Hey Phillipp, thank you for your question! Early birds get a discount in the valuation cap that is listed on their SAFE note. If we translate this into simpler terms using an example - suppose we raise the next round at some valuation that is above the cap, and then we sell the company for $1B dollars next year (hypothetically). An investor who put his money in as an early bird, will have a return on investment of:1B/6M = 167XAn investor who put his money in after the early bird period ended, will have a return on investment of:1B/8M = 125XSo the return we'll generate for early birds is going to be 33% higher than the return we generate for others who invest in this round.

Hi Raymond, you can see the list of past investments in the "details" tab. We've only raised from individual angel investors so far.

Hey Enrico, I appreciate the kind words. It's a work in progress but I think we're on the right track.Let me answer your questions out of order:1. The June numbers (through 6/26 at 00:01) have been updated.2. CAC is a blend - most of our signups are from shares, from PR, and from secondary effects (e.g. we advertise this wefunder campaign, but then most people sign up for the app to try it out before investing, too). I.e. they are "free", or rather a freebie that gets attached to something else we paid for. But we do have some paid acquisition as well - right now we can get Web and Android users for <$1 and iOS users for $1-3 depending on location. We're not running ads to improve our numbers at this point - the main reason we are running ads is to get a predictable cohort of users each week that we can run tests on (as part of our growth-hacking process to improve retention), because meaningful tests require consistent sample sizes from a consistent source. 3. Churn/retention is our big focus as well. We're facing a difficult challenge - other media companies retain users by showing them the most addictive content they can select; but we are trying to select content that maximizes the user's long-term well-being rather than their likelihood of getting addicted. This means to retain users we are solving a harder problem than others.Here's our current status. Consider the industry benchmarks: andrewchen.com/new-d…apps-do-much-better Our D7 retention right now is 43.6%. That's above average, but it is not elite just yet. According to Quettra's data, this puts us around the top-5000 but not in the top-500 yet. So we have work to do - but I'm sure we'll get there.

Hi Bruce, we're raising the current round using a SAFE (=Simple Agreement for Future Equity) note, which converts into preferred stock when the company raises a future investment round, gets acquired, or performs an IPO. This is the most common legal instrument for investing in early-stage startups, and it has some advantages over a regular stock purchase (e.g. more downside protections for the investor). Since you mentioned the Google story, here is what actually took place: en.wikipedia.org/wik…olsheim#Investments

Hey Adan, for now the plan is to grow as fast as we can, and then to gradually start adding advertising and/or subscription options (perhaps later this year). I think subscription fees are unlikely to turn people off completely - what they tend to do is make growth slow down (because even if our users like it, they find it harder to share with friends). So if we can afford to go without monetization for a while, we should, because the user-base will grow faster this way.When it's time to monetize, we can consider a freemium approach like Spotify/Hulu where the subscription is not necessary, but rather it just allows you to get an ad-free experience.

Hi Akash, I appreciate the offer, but we already have a great hiring pipeline that helps us hire superstar developers for 1/3 of their market cost.The cost you mentioned in your question (which comes from our SEC disclosures) is for 12 months. 92.5% of 50k, which is $46,250, or $3,854/month for a software engineer. I don't think this cost is too high for a great developer, and I don't think you can hire 20 developers for this budget. In fact, I am pretty sure we have a significant cost advantage over other startups in our space because we are able to hire great talent so cheaply.There are a few key positions for which we'll need to hire in the US, and a market salary for an engineer in Austin is well north of 100k per year. We've listed these planned hires in our disclosures as well. Over time, we're going to balance costs as much as we can - hiring key employees and managers in the US, and hiring individual contributors in lower-cost countries where we have great access to talent.Thanks!

Hi Kali, our initial design followed a style that is similar to Google Reader, and emphasized the ease and speed with which our early adopters (i.e. information-junkies like me) could consume as much information as possible in as little time as possible. We are now working on a redesign that will make the platform more appealing to mainstream users and allow us to cross the chasm - showing one headline at a time, with nice-looking images and swipe gestures to navigate the feed. The new design will be rolled out before the end of February.We will maintain an option for users to revert to the old design (which many of our early adopters will likely use), but I absolutely agree that mainstream users generally prefer more images and less text on the screen. We will support both options soon.

