Back to the Future

1. The Future of the Past

In the 1950s and 60s, people confidently predicted a future that mostly never arrived. Computing was the great exception: Moore's Law held, power consumption fell, connectivity kept improving, the Dick Tracy wristwatch effectively became the iPod nano, and Arthur C. Clarke's 1956 vision of a networked world came true. Almost everywhere else, the future quietly failed.

Sometimes the best way to think about the future is to think about the way the future used to be.

• Weather: prediction stayed unreliable, and cloud seeding was abandoned as too dangerous
• Transportation: flying cars never came; traffic looks the same as it did decades ago
• Robotics: we expected a general-purpose robot butler and got the Roomba
• Nuclear power: promised abundance, but it proved more dangerous than hoped and progress stalled

The retrofuture move is diagnostic, not nostalgic: find fields where progress stalled or reversed — thorium reactor research, for example, was sidelined when military priorities favored uranium and plutonium — then ask why they failed and attack the problem differently with modern tools. Copying the past directly never works.

2. Where the Future Failed: Four Frontiers

The Space Shuttle program was oddly Pareto inferior. It cost more, did less, and was more dangerous than a Saturn V rocket.

The common thread is that none of these companies copied the old dream. A cheap robot that rolls in ahead of a SWAT team — criminals often surrender to it on sight — beats a robot butler. Compressed air sidesteps the hunt for a miracle battery chemistry. And The Climate Corporation thrived partly because agricultural technology sits outside Silicon Valley's field of vision, so few people were even looking.

3. Recruiting for Hard Problems

Asked whether radical uniqueness hampers recruiting, the panel largely said the opposite. SpaceX gave a small team of young engineers real responsibility that cost-plus aerospace bureaucracies never would. The Climate Corporation competed with Google and Facebook for quants by arguing that predicting weather for farmers matters more than optimizing games about virtual livestock. RoboteX's pitch is simply that it builds robots that save lives — and customers reinforce that message daily.

• Mine adjacent industries with startup-compatible cultures: LightSail found scarce compressor talent in auto racing, after screening 1,000+ resumes and running some 400 interviews
• Sell impact and ownership, not salary — seasoned aerospace veterans with big salary expectations were harder to land than hungry young engineers
• Hiring academics works but takes patience: The Climate Corporation employed about 20 PhDs in their first jobs outside academia and had to teach research-to-product thinking
• Screen for cultural fit in both directions, and use professors' networks for honest reference checks on candidates

4. Why Can These Be Solved Now?

1. 1

   LightSail: history lost its notes

   Compressed air storage was tried in the 1870s, before the electrical grid — even the water-spray cooling idea. Why it failed is unrecorded; you can't debug the mental processes of long-dead inventors. Technology is deeply path-dependent, and modern materials make the retry worthwhile.

2. 2

   Climate Corp: compute caught up

   Crop insurance dates back to biblical times, but past attempts lacked data and computational power. With a thousand CPUs and granular remote sensing, the actuarial models finally work — the hard part now is boiling the output down into something a farmer can act on.

3. 3

   RoboteX: drop the sci-fi bias

   Science fiction biased everyone toward legs and humanoid forms, but tracks are simply better. Riding Asia's commodity computer-component supply chain and substituting plastics for machined aluminum collapsed costs to a fraction of the old approach.

4. 4

   SpaceX: rebuild the cost structure

   Incumbents wrote specs and outsourced everything, layering markups and friction. SpaceX went clean-sheet and vertically integrated, pushed composites hard, developed new welding techniques, and even resurrected a forgotten fuel-injector design. The number of innovations was staggering.

Vertical integration emerged as the shared theme. Build key parts in-house and they do exactly what you need and no more — no supplier markups, no compromise specs. Even the credible threat of in-sourcing changes supplier behavior: Fong's tactic is to show suppliers you could build it yourself, then let their competing quotes discipline each other.

5. Distribution, References, and Exits

RoboteX skipped the conventional military-contract route and sold bottom-up to local police departments instead. Staying fully private kept the company free of government strings, while small deployments generated daily user feedback, real revenue, and strategic freedom — with commercial hazmat businesses as the next ring of expansion.

• The bigger the ticket, the slower and less objective the sale; buyers always ask who else has bought, and having no references can kill the deal
• The $100M single-contract fantasy almost never happens; good enterprise startups instead grow 50–100% a year for a decade, e.g. $5M in year one, then doubling
• Your next biggest deal is realistically about 2x your current biggest — no customer signs on for 10x your largest prior deployment
• The same logic binds VCs raising funds: there is no single whale LP; investors herd, each affected by the others in hidden, unspoken ways
• Fong's exception: the perfect first customer is a desperate one whose current supplier just failed — motivated, fast, and grateful

If you want advice, ask for money. If you want money, ask for advice.

On exits: the best sales are disguised, so a founder who wants to sell should act uninterested. Boards must weigh serious offers, but M&A is driven either by efficiency grinding (merge two banks, fire half the staff) or by product synergy — and genuine synergy is rare. PayPal/eBay was the exception, not the rule; the more unique your technology, the less likely an acquirer truly complements it.

6. Funding the Unfamiliar

Asked how to fund hardcore technology, Fong's blunt answer was that it is extremely hard. Musk put roughly $100M of his own money into SpaceX and would have spent everything if needed. When NASA required outside funding in 2008, Founders Fund invested — after rockets had already been built and flown — while the rest of Silicon Valley called it crazy. Many investors recite disruption rhetoric; few partners have the internal clout to actually back it, so founders must find investors whose convictions genuinely align with the mission.

Thiel's unfamiliarity paradox: things that are truly different are hard to evaluate. When Musk pitched a rocket company, his relevant experience was zero — but so was everyone else's, since nobody had done rockets in 40 years. Contrast iPhone gaming, where every founder cites past titles, every VC cites a gaming portfolio, and competition is ferocious. In unknown territory the evaluative bar disappears — an advantage for teams that prioritize learning over process.

• Founder skin in the game substitutes for the missing evaluative framework — personal capital and commitment are the signal
• Court investors for their convictions, not their risk-taking slogans; portfolio 'diversification across different things' is often just a story
• Unfamiliar domains carry a hidden bonus: the same strangeness that repels capital also repels credible competitors — the quiet payoff of every retrofuture bet

Then vs. now (2026)

Self-check quiz

Pick an answer to reveal the explanation.