Erik Wiener

How about more kids in the workplace?

I work out of a co-working space a few days a week. There are a couple of high school tutors who operate out of there, so there are often school-age kids in the office from around 3:30 - 5:30.

When I walk by them, it usually makes me think about my own kids and miss them in a more visceral way than I do when thinking about them in the abstract. As someone who can get caught up in my work and lose track of time, this sometimes gives me the jolt I need to get out of the office and home to my family at a more reasonable time than I otherwise might. It sort of jolts me out of work mode and into life mode.

I wonder if there would be benefits (in both directions) of getting more kids into workplaces in general – for tutoring, after school classes, whatever. What if we got more businesses to open their doors to this sort of thing? A sort of everyday “bring your kid to work day”. The kids could get exposed to the realities of the modern workplace, and the adults, especially the parents, would be reminded that… kids exist.

In a previous post, I wondered why major ad-supported companies like Facebook and Twitter don’t let me pay to avoid ads.  

Amidst all the controversy around content blockers in iOS 9, it occurs to me that maybe the path to letting me pay is not necessarily that difficult. 

What if I could sign up with one of the major ad networks to ensure that whenever the network saw my cookie on any site, it would serve up… no ads. And charge me as if I were the advertiser for those “no ad” impressions. 

For my part, I would see no ads and would be billed monthly by the ad network based on the number of “no ad” impressions it delivered to me, at whatever effective CPM rate was needed to ensure my “no ads” always won the auction (or indeed whatever CPM I was willing to pay). Like any other advertiser, I would have access to a report showing the sites and impression counts for my “no ads” for full billing transparency. 

For the ad network’s part, they would treat me internally like any other advertiser, no significant change to their systems. I would be a retargeting audience of 1 for my own “no ads”. When my “no ad” was selected for delivery, they would deliver… nothing. They would bill me like any other advertiser (ok maybe with a consumer friendly version of their dashboard). 

For the publisher’s part, they would never know or care that I was a “no ad”-vertiser. Included in their revenue from the ad network would be my little payment. They might get reports from the ad network on how many unique “no ad”-vertisers they had and how many impressions. 

In terms of how much this would cost me per month as a user, it would be on the order of however much advertisers are currently paying per month for my attention. It’s been a while since I was up on ad rates, but let’s say for argument’s sake it’s $2 CPM on average. $10/month, which is about what I spend for Rdio and Netflix, would buy me 5000 impressions/month, or let’s say 100 page views/day. Seems like the right order of magnitude. 

Now, a big part of what content blockers do is block ad trackers. So to go along with this “no ad” approach, I would want my content blocker to block all trackers except for the one used by the ad network I’m participating in. So I would never see ads. But to the extent that the sites I visit use my ad tracker, they would get my money. 

Everyone wins. No web rearchitecture. No new intermediaries.

I’m not an adtech guy, so maybe this is hopelessly naive. But it sounds simple.

Who is already doing this?

Google, Twitter, Facebook: let me pay you a monthly fee for an ad-free version like I do with Rdio.

Why not?

I can think of a few reasons you might not want to let me pay:

1. Your ad model allows you to continuously optimize revenue from existing users and dial up the algorithm to hit your quarter. With a subscription model, you can only grow by adding users or raising prices, which limits your revenue management flexibility. It is harder to discover the optimal subscription price than it is to discover optimal ad placement, format and pricing because the latter has a faster feedback cycle.

2. Adding a pay tier makes it harder to convince yourself (and your users) that the ads improve the user experience. If you justify sponsored content by telling us how useful highly targeted sponsored content can be, offering to let us pay to get rid of ads confuses your story.

3. When users pay, they might expect a higher quality of service. Keeping it free probably reduces your customer support responsibilities.

4. Your accounting and financial systems would be more complicated with a pay tier and maybe you’ve decided this outweighs any benefits.

All potentially valid reasons, and I’m sure there are more.