Hey Jodie, a SAFE note gets converted to equity in the company when we raise a priced round above a certain threshold. Startups like us tend to raise funds in multiple rounds, where ideally in each round our valuation is higher than in the previous one.So, when we raise the first priced round, you could see a certain return on investment right away - for example, if we raise the round at 18M and your valuation cap was 6M, each share you receive will immediately be worth 3X what you paid for it.In each subsequent round, the value of your investment will rise or fall together with our valuation. And each time this happens, you will be notified of the new valuation, and may even be asked to vote on whether you accept the terms of this new round or not.Finally, when we reach an "exit", i.e. an IPO or acquisition, your shares will be converted to publicly-traded shares, or shares in the acquiring company, or cash - depending on the exact type of exit we have. Typically, that is the point at which most investors cash out, but you can also choose to continue to be a shareholder afterwards.

Hey Bob, I know it's an uncommon answer these days, but.. we made our own search engine. There are two main reasons for this decision:1. We think that all existing search engines are full of junk results. And there is no easy way to filter these results in real-time - NLP is slow - so to be able to filter them quickly while the user is searching, we had to index them ahead of time for at least the most common queries.2. We think that the basic UX pattern people use with search is unnatural. We're all used to doing it, but.. there's a better way. If you mention a phrase in a conversation, let's say "UX pattern", and I want to find out more about it - I will ask you "what's that?". I won't pull out my phone and type some key phrases into the address bar of a new browser window, right? That's a weird thing to do. And yet online, we're all doing it.Think about how much better "what's that?" is, compared to typing into the address bar of a new browser window. It doesn't interrupt the flow of the interaction or break context. The person I ask "what's that?" has the full context of what we were just discussing, so there won't be any ambiguity in how to interpret my query. Finally, it requires fewer actions.So, we implemented this exact UX pattern into the Otherweb. If you see any phrase you'd like to know more about, you can highlight it and search the web, or wikipedia, or research studies, etc. It's the equivalent of "what's that?" built into the news reader itself - and to do it well, we needed our own search engine under the hood.

Hey Adan, right now we aren't really competing with search engines directly. You could say we are much more competitive vs Google News or the Yahoo portal than we are with the search engine part of Google's business. It's also worth noting, Yahoo is not a search engine - it is a news portal that uses Bing API as their search engine under the hood. Here, we actually have a viable B2B play, since our own search API is comparable to Bing API (which is used by many smaller search portals like Duckduckgo, Quant, Neeva, Ecosia, etc) for some specific use-cases. It's not our focus right now, but as we grow and become large enough to support multiple product lines, we can definitely consider licensing our search API to companies like Duckduckgo.

Hey Christopher, I am not an expert on retirement accounts, but I know someone who can help with the mechanics of getting money from your DCP to Wefunder (if this is the only obstacle). Shoot me an email at alex@otherweb.com and I'll make an intro.

Hi Steve, thank you for the kind words. The reality is we don't have ads yet. We plan to turn them on in a few months (which is why the 12 months offer is still valuable), but they're not on yet.The screen you saw is an experiment to see how people would respond to such an offer if it is made early in the onboarding process - we have more than 20% uptake so far, which is a very good result, but I agree with you that it still needs work (including the option to share with more than 2 people, or to do so at a later stage).For you specifically, let's do this manually:1. Please drop me an email at alex@otherweb.com so I know your email/username and I will make a note on the account to exclude you from ads in the future.2. Please share the app with all your friends directly - via email, whatsapp, etc. The more the merrier!

Hey Tony, we think ChatGPT is a really exciting technology and we're sure it will continue to evolve further. And like all breakthrough technologies, it will introduce quite a bit of flux during the transition period.For instance:Consider what effect ChatGPT will have on the cost of creating low-quality content. It is highly deflationary, right? People will be able to generate "passable" content with less time and effort. Or consider what effect ChatGPT will have on content monetization. Search is a major source of traffic for most websites. If search engines switch from a model of linking to other websites, to a model of summarizing the answers on the SERP directly.. the amount of traffic most websites get will decrease, right?Now, add these two trends together - an increasing amount of junk content, chasing a diminishing amount of monetizable traffic - and it seems like a fair bet that ChatGPT will accelerate the trends we've already been seeing over the past 30 years. The signal-to-noise ratio on the web will continue to become lower at an ever-increasing pace. This makes what we're doing even more critical (and in-demand).

