40 comments

[ 1524 ms ] story [ 3805 ms ] thread
I wonder if the prospective company's G Suite data gets crunched.
And risk losing thousands of existing paying customers if it ever came out?

Not a smart choice, considering you're betting on the company you're illegally spying on to get profitable at some point.

It’s not illegal if it’s in the ToS.

And even the data of enterprise customers is used to improve Google’s machine learning algorithms, according to those.

So not illegal, just entirely immoral. But when has that ever stopped Google?

Exactly. These days this is exactly how Google works. Ask an AdWords, Analytics, or AdSense user who slowly realizes that Google has full access to their business data for any purpose.
I always thought if they were not doing something similar for key hires. Not for engineers as it would most likely leak eventually, but like if they are going to hire a VP maybe the founders or someone else has hire_vp_or_not.py that parses their e-mails from previous jobs, etc...
Wow. They can't bother to have people working in customer service, and now they can't even bother to have people working on where they invest their money?

Please tell me that the execs are next to be automated.

Oh cmon, like Ycombinator doesn't do the same. They will probably have a bunch of datapoints an features to evaluate each batch. Be it a human or a machine, it doens't matter. Only the accuracy.
(comment deleted)
(comment deleted)
"Can't bother"

There are numerous areas, including hiring, in which algorithms are provably better at some jobs. The fact that you choose to do something one way doesn't indicate that you "can't even bother" to do it another way.

I feel you've never dealt with Google's customer service, otherwise you wouldn't be saying this. They clearly cannot bother having competent customer support staff, or staff with the agency to actually do anything. And I feel that's largely because they rely so much on their automated systems to try and take care of it.
They can automate all executives, apart from founders and the current chairman.
> Inputs into "The Machine" include round size, syndicate partners, past investors, industry sector and the delta between prior valuation and current valuation. The algorithm then ranks deals on a 10-point scale, with green said to represent 8 or above

I'm sure there are more inputs than this, but from that list you'd imagine they basically pick deals based on who else is investing. which is not that different from many other VCs. at least in biotech, i typically see GV co-investing in deals led by other top notch VCs

> they basically pick deals based on who else is investing.

Another phrase for that is "herd" mentality, which will hopefully result in de-risked "safe" investing, but VC's are supposed to be looking for true breakout possibilities. If LP's wanted safe, they'd buy Treasury notes.

VCs are looking for breakout returns for investors. A syndicate of top VCs can be a kingmaker event that also discourages other investors from backing would-be startups.
Now they have an alibi for their herd mentality
There are huge gains to be had in herd-ing. For instance, if everyone herds towards Uber instead of Lyft, you effectively pick Uber as the winner by merely herding, instead of merit or market economics
And in that case, the worst possible outcome is to have many competing funded companies in the same space. Merit doesn’t enter into it much.
That would explain their Juicero investment.
If a significant chunk of VCs invest based on an algorithm following the investments of other VCs running the same algo... positive feedback loop. Juicebox squeezer receives $120m investment.
You can't do this with crypto startups
Given how that market is doing, that might be to its credit.
(comment deleted)
Isn't that what a lot of Excel spreadsheets do? I think investments and acquisitions have used algorithms for a long time.
By 2014, "checklists" and "investment criteria" had rebranded themselves as "algorithms" across most of finance.
And by 2018 they have been re-branded as "machine learning algorithms."
Just a fancy formula that works based on people generated input. Without people input said formula would be useless.
The article has uncovered sources that claim the algorithm makes the ultimate decision and other sources that claim it doesn’t.

If it does make the ultimate decision and not just for political reasons then this is very interesting. Having input data that is sufficiently informative is important on a number of levels. Firstly this means that it is possible to pick winners on the basis of other VCs etc Secondly it means one doesn’t have to be personally concerned with the story if others can vet it for you.

If the machine doesn’t make the ultimate decision then — as others point out — it’s just old fashioned screens and checklists with a new interface.

If the main input is quality of other VCs, then at some point a VC has to decide to invest based on fundamentals rather than what other investors are doing. a group of "fundamental" investors with good track records then would dictate what the rest of the market invests in.

you sort of see this dynamic play out in reality. YC is an example: they invest early, before other investors often, so they cant rely on other investors as a signal. they've done well though, so many investors follow them. there are more follow-on investors than successful "fundamental" investors, so there's often a valuation step up when follow on investors join that benefits the fundamental investors

same thing plays out in biotech. theres been a massive influx of capital into biotech VC, but not a big increase in the number of funded startups. most startups that go on to raise money are seeded in house by a handful of VCs. these VCs then fund the series a. they get big step-ups for series b and beyond deals and capture nice returns

im working on a more rigorous analysis to understand whether these anecdata are true in reality

I agree with your description of he dynamic at play. It raises two questions:

1. Is the money that the startup attracts responsible for its success? In other words if a mediocre company goes through Y Combinator and then attracts a $55 million round, is it more likely to succeed than a great company that does not? (Let’s day the mediocre company doesn’t squander the cash wastefully but slowly looks for the product market fit)

2. Are there fundamentals that can be distinguished from an “observer effect.” Suppose everyone believes that a company coming out of Stanford is more likely to succeed than one coming out of (say) Babson. Does believing it make it true because the company attracts more money in each round?

These two thoughts are variations on a theme of the role of signaling in picking out fundamentals.

Edit: I should also point out that GV might also use “true” fundamentals like search results, trends, etc

there is very little public data available on startups as compared to private companies. so id imagine for an algorithm to be useful there would have to be a lot of proprietary data. further, id imagine a lot of this data is somewhat subjective -- ratings of management team, market potential (when a market is still not defined enough to quantify), etc. so its possible that many of the quantitative inputs have some degree of subjectivity -- the human element is still very present, its just hiding behind data
Anyone know of open datasets around early stage startups and success? Guess might be something mashing up crunchbase, angel.co....
This is nothing new, most funds treat portfolio construction as an optimization problem, where the objective function is some risk over return metric.

"The Machine" is an optimizer, all they've done is build one that looks at early stage companies.

G also has inside access to a lot of information from search and email. Could be using that to optimise their portfolio.
Best-performing seed round investors:

     7.5724 . . . . True Ventures 

     6.6294 . . . . Accel Partners
     6.1189 . . . . Seedcamp 

     4.7562 . . . . Nxtp Labs
     4.7474 . . . . Y Combinator
     4.4846 . . . . Softtech Vc
     4.4720 . . . . Gecad Group 2
     4.4720 . . . . Radu Georgescu
     4.4246 . . . . Creandum
     4.3653 . . . . Sv Angel
     4.3636 . . . . Google Ventures
     4.3016 . . . . Greylock

     3.9381 . . . . Baseline Ventures
     3.8225 . . . . First Round Capital
     3.7816 . . . . Mitch Kapor
     3.6702 . . . . Atomico
     3.5938 . . . . Felicis Ventures
     3.3963 . . . . Freestyle Capital
     3.2857 . . . . Yee Lee
     3.2183 . . . . Naval Ravikant
     3.1193 . . . . Boldstart Ventures
     3.1038 . . . . Keadyn
     3.1007 . . . . Birchmere Ventures
     3.0376 . . . . Betaworks
     3.0012 . . . . Hyde Park Angels 

     2.8730 . . . . Ff Angel Llc
     2.8578 . . . . Slow Ventures
     2.8268 . . . . Rose Tech Ventures
     2.8052 . . . . Chicago Ventures
     2.7709 . . . . I5Invest
     2.6940 . . . . K9 Ventures
     2.6469 . . . . Founders Co Op
     2.6132 . . . . Oleg Tscheltzoff
     2.5637 . . . . Nyc Seed
     2.5558 . . . . Ta Venture
     2.5148 . . . . Geoff Ralston
     2.4748 . . . . Cnm Ventures
     2.4327 . . . . Genacast Ventures
     2.3829 . . . . Wi Harper Group
     2.3526 . . . . Allen Morgan
     2.3171 . . . . W Media Ventures
     2.2827 . . . . Oliver Jung
     2.2532 . . . . Amplify La
     2.2509 . . . . Chris Devore
     2.2481 . . . . Dharmesh Shah
     2.2291 . . . . Innospring
     2.2290 . . . . The Accelerator Group
     2.1947 . . . . Gary Vaynerchuk
     2.1791 . . . . Crunchfund
     2.1772 . . . . Jon Callaghan
     2.1700 . . . . Oca Ventures
     2.1666 . . . . Kfw
     2.1652 . . . . Plataforma Capital Partners
     2.1136 . . . . High Tech Gruenderfonds
     2.1079 . . . . Marc Simoncini
     2.0933 . . . . Tom Mcinerney
     2.0908 . . . . Steve Anderson
     2.0879 . . . . Battery Ventures
     2.0590 . . . . New York Venture Partners
     2.0256 . . . . Emerge
     2.0038 . . . . Andy Appelbaum
Worst-performing seed round investors:

    -6.9802 . . . . Wayra

    -5.8839 . . . . Start Up Chile
    -5.0805 . . . . Startupbootcamp

    -4.0593 . . . . Jumpstartinc

    -3.8784 . . . . Sosventures

    -2.7610 . . . . Start Engine
    -2.6717 . . . . Social Starts
    -2.5748 . . . . Ff Venture Capital
    -2.5641 . . . . Ben Franklin Technology Partners Of Southeastern Pennsylvania
    -2.5460 . . . . Ace And Company
    -2.4309 . . . . Quest Venture Partners
    -2.0690 . . . . Masschallenge
Best-performing series A round investors:

     7.0711 . . . . Greycroft Partners

     5.5854 . . . . Venrock
     5.4365 . . . . Trinity Ventures

     4.9683 . . . . Intel Capital
     4.9600 . . . . Scott Banister
     4.9400 . . . . Kleiner Perkins Caufield Byers
     4.8362 . . . . Redpoint Ventures
     4.6803 . . . . Ron Conway
     4.5240 . . . . Sv Angel
     4.3989 . . . . Crosslink Capital
     4.3097 . . . . Storm Ventures
     4.0204 . . . . Shasta Ventures

     3.9124 . . . . E Ventures
     3.6809 . . . . Leapfrog Ventures
     3.5279 . . . . Khosla Ventures
     3.3344 . . . . Valor Capital
     3.2867 . . . . Accel Partners
     3.2164 . . . . Austin Ventures
     3.1917 . . . . Accelerator Ventures
     3.1199 . . . . Inveready Technology Investment Group
     3.0892 . . . . Signia Venture Partners
     3.0283 . . . . Mangrove Capital Partners

     2.9607 . . . . Alliance Of Angels
     2.9039 . . . . Ventures West
     2.8911 . . . . Novartis Venture Fund
     2.8627 . . . . Carmel Ventures
     2.8622 . . . . Brightspark Ventures
     2.8440 . . . . Omnes Capital
     2.8099 . . . . Reid Hoffman
     2.7943 . . . . Alta Partners
     2.7847 . . . . Partech International
     2.7432 . . . . Balderton Capital
     2.7391 . . . . Helion Venture Partners
     2.7341 . . . . Wellington Partners
     2.6948 . . . . Mercury Fund
     2.6859 . . . . Sofinno...
If you don’t mind sharing, where did you get this data?