Hacker News Reader

Comment by JshWright | Hacker News Reader

In the world of high angle rope rescue, we have a concept called the "whistle test". The idea is that if someone were to randomly blow a whistle at _any_ point, and everyone let go of whatever they were holding, that no one would be dropped.

It takes a lot of thought and planning to make sure you're 'default alive' in all circumstances. It slows you down, and it requires you to think through the implications of every decision, big and small.

This sounds like a fairly similar notion...

View on HN

Here's an interesting thought in response: (Keep in mind I know jack about rope rescue, so I'm making assumption as to what it is)

Imagine you had 100 people do a "rope rescue race". Half of them used all the safety mechanisms (they would pass the "whistle test"). The other half didn't use any safety/redundancy (they would fail the "whistle test"), but instead relied on their skill. In the end, it's likely that a few of the high-risk ones would die. But it's also likely that many of the high-risk ones would survive, and also that they would be the winners of the race. The safer, lower risk ones would eventually get there, but far behind the high risk ones.

Startups are the high-risk ones.

You raise a good point. It's not a matter of caution and being double careful. The point of the essay is the drive towards self sufficiency, and being realistic about how investor expectations change in successive rounds.

Perhaps: don't think there's a rope backing you up if it isn't there.

> Perhaps: don't think there's a rope backing you up if it isn't there.

If you use the metaphor of rock climbing, you'd actually rehearse your route with ropes/safety system, before you would ever try the route without. When the stakes are so high that it's you're life you're gambling with, is it really worth it for most to go so blind out there?

It's either that, or you go without a rope, when the difficulty is far below your limit (in other words, you have amassed a lot of experience)

This is a good analogy in one sense, but I think the probabilities in your example are reversed from the startup case: for high-risk startups, only a few survive, but they're the winners.

And of course, that's what makes them better on average for investors (who spread risk across their portfolio) than for founders, who have all their eggs in one basket.

As pg says here, as a founder, you want to survive.

That's psychotic.

I don't rope rescue, but I sure do rock climb. You can rock climb with ropes or without. Ropes = a large safety margin, and the majority of people who climb, utilize them, as part of their safety system. Unless you're pushing your own grade level, the actual safety system isn't actuated - you're just using it "just in case" (like, "just in case you fall from a cliff, 500 feet up - gee it's nice I'm tied to something).

There are also those who climb without ropes. Perhaps you've heard of some of them [0], and most likely, because they actually climb without a rope. High Risk. Your idea of startups, let's say.

But that's just what the public sees. The free soloist has to be absolutely comfortable to go without the safety margin of a rope. Because of that, the majority of their climbing is done, still with a rope. When they go ropeless, the climbs they do are much, much easier. Yes, the risk is there, but also there is the understanding, clearly, what the risk is really about.

The death rate of climbers that are unwilling to ever use a rope, because they find that high risk = high reward would be close to 100%. But those who do survive are not going to be doing too well - they're never going to progress.

So, perhaps startups have this illusion of high risk, but maybe also they're also kidding themselves. They're burning through their lives, and offloading the real risk onto their safety systems - whoever is putting up the money. And they can do that (the investors), as they've diversified their portfolio enough that risk is spread around.

[0] https://www.youtube.com/watch?v=SR1jwwagtaQ

Reminds me of this: Make the climb like the child did - without the rope.

https://www.youtube.com/watch?v=KXxw-zXRqOs

The rope makes you weak!

The same movie also lets the protagonist heal from a incapacitating spinal injury in a dusty pit -

So I'm not sure if this is a pro- or anti-rope statement :)

The analogy breaks right at the start, start-ups are not 'fatal' in the same sense that rock-climbing is fatal. So the GP is right, the race really would be won by those willing to cut corners and sacrifice safety and procedure even though statistically quite a few of those would end up 'dead' for those exact same reasons with possible fall-out for their end-users as well (hacked, data loss and so on).

The strategy widens the bell curve and lowers it, whereas the rock-climbers try very hard to narrow it to where there are no outliers and everybody makes it to safety.

Imagine a strategy where every start-up would survive in the long term, there would be very few outliers in such a situation and that's why start-ups that are comfortable with some risks at the expense of safety but a huge increase in productivity will be more likely to be amongst the winners than those that play it safe all the time.

I don't think this is a huge problem anyway because very few people doing start-ups are risk averse.

I can't say I agree with this analogy, unless we assume that the course is a race, which isn't true for most climbs.

I'm not sure if that's an apt metaphor because the consequence of failure is much, much lower on the startup side.

If failing at a startup means you'll never have another chance to succeed, I bet the norms for risk tolerance would be entirely different than what's in the landscape today.

So it follows:

If you are betting on the outcome, then put your money on a smallish group, call it 10, of the ropeless competitors.

If you are competing, make sure you have a rope, because coming 20th or 30th is a much better personal outcome than possibly dying.

In a nutshell this is the conflict of interest between founders and investors, even though quite a few founders would rather have a 1:10 chance of a huge success than a 1:2 chance of a medium one.

In the scope of software engineering I disagree.