Hey Kyle, great question. The best solution I know of it to sample from a WIDE variety of sources from all sides of the political aisle.It's not perfect of course; some narratives can be quite pervasive. But the wider your sample, the more likely you are to see the counter-narrative to every narrative thrown your way. This is why we went with this 3-step approach of:a) Sample hundreds of sources (and not just a select few)b) Prioritize items that are purely-informative (i.e. articles that list facts without trying to tell you what your interpretation of these facts should be)c) Summarize each article into a few short bullet-points - so you can skim through as many articles as possible in as little time as possibleWe also made our AI models and datasets source-available, to make sure that no one (not even us) could bias the sample to the left or to the right. Thanks!

Hey David, I agree with your assessment of the current state. Our general approach is to separate the left-right axis from the up-down axis - these axes are (mostly) orthogonal and can be measured separately.Most of our AI models are focused on the up-down axis: detecting and measuring things that are universally viewed as bad. We're trying to catch and label things like clickbait headlines, sensationalist language, persuasion masquerading as reporting, lack of references / insufficient sourcing, etc. On this axis, "good" is basically defined as the absence of things that are known to be bad, and right now we have about 20 different bad things we know how to detect and measure. On the left-right axis, we try to choose sources that span the entire spectrum from sane right to sane left. We think left-right balance and viewpoint diversity are both positive traits of a good information diet. Of course if we had to pick one source we'd probably pick a centrist one; but since we can pick hundreds of sources - it seems better to sample the entire Overton window, and to include everything from the NY Post and Washington Examiner and Reason Magazine to MSNBC and Vox and Salon (and everything in-between).Our goal is for the reader to get high-quality right-wing articles, high-quality centrist articles, high-quality left-wing articles. And even if a particular source prints a lot more junk than others, we will still try our best to sample the few good articles they published and evaluate each article on its merits.Hope this answers your question.

Hey Curtis, great question! Here's how we think about it -There are two exit routes - M&A or IPO. Which of the two ends up happening often depends on circumstance, so our job at these earlier stages is to grow the company as fast as possible and to keep both exit routes open.The valuation in both cases is a function of user-count, engagement and growth - so the key for us right now is to grow the user-base and increase engagement. If we have 10m users by this time next year, then perhaps we'll have a decision to make on whether we prefer the M&A route or the IPO route; but until then we have to keep both options in mind, and just focus on maximizing our value and the value we bring to our users.

Thank you!

Hey Jeremy, different categories of product tend to use different metrics to track progress and valuation. Consumer apps tend to use monthly active users (MAU) and user-growth as the two primary metrics, because the assumption is that once you have users - you'll find a way to monetize them eventually (via ads, subscriptions, merchandise, etc), and the decision on whether to monetize now or later is a tactical one. Paradoxically, early monetization is often bad for valuation because it slows down user-growth, so if a startup can afford to postpone monetization and focus on just getting more users - it is sometimes (though not always) rewarded in the long run. This is why companies like Google, Facebook and Instagram postponed monetization for as long as they did. We are making the same choice for now.So, let's run through a few public examples: Whatsapp was acquired in 2014 for $20B, when it had about 400 MMAU, i.e. $50/MAU. But that was in 2014 and valuations are higher today. Facebook is valued at $463B with 2.96 BMAU, i.e. $156/MAU. Twitter was acquired for $44B with 256 MMAU, i.e. $172/MAU (that was the peak of the market; today their valuation would be lower). The numbers vary in each case because of differences in growth-rate and in the quality of the audience (e.g. a Linkedin user is obviously worth more than a Snapchat user because he/she has more disposable income, on average). But as a rule, it seems that a non-paying active user is worth somewhere between $100 and $200 right now. We are pricing this round at $80/MAU based on today's user count (and $60/MAU for early birds) to make sure that our early investors have a lot of upside and a vested interest in our success.

