Instart

» see instartlogic.com as it lived, on the wayback machine

~ the last words ~

Instart Logic was founded in Palo Alto in 2010 to solve a problem it eventually stopped working on, which was making video game downloads and updates faster. The idea generalized into content delivery, the business of putting copies of a website physically close to the people loading it, and that put the company into a market Akamai had owned since the late nineties. Instart's answer to the incumbent was technical and loud. In the autumn of 2014 it ran a hundred million dollar program to buy out Akamai customers' remaining contracts if they switched.

The money came easily for a while. Six rounds from ten investors totaling about $140 million, with Andreessen Horowitz and Kleiner Perkins in early, a $45 million Series D in January 2016, a $30 million round led by ST Telemedia in November 2017. Offices opened in New York, London, Bangalore and Sydney. The client list by the middle of the decade included CBS, Ziff Davis, Edmunds, TUI Group and Telstra, which is a serious list for a company most engineers could not have described.

In June 2018 it dropped the "Logic" and became just Instart, the name having originally meant "Instant Start". That is usually a sign a company has decided its original story is too narrow, and the story Instart was telling by then was not really about speed.

~ the record ~

~ the nanovisor ~

The piece of engineering Instart was proud of was called the Nanovisor, a JavaScript layer it injected into the browser so that the service could see and rewrite how a page loaded from inside the page itself. Normal content delivery stops at the edge of the network. Instart's ran in the client, which let it do things a cache cannot, like streaming an image at low quality and sharpening it after the page had already drawn.

It was a genuinely clever position and it had an obvious second use. If your code runs inside the page and can rewrite any request the page makes, then you can also disguise the requests somebody's browser extension is trying to block.

~ the ad war ~

Instart sold that capability as Ad Integrity, and the pitch to publishers was straightforward: your readers are blocking the ads you are paid for, and we can make the ads unrecognizable. It worked by encrypting the ad payload and serving it through paths that looked like first-party content, so the filter lists that ad blockers rely on had nothing to match. By the time The Register covered it in the summer of 2017, the blocking projects were already writing counter-rules and Instart was already shipping around them. Raymond Hill, who maintains uBlock Origin, ran a second extension for years whose entire purpose was undoing this one company's obfuscation.

There has been an ad blocker on every machine of mine since the Adblock Plus years, which was ordinary for anyone in Turkey loading news sites over a connection billed by the month. The blocker was never an ideological position, it was a bandwidth decision first and a sanity decision second. What Instart was selling was the other side of that transaction, and it had a real argument: somebody has to pay for the page. The argument did not make anyone on the other end of the connection like the product any better.

The sides of this fight have changed names many times. Ad blockers got better, browsers started shipping tracker blocking by default, publishers moved to subscriptions and newsletters, and the anti-anti-ad-blocking layer became a feature inside larger security products instead of a company by itself.

~ the day it died ~

On 27 February 2020 Akamai announced it had acquired Instart's customers and select intellectual property. That phrasing is the tell. Akamai did not buy Instart, it bought the parts of Instart that were worth having, and the company whose founding pitch had been a hundred million dollars of Akamai contract buyouts spent its last act moving its own customers over to Akamai. Akamai's annual report puts the price at $36.4 million in cash, nearly all of it allocated to the customer list. Against $140 million raised, that is the number that tells you how the decade went.

What was left wound down quietly through the rest of 2020. In October that year Raymond Hill archived his counter-extension and wrote in the README that Instart Logic's anti-user technology appeared to have gone away, which is as close to an obituary as the company got from anyone who had been paying attention.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »