HN user
Of all the times I've seen Internet votes made to have a real-world value, the bad parts of this contest were among the mildest of potentially bad outcomes there could be. Went better than expected honestly.
Really, really like this. Been promising myself I'd build something like this eventually, other projects always got in the way. Glad you did it.
Question to YC - how representative are these, as a sample of the applications YC typically sees? I've always wondered what YC's raw unfiltered application sets look like. Is this more or less skewed in any atypical ways?
Sometimes I wish HN used these nonsense stories as a gullibility test. Revoke the upvote powers of anyone who upvotes this pure nonsense.
they really just had 10 x 1 year of experience
Brilliantly said.
The whole world is full of founders. Being in the SV area is the rare advantage, not the other way around.
Yes, the FROM spoofs your name so that the person will believe it's being sent directly from you in most email clients, and the Reply-to goes directly to your email. These are the email headers:
From: YOUR_NAME <team@angel.co>
Reply-To: YOUR_NAME <YOUR_EMAIL@gmail.com>Edit: Oh cool, Hackernews still has this account rank-banned so the comments always stay at the bottom of the page no matter how many upvotes they get.
So happy that I'm not the only one who was extremely surprised and upset about Angel List's impersonation tactics.
In my experience, 100% of the people who received this impersonation email that angel.co sent, thought it was written by me. If the system wants to send a message on my behalf, it must AT LEAST show me what would be sent with my signature on it, and allow me to confirm or cancel any email that's signed with my full name before it's sent. Angel List is sending out emails with your signature and other impersonation tricks, totally behind your back.
I've had people impersonate emails from me in the past as well. Even if the emails aren't offensive, just the idea that someone else now thinks they have my permission to send whatever they want to, while impersonating me behind my back makes me feel extremely uneasy.
Here's the impersonation email they send to cofounders when they're added to a profile. As you can see, signing the email with the user's full name, as if the user had written the letter, is a clear impersonation that the user never consents to.
FROM: <MY_FULL_NAME> <team@angel.co>
Hi <RECEIVER_NAME>,
I'm building an AngelList profile for <PROJECT_NAME> and need your help.
Can you do me a favor and confirm that you're a founder by clicking here:
https://angel.co/l/54223434234234583948593475
This is important to us; confirming that you're a founder will help us build our profile.
Thanks,
<MY_FULL_NAME>
P.S. If you would like to visit our profile first, go here: https://angel.co/<PROJECT_NAME>
From: <MY_FULL_NAME> <team@angel.co>
Reply-To: <MY_FULL_NAME> <MY_EMAIL@gmail.com>Nope. They're literally measuring two different things, return rates vs "would tell a friend" rates.
Whether those two happen to coincide is quite product-dependent. Finding examples where they significantly diverge is trivial.
Retention and NPS seem orthogonal.
How skeptical? How much BTC will you, or anyone, be willing put down on either of these:
1) Just bet me that it's not real, and I'll give you some very strong evidence.
2) How about I write a web-app that exposes HN's rank-banning - who's getting rank-banned and what for. You'd be surprised at how it's being used. I found a way of proving it that I'm sure you'll agree is quite accurate, for cases where this approach can be used.
:)
Edit:
Re: lucb1e, the reason is that I'm far from being the only person who's rank-banned / dead here. Sometimes several comment in the same thread.
Re: scott_s, Hey man, no need to insult me. If you don't want to see the hard evidence, then I won't show it to you. Don't worry.
If "goodness", aka entertainment value to the reader is what you're judging based on, then yes, I suppose I agree. HN is good for reader entertainment.
In that case, HN is basically the equivalent of a tabloid. Written civilly, while communicating very bad things about whatever it's talking about, while the reader is throughly entertained.
Is that really appropriate for things that are just launched? Out of infinite possibilities, the number of things an app doesn't do are practically infinite. Is there really value in always whining about all the things a just launched app doesn't do yet?
Why am I talking about only new apps? Because that's what these kinds of sites are always going to mostly be about. Scoops on new things - news - is what drives votes on these kinds of news sites. Therefore, you're always going to get mostly new, incomplete things.
Encouraging people to point out the flaws, on a platform that only votes new flaw-filled brand new things to the top -- results in a very predictable outcome: incomplete stuff always voted to the top and people always whining about the incompleteness.
I've got to wonder to what degree this approach is really hiding the "real" language in a higher-level encoding.
E.g. you can write english in morse code, but the combination of those dots and dashes is where the "real" language is, instead of the dots and dashes themselves...
Let's be introspective. Was the "be civil" experiment really a success? Is it really the specific words that matter, or how the words make the other person feel that matters most?