Hi Vacha, please take a look at this FAQ: help.wefunder.com/#/…arted-for-investorsSpecifically, I think the section titled "Earning a Return" would help clarify things.If you have any further questions, please don't hesitate to ask!

Lifetime. As soon as we do the first rolling close these channels will be set up and will exist in perpetuity.

Hey Raymond, that's a very perceptive question. We actually had the ability to comment before, but disabled it temporarily. So the code exists, but at the moment the comment widget is hidden from users' view.Here's why - There's a difference between dealing with user-generated content and dealing with content generated by professionals (be they journalists, podcasters, wikipedia editors, research scientists, etc). Professionally-generated content comes from publishers whose identity is known, so it is pretty much guaranteed not to have illegal things like incitement to violence or child pornography in it - but comments generated by anonymous users can include all sorts of things.When the number of users was small, this wasn't a big concern. Any comment that included illegal content could just be flagged and removed very quickly. But now that the user-base is getting larger (and growing fast), we need to build an infrastructure for handling this properly.Note that it's not viewpoint censorship we're talking about here. Our primary concern is to have a mechanism in place that allows us to avoid having illegal things on our platform, since that would a serious legal liability.So we launched an effort to add some new models that detect illegal content, and to re-train our AI models to handle user comments as well, but while that effort is ongoing - we figured it's safer to just disable commenting for a while and avoid the legal liability. We will relaunch comments when the models are ready (hopefully next month).

Hey Tom, thank you for the great conversation. It was a pleasure!

Hi Travis, I share your concern - every time I hear of fact-checking, I can't help but ask - "who watches the watchers"?Even if fact-checkers had no agenda (which is unfortunately not the case), they'd still not be able to check facts in the general sense. Let's say you claim there's a lion in your kitchen right now - the only way to fact-check your claim would be to break into your house and take a look. But I can't do that, right?What I can do, however, is require a certain "form" for your claim - do you have photos? Do you have multiple eyewitness accounts or external references? Are you describing the claim in sufficient detail, and is your language primarily informative (or, alternatively, is it emotionally-loaded and trying to convince me of something)? That's a natural process we all apply to claims we hear from people. It's also what editors are taught to do in selecting what to publish, but unfortunately these standards have been slipping in most publications today. So we're applying the same process to articles online - except our editors are AI models.When we (i.e. the Otherweb) describe something as junk, we're not saying the substance of the claim is false - we're saying the claim was laid out in bad form, or is missing external references, or uses emotionally-charged language that aims to persuade rather than inform, etc.P.S.We are a public benefit corporation specifically because we want our mission to be legally-binding. This reduces the risk of the company ever losing its way and turning into leftist/rightist propaganda - because the mission is to be apolitical and informative, and not leftist or rightist.We are also making our models source-available - so people can see for themselves that there is no bias on the left-right axis. Our goal is not to move the discussion left or right. It's to move it upward.

Hey Ramesh, if you get in before the early-bird discount ends, then the cap is still $6m. There are 10m shares outstanding. So the simplified way of looking at this is that each share's price is capped at $0.60 and you're guaranteed to get at least 1667 shares for your $1000.However, that's simplified, and a SAFE with a cap is not exactly the same as a priced-round valuation - it's better for investors, and offers you more protections compared to a regular stock sale. To read more, you can check:pandadoc.com/blog/what-is-a-safe-noteycombinator.com/documents

Indeed, that's why making things source-available is probably a necessity for anyone who wants to earn the public's trust going forward. It doesn't solve all biases (RLHF can still introduce heavy bias if done carelessly, as it did in ChatGPT), but it solves a lot of it.

Hey Robert, MAU are summed up across all platforms (web + mobile + newsletter). Right now web still dominates in our product mix, because our biggest source of traffic so far has been social media (e.g. when Otherweb users share an article on twitter or reddit, their friends follow the link and come to our website). We're working on converting our web users into mobile users - but it's a work in progress. For now we're still very much web-heavy.