walls.corpus

Immature optimization

Wednesday, 4 July, 2012 — software business

My wife and decided on dinner out Tuesday night. We picked Olive Garden.

It’s an easy option to nitpick. It’s a chain. It isn’t authentic Italian. The portions are ginormous. The pasta’s overcooked. All true. We went anyway because we typically enjoy the heck out of the salad and salty breadsticks.

Now, we’re both aware of the chain’s ordinary course of business, but in many restaurants, minor customizations to the menu aren’t problematic. So, we ordered two entrees. Capellini pomodoro for me, eggplant parmesan for Robin. Knowing what we would otherwise get, we specified al dente. Our server confirmed that as she put the orders into the system.

When the plates arrived roughly 15 minutes after our order was put in, it was evident by sight that the pasta was overcooked, entirely contrary to what we and our server understood we wanted.

How did this happen?

When we’ve gone to Olive Garden before and I’ve made a point of asking for al dente, the response I got from the server was effectively, “What’s that?” So, when my order arrived overcooked, I assumed the server didn’t relay my request to the kitchen. This time was entirely different. Our server made the specific point of confirming detail — acceptance criteria, if you will — as she put the order in.

My wife and I began speculation about why. We didn’t get very far when our server returned a couple of minutes later and asked if the pasta was correct. We both said no. That’s when we learned from our server that Olive Garden precooks pasta in the morning as part of the preparation process. When a dish is ordered, the pasta is warmed up for a minute or two in hot water.

We stopped. We both understand that running a restaurant means preparing the kitchen before opening. My expectation of early preparation is getting sauces made, vegetables cleaned and broken down, cooking stations prepped for use. I could understand making the lasagna early, then finishing in the oven when ordered. But precooking all of the pasta? Unbelievable. We were told it was in the interest of time. The kitchen is incredibly busy.

Here’s what I understand about pasta. It’s best fresh, but good pasta from dried is very doable. It’s best served with some resistance in the noodle, that’s the “to the tooth” of al dente. You can keep a boiling pot of water going in the kitchen and drop pasta a few minutes before the order needs to go into the window for service. The water’s boiling, seasoned and ready and you can keep using that boiling pot of water all night. The actual noodle cooking doesn’t take that long. What’s key is knowing when to pull it.

John Siracusa, best known for exceptionally comprehensive reviews of Mac OS X releases on Ars Technica, wrote the following about cooking pasta:

As you gain experience, you’ll be able to tell when pasta is ready by “feel” (with a pair of tongs or a stirring spoon). But the old fashioned way is still the most reliable: taste a piece. Drop the pasta in the boiling water (see the next section for more on that), set a timer for 1-2 minutes less than the time on the box of your trusted dried pasta brand, and start tasting when it goes off.

There’s an old saying about cooking eggs: done in the pan, overdone on the plate. The same goes for pasta. It will continue to cook after you remove it from the pot, and even more so when you put it directly into another hot pan or combine it with other hot, moist ingredients.

Even setting aside the fact that, in the usual course of business, Olive Garden intentionally overcooks pasta, cooking a batch to customer taste shouldn’t wreck the kitchen’s timing. Might it take longer to order a fresh batch of pasta? With thinner pastas like the capellini I ordered, they’re probably doubling the time they need to cook the dish by handling it twice. Two minutes in water might be over doing it. Capellini pomodoro is dead easy to make.

Strictly for sake of argument in this paragraph alone, let’s accept precooking pasta as beneficial to the business overall. Is it then unreasonable to expect the kitchen to set-up a pot to boil water to accomodate variance? No. We’ve already established they already have water boiling to reheat previously cooked pasta.

Restaurants can’t offer every variation of ingredients and cuisine. You’re unlikely to get grilled cheese at a teriyaki house. Software also can’t be all things to all people. At the same time, a smart business understands and embraces adaptation to meet customer needs. Restaurants handle this frequently to work with dietary restriction, food allergies or, as offered at Olive Garden, cooking a steak to order off of a standard of medium.

The desire to optimize is strong. Businesses have it with operating process. Chipotle precooks some of their meat in a central commissary, then ships it. Chain fast casual restaurants have plate plans, showing how every dish should be assembled at every restaurant, every time. Similarly software engineering focuses on optimizing process and development through use of automation, frameworks, higher-level languages and so on.