For software, the rope is the engineering. Whenever I need to move fast in a production environment I make damm sure my software is as good as I can make it because I know something will come and bite me in the ass otherwise. I know patience and carefull design makes me fast in the long run. I've done projects in short span and they were successfull precisely because of the quality I put in the code, and aggressive application of YAGNI.

Note: good here means understandable and simple, not some god awful super-elegant object abstraction scheme.

It might seem people with cowboy coding skills do well but it's only an illusion - they have faced as many bugs but they just code around them or do a quick fix.

The result from the latter is unmaintainable spaghetti.

Been there. It was a death spiral. The investors (which where customers, too) always wanted more features (of which not half of them made sense) so the directive given was just throw something at them that remotely resembles that requested feature to get the money. In the end the product was very unappealing and way too complex to sell to anyone else than this particular investor/customer. And as I heard recently they hired two more sales people in order to fix that. So what I'm trying to say is that not only you have to make sure your code base has a certain level of quality (fsloth's point) but this goes also for the product.

I totally agree that going slower is faster.

I disagree that the problem is cowboy coding. I'd identify the rope as carefully understanding the actual goals of the software and the way it functions in the world. Good engineering teams going "ropeless" rarely fuck up in a technical sense. I've been writing software long enough that my engineering judgement is usually pretty solid. Where they fuck up is building tons of extraneous functionality with the extra capacity, losing touch with what their software actually does in the world.

It very much depends on the situation. In an engineering environment (industrial control, regular applications, aerospace and so on) slower really is faster. But in a SAAS prototyping race with a whole bunch of like-minded characters those who cut corners may very well win the race. This is frustrating to those of us (including me!) who would rather do a proper job but the 'cowboys' seem to be both getting the funding and seem to be statistically doing better at success in the longer term. Of course plenty of the cowboys die with self inflicted bullet holes but the press does not seem to care and neither do investors. It's there on every level from 'go big or go home' to 'don't be scared to break stuff' to 'oh, we went bust, sorry customers, we'll try better next time'.

eh, the thing about margins of safety is that when done properly, they let you recover from mistakes, which lets you take risks more often. I see a break-even business as a 'save point' almost - you can go and put effort then into this or that or the other that might or might not work out, but as long as you've got the break even business paying everyone's rent, you fall back to a place where you still have all your infrastructure ready to go for your next try.

If your company is 'default dead' as it were, a schedule slip (a slip past the end of your runway, that is) means, at best, bankruptcy and starting over.

And really, you want some safety built into that too- if you screw up your taxes, or screw up your corporation badly enough, failure can mean personal bankruptcy, or if it is a tax matter, something much worse than personal bankruptcy.

Your premise is based on the assumption that those who win in the marketplace are those who move fastest, not those who have the better product, or make something people love.

I think that relatively few startups would get started if they all had to be 'default alive' - founding a startup would be limited to folks who were already independently wealthy. The point of the article, I think, is that if you're 'default dead', you should know that you're default dead, and your attention should be focused on getting to a state where you're 'default alive'.

Absolutely. My point in making the analogy is that 'default alive' is _very_ expensive, and a lot of work. It's probably not worth it in an early stage startup. The stakes just aren't that high.

I think it heavily depends on your product. If you have something that can be used by a single person (or small group) without a million others to use it, too, than you're IMHO default alive (Slack, Trello, ...). I think as a rule of thumb if you have to first get traction and then be able to monetize than you're default dead. Instagram's social component was default dead but the feature where you can alter your photos was default alive and that pulled them through.

On the other hand, Facebook and Google were both founded as "default alive" - the startup part didn't happen until the founders were certain they had something so good they should drop out of school to do it.

They also didn't have much in the way of worthy competition. That really helped them out a lot. It's different now. So much saturation in many areas things have moved from the era of bold ideas to the era of carving out niches.

Didn't google enter search market when it was considered full and pretty boring already?

From Bessemer's anti-portfolio: Cowan’s college friend rented her garage to Sergey and Larry for their first year. In 1999 and 2000 she tried to introduce Cowan to “these two really smart Stanford students writing a search engine”. Students? A new search engine? In the most important moment ever for Bessemer’s anti-portfolio, Cowan asked her, “How can I get out of this house without going anywhere near your garage?”

Facebook (launch Feb 2004) was founded after Myspace (launch Aug 2003), which was founded after Friendster (2002). So there was certainly competition.

Was it 'worthy competition'? Hard to say without hindsight bias (only the winners 10 years later were 'worthy'), but I doubt Mark Zuckerberg or the Google founders felt there was no worthy competition. It seems fairly clear to me that Facebook (and Google) succeeded because they made better decisions (on average) at a gazillion small decision points, not because they had some great idea that no-one else had thought of.

Oddly enough, Facebook was founded at almost exactly the same time as a fast-growing social networking site backed by a famous company with global reach. That was Google's Orkut.

I remember seeing a talk by Zuckerberg (or another early Facebook person) where they talked about clones of Facebook, and how they were racing to win over different colleges / universities before other social networks did.

