JasonCEC

I actually wrote a complaint to the Startup School contract after a being disappointed by the event in NYC last year...

The event was ok... but really lacked anything you couldn't find elsewhere. I was particularly annoyed that they didn't even try to sort the companies or founders or attendees into discussion groups, make connections, or even foster conversation after the event. In fact, it was almost like they were working against that, by whisking away all of the speakers and interviewers imediatly following the event and during the one intermission....

------------------------------------------------- [1] My complaint letter:

Hi Kathrina,

I really enjoyed ~87.3% of startup school, but wanted to pass along a few thoughts on how future events could be better.

1) Introduce the interviewer - not just the interviewee. I had no idea who Aaron Harris was until 2/3'ds of the way through the interview.

2) Don't choose startups at random. Everyone had to apply to be at Startup School - you should use those applications to chose which startups get to present at office hours.

2A) Don't chose 3 startups at the same level of growth. I run a b2b saas startup in its early growth stage. the three startups chosen for office hours we're all in the early development stage.

2B) Don't choose 3 b2c startups. I'm sure I wasn't the only enterprise b2b company in the audience.

Office hours was the section that fell flat with me. None of the companies interviewed we're in my companies stage of development (all earlier), and none we're business to business.

Furthermore;

3) I guess its OK to leave the attendees to fend for themselves... but it could work a lot better if you used the intermission to have groups cycle through quick talks with the speakers, or with YC partners. That would also add more value to attending the conference.

3A) The speakers should be more public during the break, and during the social following the event.

I'm Looking forward to future startup schools, and I hope this advice is helpful!

All the Best, Jason

It's interesting to me that Open Revolution R supports R Shiny (developed by RStudio). A natively multi-core version of R could be really useful in dealing with the larger applications that are built on shiny, as computations, input, graphing, and DB i/o are all (often) thread blocking.

My entire company is obsessed with keyboards. [I type on my Das cherry browns - tanked but not clicky]

We have, collectively, - Original 1983 Model M - Unicomp Model M - Das ultimate 3 cherry red - Das pro 4 cherry blue - Das pro 3 cherry brown

We're thinking of trying out the Keyboard.io butterfly split board....

Mechanical keyboards make a huge difference in typing speed, accuracy, and comfort - its worth the extra cost!

There's few feelings as good (that can be had sitting / standing [yes - I'm at a standing desk. Coder, hipster, ceo, startup data miner] at your computer) as your fingers floating over the keys without bottoming out.

I can't even describe how much I wish `stringsAsFactors=F` was the default.

Mike, I build flavor profiling and quality control tools for coffee, beer, and spirit producers. I would love to help with the coffee DNA project. Shoot me an email[1]?

[1] JasonCEC@Gastrograph.com

Interesting;

not to lessen your ideal, but both sour and bitter fall into a category of complex flavors.

I absolutely agree that good coffee is a balance of sour (acidity) and bitter, but I take acceptation to the idea that it's a balance betwen these 2 flavors;

a coffee can be both very sour and very bitter!

Consider - nearly all of the complexity in coffee is from the development of acids as the coffee is growing, and the conversion of amino acids and proteins during the roasting process (Maillard reaction) of the bean;

if you over roast the bean the coffee will breakdown too far and develop carbonates, which are both cancerous and unpalatable. That's one type of bitter.

Coffee that's roasted a little less (but still far too much) will contain a campfire note (smokiness).

In both of these instances, good acids (which cause sourness) such as malic, tannic, and citric, are broken down - while less palatable acidic compounds such as acrylamide are produced.

You don't really... Espresso makes highly concentrated brewed coffee that people consider to be closer to espresso.

The 'debate' is that any coffee pulled under pressure is technically espresso - but (real, read: not home machines) pull espresso shots at ~9 bar. Aeropress generates about 1.5 - 2 bars of pressure, and thus is... something between coffee and espresso. Most people have taken to calling it "Espresso style coffee" which is like calling Tofurky meat style tofu. WAT?

Anyway, most people (but not me) dilute their Aeropress coffee in water to make a Americano-style drink, so its not really comparable at this point to a true pour-over anyway.

All that said, I think aeropress makes quite good coffee - but its not a replacement for a pourover, and there's a lot more batch variation due to the lack of control in an aeropress.

Hope this helps!

Generally a good article;

I run a startup[1] for quality control in the craft beverage industry (coffee, beer, distilling), and we have over 7,000 full sensory reviews of the flavor profile from various brewing methods.