Just as there are known optimizations and patterns, there are also antipatterns, such as premature optimization. Premature optimization is a potential fit for my Olive Garden experience. But, what slays me isn’t precooking the pasta. It’s the fact that it somehow precluded me from getting what I asked for, even after it was implied I could have it. Smart, mature businesses don’t screw that up.

Don’t “save time” to deliver the exact wrong thing.

Video stores competing with Netflix and Redbox

Saturday, 30 July, 2011 — business community

The New York Times’ Nicole LaPorte has a neat pieces about independent video stores retooling their business in the sunset of Blockbuster and the ascent of Netflix, Redbox and other online options for movie watching.

They’ve done what a lot of declining industries like newspapers or record stores failed to do for a long time — recognize that just being what they were was not going to keep them in business. Being a commodity means when someone improves the commodity business, if you can’t match, you’re screwed.

I think this is a great deal to do with why Blockbuster, Borders, many of the chain record stores and so on are gone or fading quickly. They simply aren’t equipped to reflect their communities and be something other than a brick-and-mortar when Amazon Prime couldn’t get it to you fast enough.

The bookstores that are going to thrive in the time of Amazon, the record and video stores that are going thrive in the time of iTunes are going to do things like this, regularly and with excellence:

A campy sing-along night is just one component of their plan. Since Vidiots, a beloved institution among the area’s movie cognoscenti and stars, opened a sleek space called the Annex a year ago, it has offered a “Film Studies” program. It has had classes on anime mythology; lectures by filmmakers like Larry Clark (“Kids”); and spoken-word events, known as Tail Spin, where participants deliver improvised monologues on a theme (for example, “the stranger”) for five minutes before the thread is picked up by someone else.

Physically, too, the Annex symbolizes a new era. Its clean, modern design bears no resemblance to the graffiti-covered walls of the video store, which feels more like a basement clubhouse.

The special events have been integral to Vidiots’ transformation from a strictly retail business to a cultural hub and community center. They are intended as a riposte to what the store’s fans regard as the nameless, faceless quality of services like Netflix.

Laziness is a (business) virtue. Sometimes.

Thursday, 7 January, 2010 — software business

Rafe Colburn on what my previous job called “a buy vs. build analysis”:

The question people need to be asking is how little custom software can they get away with having. The ideal number is zero. If you’re working at one of those web design [firms] that rolls a new content management system for every customer, you’re doing your customers a disservice. Honestly, if you’re selling them your own proprietary CMS you’re probably doing them a disservice.

Software developers like to build things. And most developers are confident that they can provide something that perfectly solves whatever problem they’re confronted with as long as they can write it from scratch. Developers are horrible at estimating the long term costs of building applications yourself. And they have an incentive to be bad at it, because if they were good at it, nobody would let them loose to write custom software.

When you do bend the business to fit the software, you can end up writing a lot of glue code to get one API to talk to another API; to push or pull data whichever way it needs to go. That’s probably OK, when you’re dealing with a competent developer community or vendor.

The counterpoint is lack of confidence in the “off-the-shelf” solution. At my last gig, my coworkers and I preferred in-house development because we had just about zero confidence in the abilities of our vendors to provide a stable API, accurate documentation and reasonable integration cost. This collective opinion came from years of bad experience. In our minds, it was lazier to build a solution ourselves than continually fix what we bought from someone else. We were far more trusting of RubyForge and CPAN (and building on top of that) than anything that involved a contract.

Also, pray the company answer to the central question, “Do you want to be in the software business?” is either “yes” or “no.” “Maybe?” is a really scary spot to be.

I hearticon new products

Saturday, 7 March, 2009 — business

CrazyLikeThat.com has two brand new products; a heart emoticon t-shirt and paired greeting card. The t-shirts feature a red <3 on the left side, printed on a black, 100-percent cotton, Gildan T. The card is thick stock with just a little bit of gloss. How much? T-shirts are $16 in sizes Small through XX-Large. Our first greeting card is $3.

It’s fun codeveloping products with 5x5. One of us mentions an idea and the other has a slight modification that makes it better. We can usually refine an idea in about five minutes. A few weeks later, we can offer them for sale. That’s pretty sweet in my book.