I remember one particular college-targeted social media site (besides Facebook) at the time called CollegeClub. Met a couple women through that. It didn't have anywhere near the reach that Facebook did a couple years later, though.

Wrong on both counts. Both companies entered markets considered completely saturated.

Friendster and Myspace were the big players for Facebook, Lycos, AltaVista, and HotBot were competitors for Google. A new social network or search engine didn't get noticed back then.

Google was founded 'default dead' (they only changed that when they got into advertising), Youtube was default dead for many years even after being acquired.

This is absolutely wrong.

Facebook was founded in 2004 and didn't start showing ads until 2012[1]. Until then they were making losses (default dead).

Google is more complicated, but they were so desperate for money at one point they tried to sell to Lycos.

[1] http://www.thewire.com/technology/2012/12/2012-year-facebook...

> founding a startup would be limited to folks who were already independently wealthy

Isn't that already mostly true (with a few outlier counter-examples)?

What do you count as independently wealthy? Middle class (1-3 years savings, good education, professional job prospects) to successful startup founder is not uncommon enough to call an outlier - I personally know at least half a dozen people with that background and story. It's not exactly easy, but people can do it.

It is much harder for working-class folks (no savings, no college degree, no unique skills) to make the jump.

Pretty much by definition, yea. If you can actually afford to leave your job (in other words are independent of it) for 1-3 years and go off and start something that might not even make money, you're pretty wealthy.

In a much less serious vein, I had a boss who used to joke about throwing our laptops on the ground at work, saying "if you're not up and running on another machine in 10 minutes, you're doing it wrong"

I can do that at work, because all my code is on an EC2 instance, and I use emacs on the remote box as my editor. I've accidentally logged out of my X session and been back up and running in less than a minute before.

Then just replace "throw laptop on ground" with "delete EC2 instance".

That would take longer than 10 minutes but definitely under an hour.

How often is this put to the test? How do you automatically record configuration changes like the occasional "apt-get" to bring in something you're missing?

We use chef to record all our config. Our chef configuration is pretty stable these days, but it gets tested every time we build a new dev box, which happens at least as often as we hire new engineers. I could completely rebuild my dev box at any minute and the only thing I'd lose would be unpushed changes I was working on at that minute.

0 seconds. Just continue using laptop :)

What's the most common reason to fail a whistle test? If a real life "whistle test" situation were to occur and everyone let go without being 100% secure, what would happen? how many people would fall?

This idea is really interesting to me, but I'm not sure why.

The general idea is that whatever mechanism is allowing the ropes to move is something that 'fails closed'. The simplest mechanism is to wrap two prusiks[1] around the rope that is moving. Those prusiks are then securely attached to an anchor or some sort. Left alone, the prusiks will 'grab' the rope and prevent it from moving. Someone has to be actively holding them in position to allow the rope to move (otherwise they'll slide with the rope and pull taught).

In other cases, it's a question of how you set up things like descenders[2]. The same basic principle applies though. You need to be actively holding it open in order for the rope to move through it. There are a few exceptions to this (things like figure-8 descenders[3]), but in those cases, they are being used by a single operator to control their own descent.

The most common situations where this goes wrong are during transitions. The system used to lower someone over the edge is different from the system used to haul them back up. This exercise is fairly simple, but it requires a great deal of attention to detail, and the steps have to be done in the proper order to ensure there are no 'gaps'.

The number of people who would fall would depend completely on the operation. Generally speaking you have a single medic over the edge, who may have the victim clipped in to them if they're on the way back up. In some cases, you may have one or more additional rescuers.

[1] https://en.wikipedia.org/wiki/Prusik [2] http://www.petzl.com/en/Professional/Descenders/RIG [3] http://www.cmcrescue.com/equipment/rescue-8/

Thank you! This is reallyl cool. :)

The links are amazing. I just spent a bit of time looking through most items and figuring out how they are used. I guess I like this kind of stuff.

Where do you do rescue work? I do mountain rescue in southeast AK. I wouldn't work with a team that didn't use the whistle test.

Upstate New York

In the world of industrial rope-access, it'd be illegal in most jurisdictions to use any rope system that wasn't 100% safe to fall in to at any given time.

PSA: Rather than using underscores, _like this_, at HN you can get emphasized text using asterisks, like this. (I see _emphasis_ almost every day on HN, so IMHO it should support both. But until it does, asterisks are definitely the way to go.)

Everywhere else I write I use * except here because I want *, not italics.

\Escaping should work (but doesn't)\

Edit: That test failed, nothing to see here...

>so IMHO it should support both

some of us think it should support neither (because _emphasis_ is only slightly less legible than emphasis whereas erroneous use of asterisks is very much less legible than uses of asterisks would be if asterisks were not interpreted -- and because erroneous use of asterisks is fairly common).

But _if_ the present situation continues in which asterisks (or asterisks and underscores) are interpreted, at least the interface should provide a link next to the "ago", "parent", and "flag" links that show the raw, uninterpreted comment the way the user entered it.

Reply on news.ycombinator.com