But still, come on, make me happy, let me pay.

That sound from Intellivision D&D when the pieces of the crown come together at the end — that came into my mind yesterday in a flow state of programming, the moment when you have every piece of the engine arrayed out on the floor after a whirlwind refactoring — when everything couldn’t be more broken but you have it all mapped out in your mind — and then in a rush it all snaps into place and… it works. And it’s not just that the code works. It’s that where there was chaos, before the refactoring, there is now order and beauty. There is peace in your little kingdom.

What would you do differently — what would society do differently — if we knew with certainty that the world would end in 10000 years? 1000? 100? 10?

Similarly, what if you knew with 50% certainty that the world would end in 50 years? 10% certainty? 1% certainty?

I like to do these thought experiments, picking various time scales and/or probabilities along the spectrum, until I find the points at which “obviously everything would change” becomes “some things might change” becomes “nothing would change”.

A fun way to unwind the time preferences and sources of meaning in life.

Somewhat different question, but add it to the mix: what if it was a potentially avoidable catastrophe requiring a large effort and high cost to combat. For example, on course to collide with a giant asteroid with 100% certainty in 1000 years. 100. 10. At what point does humanity start preparing a defense?

Every time I read something about a big co acquiring a startup and immediately shutting down its service (so the founders can focus on making even better things for you!) or a company sunsetting a new product after two years of hyperbolic marketing, I get embarrassed on behalf of this industry for how fragile it is and how poorly it treats its users.

I get it, the software biz, especially the consumer software biz, is one long experiment, but pity the foolish user who invests their time or data or energy into any of it. The more so for free services that seek to host or mediate some important portion of users’ online lives. You’ll be lucky if you get two years before the founders fail and shut down or succeed and shut down. Now you can forget about that data or try to migrate it.

However, even very successful services, or at least long lasting ones (Flickr, etc) will not last for that long in the scheme of things. Likely not half as long as a paper photo album. The old services will inevitably start falling behind newer entrants in fashion and capability and there’s only so many times a service can reboot or get bought and revitalized before the last manager left holding the bag makes the calculation to shut down.

So whether it’s 2 years, 5 years, 10 years or 30 years, your cloud data will die.

That presents a business opportunity for data continuity services. Businesses that handle the task of porting your data from one service to the next. No promises of compatibility with old API’s and tools, but at least the assurance of data integrity and continuing availability in the latest best of breed service.

Basically, insurance against stranded assets. Which might make users feel freer to invest data into unproven services.

This would be a challenging business in itself to run because of the very nature of its long term value proposition. It’s a service you might not need for 5 or 10 years. But, in that sense, not necessarily different from an online backup service or an insurance company.

How would it work? Would you proactively mirror users’ cloud service data into your own “continuity cloud”? Like backing up everybody’s Dropbox and Instagram and Facebook and gmail data into your own storage cloud so you’re ready in advance to migrate users? Or would you migrate on demand? With the former, it’s easier to give users proof that your service is doing something. But it also may create the impression that you’re now just yet one more cloud service holding users’ data that’s just as likely to fail as any of the services you’re purporting to protect.

I would think on-demand migration might have better optics. In the sense that one of your selling points is that users don’t actually have to trust you to safeguard anything over time and hence the cost of you failing is low. Low switching costs. Companies usually don’t like to promote low switching costs, that’s the opposite of what they want. But this shouldn’t be a usual company. Perversely, part of your value proposition should be that you have a meta process for providing continuity of your own continuity service in case you fail. And perhaps this means that there must exist multiple competing continuity services who also cooperate to insure each other, in the same way that banks privately fund the FDIC insurance pool through premiums. (A sort of federal data insurance corporation.) The continuity of the continuity service would take the form of a surviving continuity provider seamlessly taking over the accounts of the failed provider.

How would a continuity service make money? You could charge users directly with a yearly subscription, or premium. Or you could charge the online services as sort of a data bonding service, and the online service could promote the fact that they are bonded for data continuity in the same way some ecommerce companies promote the fact that their security has been audited to create trust. Perhaps bonding for data continuity would become table stakes for getting VC funding. Perhaps some states or countries might require it by law.

A follow up to my earlier Cake Walk post.

My idea after last year’s failure:

What if I have them start in an inner circle and then lead the line into an outer circle that has twice the number of slots for the actual cake walk?

Well, we tried exactly that at this year’s Mayfair last Saturday. And… success!

However, we didn’t quite have twice the number of slots in the outer circle. More like only about 20% more. When the inner circle wasn’t quite full in the first couple rounds of the day there was no traffic at all. When it was full, we still had some traffic, though not as bad as previous years, and of course we always had at least one full revolution without any traffic as the inner circle filed into the outer.

So for next year the final refinement: put 45 slots in the inner starting circle and put 70 in the outer, leaving 25 gaps.

I have a cake walk problem. As problems go, it’s not a big problem. But it has been bothering me for the last several years.

Background

At the annual school Mayfair, we always have a cakewalk. Throughout the day we run about a dozen of them. Each winner takes home one of the cakes baked for the cake decorating contest.

About 30 people join in on the walks, each starting on a numbered slot drawn in chalk around a big circle. When the music stops, a random number is drawn from a can and the person standing on that slot wins.

The Problem

When the music starts, no one moves. With some cajoling, the circle eventually moves in fits and starts, but it’s a big traffic jam. Not much fun.

For a couple years I observed this and thought there must be a better way. Last year I volunteered to run the cake walk in order to solve it. I failed.

Hypothesis

My theory was that there were two problems: people were just too close together and also afraid to step off of their slot (erroneously) thinking this was musical chairs and they would lose if the music stopped while they were between slots.

The Experiment

First, to address the musical chairs anxiety, eliminate the space between the slots. Rather than drawing individual circles for the slots as in previous years, I drew a ring with 30-something contiguous segments. No way to land outside of a slot.

Second, to address the traffic crowding, I took a cue from my father’s 1965 PhD statistics thesis on on-ramp metering lights and I tried to introduce gaps in the line. My plan was to prevent people from starting on four of the slots spaced evenly around the circle. Crucially, those spots still need to be numbered so people knew they could land on them once the music started. This meant there would always be four extra numbered slots, which I thought would be enough to create a sufficient gap to prevent a jam.

Result

Traffic jam. Not much better than the previous year

First, with the contiguous slots, the reluctance to step off your slot before the person in front of you stepped off theirs was replaced with a reluctance to step into their slot before they stepped out.

Second, the gap strategy ended up being hard to implement and even when it was implemented, didn’t help enough.

First I tried just marking the gap slots with an arrow saying “Don’t start here”. People either didn’t see that, didn’t understand it or didn’t care. There’s a lot of commotion, a lot of young kids and there’s a mad rush to grab a free slot before each cake walk starts. Too much work to explain to people why they couldn’t take those gap slots at the beginning and once a kid was standing there next to their friend, you couldn’t really tell them to move.

So then I tried having slot-fillers take the gap slots and step out just before the music started. Worked a little better, but still a pain to recruit the slot fillers and they would sometimes still give up their slot in the end to a kid who really wanted to get in. Plus it was a pain having to explain all of this every time to the slot fillers.

Next Steps

While it may seem that just spacing the slots out more or having fewer people in the line might work, I think the musical chairs anxiety would still keep people in gridlock.

I suspect the gap approach will work, I just need more gap slots and a workable approach to keep them free in the beginning.

Whatever the solution, it has to be simple and foolproof. You don’t want to have to explain or enforce complicated rules.

Thinking along the lines of metering lights…

What if I have them start in an inner circle and then lead the line into an outer circle that has twice the number of slots for the actual cake walk? Or have the outer circle loop back to the inner circle?

Time for another experiment.