I'd contrast this approach with e.g. Product Hunt's community standards, where you're expected to be inclusive, make others feel good, and try to work as a group of makers who are "all in it together" instead of constantly arguing and putting each other down. Little mention of civility of the message delivery is mentioned at all - it's all about how the message affects the other person.
I know I'll get flack for saying this here, but I've found PH's approach to make far more sense and be 1000x better in practice.
(By the way, last I checked, I'm still ranked-banned for pointing out YC's reversal on their stance about board members, so this comment will stay at the bottom of the page no matter how you vote. - Edit, or the comment will tie for bottom of the page, along with the other rank-banned and dead users.)
I'm genuinely confused as to what experience he's talking about, as it sounds like the complete opposite from what I've experienced. I do know that different companies may have implemented their flat structure in completely different ways though, so that's why I ask.
The question was actually not a rhetorical jab :) But thanks, I'll keep that in mind.
You've cut off the part of the sentence that I was replying to.
Did somebody hire you? He's your boss. Can somebody fire you? He's your boss. Can they tell you to show up at work at certain times, or prevent you from taking the afternoon off to take your kid to the doctor? He's your boss. Can someone control whether or not you get a raise? They're your boss.
What are you talking about? In the flat-managed places I've worked in, a group a peers periodically voted on everyone else's performance. If you had to leave the office for something, you announced it (but never asked permission for it ) on the group mailing list.
Does somebody extract the excess value from the of the labor you provide, and claim the right to do so by more efficiently organizing the workplace so as to produce value most efficiently?
The one I'm talking about was a publicly traded company, so no. The technical "owner" of the company had little to no say in the management. Maybe this place was not the typical employer. It was in a field where talent is extremely valued. The employees ruled.
If you ask me, it's quite a bit harder to brown-nose 20 of your peers, than just one boss. The dynamics were vastly different than a non-flat, 1-boss system.
Do you not believe that's how mature businesses work?
The people demanding you pay them tens of thousands per year, for them to tell you how to learn, say it is. We can trust them, right?
Likewise, a cap on a convertible note may be used by investors in the next round as an anchoring point to restrict the valuation in the next round.
Both of these effects are what a layman would think "shouldn't be real". From experience, both are extremely real.
Too bad that the TV shows present these as sort of jokes. That disappoints me. No joke, these are dead real.
Companies valued purely based on revenue streams alone will typically sell for 2x to 5x their yearly revenue.
This is a big part of why it's SMART for most consumer startups to avoid earning revenue at all. Doing so will often cap their potential exit price, in practice.
Still, this at least was at the higher end, for purely revenue-based valuations.
Just doesn't make sense. Sure, quickly eliminate someone's role, but why immediately fire the person? Why not even try to transition the person to a new role? Was she employed as an independent contractor?
Reddit clearly has both plenty of money, and plenty of work to be done. This explanation doesn't add up at all.
Extremely inefficient markets. Traditionally, VCs have been able to collude together to prevent startups from increasing in valuation "too fast". Unlike more efficient and less manipulated markets, which tend move in fast discontinuous jumps, VCs have been able to slow the top-performing companies to a slower exponential curve.
Because they've typically been able to slow this curve, waiting until the next round usually hasn't caused them to lose nearly as much upside loss as it would, were the market more efficient.
With IPOs being pushed back, and more sophisticated IB investors moving into much earlier rounds now though, this is finally starting to change.
Keep this in mind whenever you hear VCs shouting "bubble", just because some top startups aren't following the smooth exponential curve in valuations, that any competent trader or economist could tell you should never exist in an efficient market.
Strongly disagree. Beautiful design is a failure anti-pattern. It's no coincidence that many of the best tools are ugly. AWS is a set of tools.
It's really the most fundamental rule for every type of machine learning task -- there should be both a (1) test set and (2) a separate holdout validation set. No matter if the task is classification, sequence prediction, regression, manifold learning, clustering, whatever.
Really surprising how so few amateur traders understand this fundamental principal. Scientists too.
Oh, I'm sure that startups know very well how poor and incomplete that methodology is. They just present this non-forward tested approach to fool the end users who don't know any better.
Can you show the forward performance, starting from the date when you chose those 4 parameters - 165.12, 172.44, 164.23, 156.02?
As you know, backtested results with carefully tuned magic numbers can potentially mean nothing.
This puzzled me that, for the past few years, YC has wanted so badly to invest in more or less traditional style wealth management companies. It's so extremely hard to outperform existing methods in terms of returns. For user experience, these are the kind of things that are best if totally automatic, with no interface at all.
So I guess they just compete purely for growth and that's it - like the commercials for popular drugs on TV, it's an industry 99% about advertising?
SV tech + extremely risk averse, mature, competitive space = doesn't seem to be a good fit.
Looking at those generated images, I realized that many of those training photos seem to apply: