Launch HN: Foster (YC W21) – Improve your writing with on-demand editing (foster.co)
Most online writing today is produced in a broken way. Writers are commonly told to publish a lot, see how readers respond, and adjust. But most internet readers don’t comment or respond. They just get whatever they can and move on. They bail on an article the minute it fails to hold their attention, and you don’t get a second chance. Thus the most common response to writing online is silence. Silence is a terrible teacher!
What writers need is feedback before they hit publish, so that readers receive the best-expressed version of their ideas. But it’s hard to get quality feedback on a draft. Asking friends tends to be unsustainable—you burn an ask each time—and hiring an editor is complicated and expensive. That’s the problem we’re solving with Foster.
I published a personal blog for ~2 years that basically nobody read. One day, I decided to start asking friends for feedback on my writing. I was pretty surprised to discover how much my writing sucked. The first post I published after an extensive round of peer edits became my most popular piece.
At the time, I was working on another startup and blogging for fun on the side but the experience was so profound that I wondered if this might be something I could help more people experience one day. A year later, I reconnected with my friend (and now co-founder) Dan and we realized neither of us were passionate about our startups at the time. We started to talk about our mutual love of writing and realized this might be our chance to build something we really cared about.
We launched Foster as a Slack group that writers could join to swap feedback on each others’ drafts. But over time we learned that 'writer' and 'editor' really aren't the same thing. Most of our members primarily loved to write, and while editing other people’s work could be helpful and interesting, they ultimately wanted to just get the best possible feedback on their own work.
We started to redesign the experience so that writers didn’t have to join a full-on community and could instead share and track their drafts using a simple web app. On the other side, we recruited professional editors to specialize in improving drafts.
Eventually, we realized we should just bring the entire experience to the draft itself since that’s where the writer lives, so we built a Chrome extension that works with Google Docs. Writers see a Foster button directly in their doc that they can tap to submit the work to Foster. Editing help then comes directly to their document without them ever leaving.
Unlike hiring an editor via marketplaces like Upwork, you don’t need to do any job-posting or back-and-forth. You hit a button in Google Docs and high-quality collaborators jump in to help. (The majority of the writers we spoke with early on use Google Docs, but we plan to expand to other writing platforms in the future.)
We’ve built a collective of editors, writers, and experts who enjoy jumping into first drafts and helping to improve them. Foster contributors do more than spot a missing comma—they give you suggestions on how to make a story funnier or how to make your argument air-tight. People tend to underestimate how much better their writing can be. We’ve found it's hardest to get somebody to post a first draft in Foster, because often they don't think they need an editor. But once they do, more than half post another draft.
A few of our early users are Hacker News regulars who have used Foster to improve their work before submitting it here. These were both sent through Foster prior to being submitted to HN: https://news...
58 comments
[ 3.4 ms ] story [ 110 ms ] threadAnother thing that might be worth clarifying is that most drafts get 2-3 contributors, each of whom has a specific skillset or expertise that is relevant for the draft. So, the cost per-editor on a draft tends to skew downwards quite a bit.
We had originally assumed that editors would not like this sort of variability, but two things have ended up happening: 1) their per-hour rate ends up being the same on Foster given that the work comes directly to them (no selling themselves, doing onboarding calls, etc.) and 2) they strongly value the freedom to select drafts that intrinsically interest them. It also leads to a much better experience for the writer since there tends to be more of a missionary spirit, rather than a purely mercenary one.
What percentage do you get to keep out of the $55?
Or, perhaps a better question is, since you are running a two-sided marketplace, do you make money off of buyers (writers) or suppliers (editors) or both parties?
Can you elaborate more on why you don’t think having writing that performs better justifies the cost?
Example: you pay for our $99/month plan and over six months you write a 1,000-word post each month and submitted it on HN. If one of them ended up on the home page, you don’t think it would generate at least ~$600 in value for the author?
Who is willing to spend that much for proofreading one page a month?
We think Grammarly and other tools will continue to be helpful for proofreading and adding some final polish.
I think editing is _extremely_ valuable -- a crazy high percentage of pieces we edited wound up on the front page of HN.
Ultimately, I dropped the idea because most non-writers think of editing as an ex post faco activity. But talk to any editor, especially ones who work for major publications, and you'll quickly learn they are most effective when they are involved in the process from the very beginning: from identifying unique stories, crafting narratives, shaping the research process. Polishing prose is only the final, and ultimately, most trivial step.