Neighborhoods don't scale

Sunday, 1 March, 2009 — business journalism social+media

There is so much cross-pollination between starting a neighborhood site and running a small business. Jeff Jarvis posted about CUNY and the New York Times beginning a partnership for a network of hyperlocal sites. Howard Owens, formerly of Gatehouse Media, is taking over The Batavian as owner and got into an interesting argument with Jarvis. He expanded on his point of view in a full post, “VCs chasing fool’s gold in funding ‘hyperlocal’ projects that 'scale’”:

Now scale is being applied to “hyperlocal” start ups.  And the meaning in this context, as I take it, is that a “hyperlocal” business needs to have the capability to expand in multiple towns and neighborhoods rapidly at a very low cost.

…

The “hyperlocal” approaches that supposedly “scale” don’t scale in one very important aspect: building new audience for community news.

Sure, they might appeal to a segment of the population that is already involved in a community, but they’re not tackling the “Bowling Alone” problem.

From the number and visibility of venture-backed, industry-supported hyperlocal flameouts, it’d be tempting to think that there isn’t money/eyeballs in being hyperlocal. Heck, the Washington Post failed to make a go of it. But it seems like all of the existing big media companies and venture capital-backed startups are trying to attack the problem the same way. Throw money at building a platform, put it in a lot of places. When it doesn’t work, it’s a failure. But that doesn’t mean there isn’t a solution. Just that the way a lot of people have been trying to go about solving it isn’t working.

One of the things Jason Fried and David Heinemeier Hansson of 37signals talk about frequently is doing only those things you absolutely have to do. Trying to instantly scale the platform is premature and distracts from learning what the business needs to work and executing on it.

I was ecstatic to read David Westphal’s piece at OJR.org, talking to various local sites and how well they’re doing:

Local news sites come in all sizes and shapes. Some are non-profits. Some aren’t trying to live off the operation. But for those who are, some survivable wages are being earned.

Tracy Record and Patrick Sand, another husband/wife team who operate West Seattle Blog, are getting revenue in the high five figures. Debbie Galant, co-owner of Baristanet, earned more from the site than she did from her free-lance writing business last year. And Bob Gough, who runs Quincy News, pockets $1,000 a week in wages from his startup that serves an Illinois community of only 40,000.

Right now, there are a lot of companies predicting doom and gloom and continuing to do exactly what they’ve done. There are also entrepreneurs who “don’t know” that they’re not supposed to succeed at providing a service and earning money where others have been before. But why are these sites succeeding?

Success for these sites looks different than it does for an established media company. They’re focused on solving a smaller problem and making it the problem they’re addressing. They’re risking the business on solving it. They’re climbing Everest by climbing Everest, not thinking about how they’re going to summit then climb K2 or the Matterhorn or Denali in the exact same way with the exact same tools and the exact same resources. They’re all mountains in the way that Raleigh, Miami, Paris, Mumbai and Houston are all cities.

Within those cities, distinct areas and boundaries. I’d pursue a site covering SoMA in San Francisco differently than I would one covering North Beach or the Sunset. In Paris, the Marais is different than the 5éme arrondissement. In Raleigh, Glenwood South isn’t North Hills or Oakwood. Sure, there are similarities, but you can’t treat them as the same thing. Neighborhoods don’t scale.

I suspect beyond the stories these sites are writing, they’re solving the advertising needs of smaller businesses, the sorts of businesses that live in smaller spaces along city blocks and next to strip mall anchor stores. These are, coincidentally, the sorts of businesses that don’t tend to buy any or much advertising at the major daily. Why not? Well, drawing on my own case, buying a print or section-specific ad at a typical newspaper would be beyond budget. I’m not going to reach the right people for my business, either. A neighborhood site is a much easier way to draw and address a specific audience. Plus, it’s far more likely I can see and talk to who’s selling me the advertising. With not a whole lot of luck, they’re probably a customer, too.

Maybe there is a way to abstract a platform and aggregate neighborhood sites, but, just as mountains have their own weather, neighborhoods are unique and not taking the time to dive into them and understand them is a mistake. The large, monolithic approach is not the workable one. There’s no rule saying there must be a way to build and sustain a larger business out of “hyperlocal” content.