entitlement risk here. If an author scratches their own itch, and it does what they wanted, and they decide to open source it, then good on them. Choosing to do so doesn't create any requirement to make it useful to anyone else. This is the root of a chunk of angst today: consumer X finds open source project Y, discovers it doesn't do Z, and throws a twitter/issue hissy fit. There's *zero* requirement on the author to fix the perceived deficiency, nor should there be any expectation. If X wants it addressed, then - as noted elsewhere in this topic - a founding principle of open source is that X is free to fork and fix. Now, of course, some authors create open source and actively solicit use by others. In that case, they do need to consider feedback from their users. But again, there's no divine right conferred on users that the author will address their every whim. Fork and fix should be the first option - with no expectation it'll be upstreamed.

fouc 5y ago

I think this is telling us to drop X windows / Wayland, it's time to acknowledge that's a failed experiment. We should go back to just terminals and load firefox in framebuffer when we need that. ;)

arp242 5y ago

> We already have various tools for enabling growth: the freedom to use the software for any purpose being one of the most powerful. This has been tried for the last 30 years; with some success, but very limited, and probably not really primarily attributable to "Software Freedom" either. Essentially this article is "keep doing what we've been doing for the last 30 years", which doesn't strike me as a very good strategy. I do think we should not just "copy" whatever others are doing. Actually, I call this the "new coke fallacy": 1. Project X is more popular than us. 2. Project X does Y. 3. Therefore, we must do Y. There are all sorts of problems with this: you may misidentify Y, X probably does a good job solving Y already (not always though!), and exactly because it's not X and does Z instead of Y is why you have (part of) your user/customer base in the first place. It didn't work out very well for Coca Cola, and unless you're really sure what you're doing it's probably also not going to work for you (but if you can do Y substantially better, it might be a good idea).