I also don't think that this is venture scale. I think there is a market for talented editors serving serious professionals who can afford a premium product, but scaling this business is very hard.
The Google Docs plugin idea is very clever. I never really figured out a good onboarding funnel beyond "do it manually".
Good luck to the team, really hope that someone makes this work!
I agree that most non-writers tend to think of editing as either an afterthought, unnecessary, or only something that professionals do. It makes positioning here really tough.
One thing we've experimented with isn't selling editing per se, but trying to counter-position ourselves to the publishing into the void problem. Like, "hey - you're probably busting your ass writing and publishing a bunch, but nothing is happening. That's because silence is a crappy teacher. There's a better way to write..."
Most of the writers who have used Foster were not working with an editor before they joined, but once they submit their first draft they often say they can't believe they ever wrote without getting feedback first.
I think if we can figure out positioning this in a new way and if we deliver a genuinely new editing experience (the input you get reliably goes far beyond catching grammar errors), it will unlock a ton of latent demand and be big. If not, I agree with you that selling editing is hard to scale.
Specifically, we think it'd be cool for contributors (editors) to earn more than just money and reputation by editing on Foster. It'd be exciting for them to have a direct say in things like what types of writing we want to collectively support, new ways we might support writers, and how we generally build a passionate contributor community. We want to do anything to make it rewarding and worthwhile to be a contributor and DAOs offer some intriguing new ideas and tools for how we might do that.
As we got into the weeds of what it'd take to build for a few large customers first, it seemed that we'd have a lot of custom functionality and servicing required that wouldn't necessarily generalize to other customers.
Our theory is that if we can build a product that works on the individual atomic unit of writing (the draft) then we'll be able to move upmarket – whereas the inverse may not be as likely. On some basic level, a team just produces a higher volume of drafts than an individual.
Focusing on individual writers and individual drafts allows us to stay much more focused on the atomic experience for now.
Great - you bring GPT3 DaVinci to Google Docs. Or Chrome. Or whatever.
You think this hasn't occurred to Google. Or that your prompt engineering is beyond Google? (In reality they're waiting for their own model, or trying to negotiate a large-scale deal as Microsoft did.)
"Extend a popular application of your choice to support GPT3" is likely already on curriculum everywhere for people learning to code.
There's no value-add being one of 1,000 people implementing the most-obvious use case when an API gets released to the world at the exact same time.
And clearly in 2 years time this is wrapped directly by Google & provided by OpenAI directly & also will be a very popular beginner code tutorial project.
I was overly pattern matching because of the eye-rolling number of GPT3 hello-world companies.
I'll leave my comment up for the MANY people who will pattern-match the same way.
In our team (small SaaS startup) I am the only one writing product updates, documentation, blog content, social posts content, help sections.
I've heard about Foster before but seeing the landing page now made me sign up instantly. I think this is exactly what I was looking for! Excited to try it out!
Soylent: A Word Processor with a Crowd Inside https://people.csail.mit.edu/msbernst/papers/soylent-uist201...
Human feedback loops can adjust nicely to pretty much anyone’s skill level.
Also, the subscription pricing with word limits seems wonky. Why not just go with usage pricing and discounts at higher volumes? (eg. $0.10 / word.) You'd have a much more likely chance of keeping me as a customer if I don't need to cancel when I stop using you for a while.
On pricing, that’s helpful feedback. We’ve found that the writers who get the most value out of Foster tend to be ones that write regularly. The monthly pricing helps us self-select for those customers and we’ve found it created a self-fulfilling mechanism for encouraging writers to submit more work (and often to write more: https://twitter.com/liuxi/status/1511191555796729862).
But it’s not perfect and I can imagine a world where we can support work on a more one-off basis.
If I have a good editorial experience I am more likely to "ship" my writing, and move on to the next essay/piece/post, and keep using the editorial service as a part of my "shipping" process (starting from my first draft of the post). Still, I would definitely be slower initially as I reforge my writing habits around an on-demand editorial service, but eventually I expect I would write often.
Any pricing structure that relates more to the amount of writing/editorial work I get done rather than to units of time would make me far more likely to buy in — non-expiring "credits" I can buy, or per-word pricing, or flat pricing per essay/post/doc/<10k words bloc.
