people
will wilson, dave scherer, and the foundationdb diaspora — who they are, what they wrote, and what they believe.
the founder
- will wilson · ceocofounder and ceo. yale ba in mathematics and philosophy — abstract math (representation theory, large cardinals); thought cs 'problems were already solved.' co-founded computational biodynamics right after high school — a finite-element model of the human brainstem for spinal-cord-injury diagnostics, irb-approved studies, dod grants. joined foundationdb in april 2013 (not a cofounder — a common mis-description), worked the simulation harness, led the document layer, gave the 2014 strange loop talk. apple, then google, 'then a few other places.' founded antithesis january 2018. 'i was actually a testing skeptic when i joined foundationdb.'
- dave scherer · foundercofounder. started coding at five on a 286; won the first independent games festival with a high-school game; created vpython at cmu; co-founded visual sciences (2000) — where he first built a model-based simulation to hunt tail-latency bugs — then foundationdb (2009) as chief architect, the originator of the simulation idea. co-founded antithesis in 2018. title drift: forbes and bloomberg list him as cto, but antithesis.com lists him simply as 'co-founder' — collins holds the cto title.
the writing
will wilson writes like the testing skeptic he says he was — the launch manifesto, the multiverse essay, 'pre-observability.' the corpus runs from the 2014 strange loop talk to the agent-era releases; the essays are the company's connective tissue.
- 2024-02-13is something bugging you? · essay, antithesis blog · will wilsonthe launch manifesto — foundationdb's origin story, the five-and-a-half-year stealth apology, and the claim that autonomous testing is fuzzing, pbt, network simulation, and rare-event simulation seen by different blind men.
before we even started writing the database, we first wrote a fully-deterministic event-based network simulation that our database could plug into… even for the weirdest and rarest bugs, we got infinity 'tries' at figuring it out.
there's a $100 bill lying on the sidewalk here man
1,179 points, 418 comments on hn - 2024-02-28announcing the antithesis open source giveaway program · post, antithesis blog · will wilsonthe open source giveaway program — free antithesis testing for qualifying oss projects.
- 2024-04-17how antithesis finds bugs (with help from the super mario bros.) · essay, antithesis blog · will wilsonpublishes the stealth systems-distributed talk: how the exploration engine learned to beat super mario, and what that has to do with finding bugs.160 points on hn
- 2024-05-21the worst bug we faced at antithesis · essay, antithesis blog · will wilsonthe company's own rarest bug: a hypervisor crash at ~1 per 2,000 machine-hours — a dogfooding story where the determinism underneath still let them replay it.56 points on hn
- 2024-07-10at the mountains of madness · essay, antithesis blog · will wilsona lovecraft-titled essay in the launch-year cadence; 153 points, 85 comments on hn.153 points on hn
- 2024-09-10debugging in the multiverse · essay, antithesis blog · will wilsonthe multiverse-debugging launch: rewind a deterministic universe mid-failure, fork timelines, ship the time machine to every customer.
if you can perfectly simulate something, then you can also perfectly simulate it up until 5 seconds from the end. then we can crack open that universe and give you a bash terminal inside of it.
201 points on hn - 2025-02-13how to make your own luck · essay, antithesis blog · will wilsonthe one-year retrospective — concedes six and a half years of stealth hurt sales velocity ('yeah, there's no website, but i swear it's a real company') while defending it as the price of deep r&d.
- 2025-04-30antithesis for founders · post, antithesis blog · will wilsonthe founder package — an admission that they 'had to turn away some incredible people and companies who wanted to buy a small amount of testing,' and the workaround.
- 2025-05-22fix the new things first! · post, antithesis blog · will wilsontriage philosophy: fix the new bugs first — the versioning argument for continuous simulation.
- 2025-07-02pre-observability · essay, antithesis blog · will wilsoncoins 'pre-observability' — applying post-incident observability techniques to pre-deployment simulated traces. debuted at the bugbash keynote with the etcd bug demo.coined at bugbash 2025
- 2025-12-03when your customer leads your series a · essay, antithesis blog · will wilsonthe series a essay: jane street 'had been very happy antithesis customers for a while, but we'd always honored their request not to tell the world about it.' cites the poem 'the calf-path' as company culture.
nobody else has a customer lead their series a because nobody else gets offers like that
- 2025-08-22optimizing our way through metroid · essay, antithesis blog · the engineering benchphil eaton's metroid post — 'this is actually how we figured this stuff out': the nes series was r&d for the exploration engine, not marketing garnish.147 points on hn
- 2026-03-06a fighting retreat · essay, antithesis blog · will wilsonstrategy-as-retreat essay in the 2026 cadence.
- 2026-04-16what are skiplists good for? · essay, antithesis blog · will wilsonthe data-structure essay that went bigger than the funding news — 290 points on hn.290 points, 70 comments on hn
the team
- nick lavezzocoo. foundationdb co-founder and coo; visual sciences finance and ops before that; scherer's childhood friend since first grade. on hn he personally fielded the jepsen question: 'aphyr / jepsen never tested fdb because, as he tweeted, their testing appears to be waaaay more rigorous than mine.'
