Dear Studio Fam,
The future of the tech industry seems likely to be defined in a court room. This week we bring you news of four separate tech battles in US courts. We start with the latest crypto fraudster's arrest and then take a look at the first major AI lawsuit brought by writers. Then we look at a split decision in Federal court on the legality of a specific crypto token and hear about Twitter's new lawsuit against data scraping thieves in Texas. Finally, we've included another post from our new Knowledge Base that should be helpful to any early stage startup founder.
Celsius CEO Alex Machisnky Arrested for Fraud
Another prominent crypto executive is facing jail time. Celsius founder and former CEO Alex Machinsky was arrested this week on charges of wire fraud, making materially false statements to Federal regulators, and other crimes in the aftermath of his crypto exchange’s collapse last year. Similar to other now-defunct crypto services like Blockfi and FTX, Celsius promised its users high returns on crypto deposits. According to Federal authorities, those returns were all fake and the byproduct of Machinsky’s manipulation of his proprietary crypto token CEL and other crypto products like TerraUSD.
Celsius promised returns as high as 17% to users of its “Earn Interest Program.” What was described as a cutting edge marketplace for crypto products funded by the fees paid by unidentified “institutional investors” was actually just a Ponzi scheme, according to the indictment, in which new customer deposits were used to fund withdrawals. When deposits slowed, Celsius inflated the value of its proprietary token with wash trades and limited customer withdrawals. Machinsky also faces multiple civil suits from other Federal regulators for making false statements about the health of his company.
Machinsky was released from jail on a $40 million bond.
Sarah Silverman sues OpenAI for Copyright Infringement
A group of authors, including standup comedian and writer Sarah Silverman, has filed a lawsuit against OpenAI for violating copyright laws. Silverman says that she never gave permission to OpenAI to train its AI models on her work yet ChatGPT is able to produce summaries of the short stories contained in her book The Bedwetter, as well as create new content in her voice and style.
“This is an open, dirty secret of the whole machine learning industry,” said Matthew Butterick, one of the lawyers representing Silverman, according to AP. “They love book data and they get it from these illicit sites. We’re kind of blowing the whistle on that whole practice.”
Generative AI technology like OpenAI’s ChatGPT and Google’s Bard work by “training” computer systems on large volumes of data, in this case the written works for countless authors, and then charge users for access to the systems that can answer questions based on the trained data. Similar lawsuits brought against big tech companies for scanning authors’ works without permission were not successful, such as when Google was sued for digitizing millions of books and making small snippets of them available in search results.
However users never paid for these results, and Google maintained a system that would let authors request to remove their work from the system.
Judge Issues Split Decision on XRP Crypto Sales
A Federal judge made a split decision in a long running securities fraud case this week, ruling that some transactions of the crypto currency XRP constituted the illegal dealing of unregistered securities while some other transactions did not. Despite the mixed news, crypto promoters heralded the decision with unbridled enthusiasm and the price of the XRP token increased dramatically in the minutes after the decision was announced.
The court ruled that “institutional” transactions of XRP, such as when a hedge fund or venture capital firm bought XRP directly from its owner Ripple constituted unregistered securities sales in violation of the law. But the court also ruled that when “programmatic” sales occurred, such as when a Coinbase user purchased XRP blindly through an exchange, no such securities law violation occurred because the buyer “did not know to whom or what it was paying its money.”
The SEC declared that the decision ultimately vindicates its regulation of the cryptocurrency industry as the court ruled that the traditional Howey Test applies to crypto as much as it does any other investment opportunity. The court’s decision also declared over $700 million of XRP sales were illegal, and that the CEO and Executive Chairman of Ripple should be put on trial to determine their personal responsibility in the matter. The judge’s distinction between “institutional” and “programmatic” sales is also a novel ruling and likely to be appealed.
Twitter Sues Scrapers for Stealing Twitter Data
Twitter, now officially known as X Corp., sued four unknown data scrapers in Texas court this week for “severely taxing” its servers and “degrading the social network’s user experience.” The identity of the scrapers is currently unknown, and they are identified only by the IP addresses associated with the accounts they used to pull data from Twitter’s systems.
The lawsuit comes in the wake of significant changes to Twitter’s API and user rate limits that are designed to combat the use of automated bot accounts, increase adoption of Twitter’s paid features, as well as limit the ability of AI companies to train their technology on Twitter user data without paying for API access.
"Several entities tried to scrape every tweet ever made in a short period of time. That is why we had to put rate limits in place," Musk wrote in a tweet about the lawsuit.
Studio Knowledge Base: Demonstrating Product Market Fit
We’re happy to share with you another entry from the Studio Knowledge Base, our newly launched resource of inspiration and guides for startup founders.
This guide is designed for startup founders to better understand the metrics that define product-market fit (PMF). PMF is the degree to which a startup's product meets a market demand.
To early stage founders, the search for PMF is indistinguishable from the business. While seed financing can be raised on ideas and experience, demonstrating PMF is essential to raising later rounds of capital.
Every startup is unique so there is no general formula for PMF. But all formulas must include measurements not just of your startup's product (e.g., user adoption, retention, and feedback) but the market into which you are selling your product (e.g., market size, demand, and competition).
Product-specific indicators of PMF are directly under your control. The rate at which new users adopt your product signals the willingness of the market to try your product but doesn't demonstrate PMF unto itself. It's arguably more important to assess your product's ability to retain customers than it is to get them to try your product. High adoption means you have a great sales team. But if retention is low, you probably need to focus on customer experience and customer service.
Collecting and analyzing customer feedback is one of the most important ways to boost retention and further refine your PMF. A common metric for assessing customer satisfaction, the Net Promoter Score (NPS), measures how likely are your users to recommend your product. Because NPS subtracts the less likely customers from the most likely customers, NPS is not simply a measurement of how many users like your product. It's a net measurement of loyalty.
Investors like NPS because it's a standardized metric that is hard to distort. But NPS is not perfect. Your customers might overwhelmingly recommend your product but what if they could live without it? This could be a major concern if your product is competing in a highly saturated market.
Luckily, you're a founder and can do whatever you want to customize metrics to better reflect your unique value proposition. Instead of the standard NPS, perhaps you will ask your customers how they would feel if they could no longer use your product. If a high percentage of customers report they would be very disappointed if they lost access to your product, you might be well on the way towards achieving PMF even if your NPS is mediocre.
Market-specific indicators of PMF are outside your control, but up to you to prioritize. If you're selling email software, you don't control the number of people who use email. But you will decide which segment of the email software market in which you compete. Your growth into that segment plus the willingness of users to switch from competing products are both great indicators that you are closing in on PMF.
Remember that the journey of demonstrating PMF is never complete. You cannot "achieve" it. Pursuing PMF is an ongoing process of constant refinement and response to changing market conditions. It's up to you to determine which metrics are your guide, and even invent some new ones.