Is this an HN bug, or is there some strange thread merging behavior happening here?
Edit: Oh, it's an artifact of us re-upping the thread, which we do with Launch HNs as long as commenters are staying engaged. We use the same software as when putting submissions in the second-chance pool (explained at https://news.ycombinator.com/item?id=26998308), and that relativizes timestamps as I've described here: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que....
Actually that arguably is a bug, because we don't need to do anything with comment timestamps in cases like this. We do it because when we first started the second chance pool, people would get confused by submissions on the front page with timestamps like "3 days ago" or whatever, and the same if there are comments in the thread saying "3 days ago".
It doesn't change the official timestamps, though, which you can see by hovering over the "x minutes/hours/days" ago text. So in that sense physics remains intact.
For example, a user sends in an academic draft, represents it as their own, and does so with the intent (explicitly voiced or not) of using the service to get past automated plagiarism software / filters?
If you're familiar with academic writing, it's a pretty plausible thing that could happen.
Or is that type of abuse-prone material automatically turned down? Couldn't tell from the site.
For non-STEM writing - your typical business and humanities-oriented material - you'd be surprised how frequently this issue comes up with instructors, teaching assistants, etc.
The pricing page has a notice saying “Use the code FIRSTDRAFT for a free month of the Hobbyist or Basic plan,” but the only plan described on the page when I access it is the Hobbyist plan at $55/month for up to 500 words. Is information about other plans also supposed to appear on that page?
Also, how are you and your contributors dealing with people who are writing in English as a second language? Such writers can benefit from the same kind of higher-level feedback on content, organization, audience awareness, etc. as native-speaker writers, but they also need help with grammar, word choice, cultural assumptions, etc. It’s difficult to provide higher-level support to nonnative writers without also dealing with grammatical and vocabulary issues, but the latter issues can take a lot of the editor’s time.
Knowledge of the writer’s first language is also often helpful for identifying the reasons for grammar and vocabulary issues. Are you making any effort to match your editors to writers based on the editors’ knowledge of non-English languages?
One more point (added later): Have you looked into the one-on-one tutorial model for supporting writers that is used in high-school and university writing centers? Writing centers in educational contexts are typically focused on making people better writers, not just fixing specific texts. This is usually done by the tutor and writer discussing the writer’s text together, with the tutor asking questions as often as making suggestions.
While it’s possible to do tutorials through comments on online documents, it’s much more productive to do it interactively. You might want to look into offering video tutorials between the writer and editor in addition to the Google Docs integration.
We do have some writers for whom English is a second language. We also have a handful of collaborators who have experience working with those folks. So far, those writers tend to be up front that English is their second language and so the contributors who jump into their work tend to self-select based on if they feel qualified to help. For now, this sort of emergent pairing of writers and contributors tends to give us more flexibility in what we can handle (as you noted, there's a lot of nuance within just this one use case).
And the tutor / tutorial model is certainly interesting (we've thought about it in a different context: hiring a writing coach), our general sense was that it is different enough from our core draft-level editing service that it's out of scope for the foreseeable future. I do agree there are certain things you can do live that are much harder (or impossible) to do asynchronously in doc comments.
Oh, I see. I actually did drag that slider a bit to the right the first time I accessed the page, but not far enough to go to the next plan, so I didn’t understand the purpose of the slider. If you get similar reactions to mine, or if your access data reveals that many human visitors to the page do not touch the slider, you might want to rethink your interface. Boxes displaying all of the plans side-by-side (vertically on mobile) would have been fine for me.
Thanks for your responses to my other questions. I wish you the best of success.
But your current pricing seems very high. Curious to see if there will be enough users to support this.
These AI platforms are targeting improved writing as well. Not just fixing a missing comma, but suggesting sentences be re-written to be made clearer. Also offering suggestions on how these can be re-written.
Something I'm seeing in YC companies is often completely ignoring existing competitors or would be competitors in the next year or two.
How could you write all about your product and no mention any of the existing AI tools, even if just to say "we're better than using X because [a,b,c]"?
Perhaps you have something so completely different and better, but if you don't address that, it seems like you've just got blinders on.
And what do the editors specialize it? Can I write cooking blogs? Literature blogs? Computer science paper reviews?
Can you show me a post before editing vs after editing? Only then I'll understand the value provided.