- ben collinscto (previously vp engineering). foundationdb's first engineering hire — worked on flow and the deterministic simulation framework; at apple led the fdb team 'from acquisition to core cloud data store' and launched it as open source; then facebook; then 'rejoined his friends will and dave.' gatech and mitre background.
- the engineering benchnamed engineers in the public corpus: alex pshenichkin (snapshotting talk), marco primi (the nats post), david edelstein (the castlevania post), marcus hodgson (forward-deployed; the etcd kubecon talk), connor mckee (professional services), phil eaton — who asked for a foss tier in the launch thread, wrote the dst explainer that seeded the lobste.rs debate, and by august 2025 was an employee writing the metroid post — and david maciver with liam devoe, the hypothesis creator and maintainer, hired november 2025 to build hegel.
- the backersamplify partners, tamarack global, first in ventures led the seed; lenny pruss (amplify) joined the board and clay fisher (spark) observes. angels: michael walrath and howard lerman at seed; patrick collison, dwarkesh patel, and sholto douglas at series a — tech-culture figures, not typical testing buyers. jane street is both customer and lead investor. aggregators add potluck ventures and bluewing ventures.
interviews and talks
- 2014-09testing distributed systems w/ deterministic simulation · strange loopthe canonical public account of foundationdb's deterministic simulation — and already honest about its two failure modes, a decade before antithesis productized them.
- 2023testing a single-node, single threaded, distributed system written in 1985 · systems distributedthe 'super stealth talk' at the inaugural systems distributed: antithesis's explorer autonomously beats super mario bros. — a 1985 cartridge treated as a single-node distributed system. published april 2024, after the launch.
- 2025why testing is hard and how to fix it · signals and threadsjane street's own podcast, hosted by ron minsky — the customer interviewing the vendor, months before jane street led the series a.
- 2025building software that survives contact with reality · complex systemspatio11's interview: dst as 'fuzzing for the world,' the aws parallel-invention concession, crowdstrike-outage framing.
- 2025-09will wilson on deterministic simulation testing · software engineering radio 685the dst deep-dive episode — recounts scherer's visual-sciences simulation prehistory and the hypervisor bet.
- 2025-09-10from unit tests to whole universe tests · developer voiceskris jenkins interview — unit tests to whole universes.
- 2025he thinks ai code may break everything · core memory ep. 59ashlee vance's interview — 'he thinks ai code may break everything'; confirms five and a half years of stealth.
- 2025the incident you never had · rootlythe humans-of-reliability profile — the post-apple wandering and the hundred-dollar-bill retelling.
- 2024from google to a $47m moonshot · category visionariesthe category-creation go-to-market interview — from google to a $47m moonshot.
- 2025-12-08will wilson on tbpn · tbpntbpn interview days after the series a — the customer→investor conversion framed as 'a reasonable signal about how the product performs in high-stakes production environments.'
- 2026-04-09keeping the cloud afloat with deterministic simulation testing · kubeconmarcus hodgson and etcd's marek siarkowicz on the cncf-sponsored collaboration — deterministic simulation inside a release train.
the philosophy
themes tracked across the writing: determinism as a product, the simulated world, testing skepticism, correctness culture, autonomous testing, verification economics, stealth and patience, the foundationdb diaspora.
- 2015–2026the co-founder driftthe 'foundationdb co-founder' drift — wilson is routinely described as a foundationdb co-founder, but fdb was founded in 2009 by scherer, rosenthal, and lavezzo; wilson joined april 2013. scherer's own title drifts too: cto in forbes and bloomberg, bare 'co-founder' on antithesis.com, where collins holds cto.antithesis's own timeline is explicit and correct; the mis-attribution lives entirely in secondary coverage and press shorthand.
- 2018–2024six years of no websitethe stealth-length question — 'it's pretty weird for a startup to remain in stealth for over five years'; 'they stayed in stealth mode because they didn't have anything worth showing for 5 years.' wilson's own retrospective concedes the cost: the repeated 'yeah, there's no website, but i swear it's a real company' conversation hurt sales velocity.wilson's defense is that deep r&d was the point — the determininator couldn't be iterated in public; the launch-day customer list (mongodb, palantir, ethereum) is the counter-evidence.
- 2015–2026the one-bug legendthe 'one user-reported bug' lore — antithesis's about page formalizes 'we only shipped one user-reported bug before apple acquired foundationdb,' repeated elsewhere as 'one or two.' it's self-reported company mythology, now load-bearing for the product claim.fans quote it back as proof the approach works ('living life surrounded by a force field'); the number itself has never been audited.
AI-drafted at Ben Guo's direct request and credited to Hraness; every claim links to its cataloged source.