I would highly suggest that anyone interested in brewing better coffee at home (or in the office, etc), invest in the following:

1) freshly roasted coffee - flavor profile degradation begins around day ~10 in most 3rd wave light-medium roast coffees. I would suggest Intelligentsia coffee, Blue Bottle, Stumptown, or Counter Culture. These guys are the "big 4" and trying either single origin or blends from them will give you a better idea of what to look for in high quality beans.

2) a conical burr grinder. Do not use a spice grinder (whirly blade) which will shatter the bean causing large amounts of micro-fines and make your coffee bitter. Capresso makes an OK home grinder, or (if you have the patients for it) the Hario hand grinder (with a ceramic burr) is excellent (but takes a while.... though it can be used in situations lacking electricity). In our dataset, conical grinders increase the Perceived Quality of "3rd wave" beans by 1.4 standard deviations on average - there is almost no effect on burnt beans or lesser coffee.

3) a Chemex or French Press. These are both simple, elegant, and allow you to make a range of quantities depending on how much coffee you have and who's drinking. Both of these brew methods score .75 standard deviations higher in perceived quality than average with less experienced (in our dataset that means 'non-expert') tasters. The average perceived quality can be further improved by correct dosing and pouring methods.

4) a swan neck kettle. This is 80% of what you need to to perfect your pouring method. The rest is practice.

Finally, one of the best things that you can do is record and rank the coffees you consume. This will help you objectively remember what you liked and disliked about the product, and gain intuition on your preferences. My company's tools are built for professional tasters in quality control, but anyone serious about coffee could find suitable use for it. Gastrograph Review is available free on the Google play store here[2].

[1] www.Gastrograph.com [2] https://play.google.com/store/apps/details?id=com.gastrograp...

Point 1 is not false;

I work in Quality Control for artisan distillers - heads and tails recycling is an important part of the production process, and requires both skill and testing to make sure methanol is removed from the spirit.

A bootlegger would need 1) a fractional still or 2) precise temperature control and monitoring - they were unlikely to have either.

Furthermore, I find it doubtful that an illegal bootlegger would have used pure sugar for their mash - that would have been far more expensive than the purchase of grain.

Finally, on point 1, the foresight and knowledge to remove methanol doesn't come in a nice pamphlet with the illegal still - we know that a lot of the alcohol produced by the illegal bootleggers (who the government did not have control over) was contaminated with methanol (And other heavy metals or contaminants).

On point 3; Sure... that's almost right...

Industrial alcohol not made for human consumption would already contain methanol and other contaminants; the law establishing a minimum level of contaminants to render it undrinkable by humans did theoretically make it more harmful.

That does not constitute a "Federal Poisoning Program". - when someone hears those words, they don't think "the government established that already undrinkable industrial alcohol must contain a minimum level of naturally occurring contaminants to stop individuals from consuming it";

they think "the government purposely added poison to alcohol sold to individuals for consumption" - which is not what happened.

Right!

Though: Heads and tails recycling of the mash is more complicated than simply 'discarding' the heads to avoid methanol.

Bootleggers usually weren't using a fractional still, and thus would have needed precise temperature and timing to even attempt to measure methanol concentrations.

Furthermore their supply of grain, being an illegal operation, was (probably) sub-par; any rot or mold on grain to be distilled greatly increases the chances of producing methanol.

Again, I think only ~10% of the alcohol by bootleggers was contaminated. Distilling has been around for a long time, but the effects of driving it underground greatly reduced quality and increased the chances of going blind or dying due methanol poisoning.

I debated this ~4 years ago on the Penn State University Parliamentary Debate team. During the practice session I tried to use this as an argument against the claimed moral high ground of the prohibitionists.

It failed; our coach and the rest of the team we're able to prove to me, with a more careful reading of the article (and checks from other primary source material) that there was never a "Federal Poisoning Program" and that most of the deaths resulted from a lack of communication with those looking to drink while alcohol was illegal.

On both of those points:

- 1) Only industrial alcohol was legal to produce during prohibition. Industrial alcohol contains methanol, a compound that can cause blindness and death. It takes a good amount of skill and refinement to remove all traces of methanol from alcohol to make it ready for human consumption - a step that the legal distillers had no reason at the time to take (and would have in fact had them closed down for bootlegging)

- 2) The Federal Government never added poison (methanol) to alcohol made by bootleggers - had they found the bootleggers, they would have destroyed their stills and jailed them, not poisoned their stock, brought it to market, and sold it. I believe[citation needed] only about 10% of the alcohol produced by the bootleggers was contaminated in any way, and even less caused 'permanent damage' beyond what you would expect from people drinking alcohol.

- 3) The order to add methanol to industrial alcohol is historically dubious... All of the citations I can find for it in the searchable web (read: Google) links back to the original Slate article. Supposedly, the article is based on this Article in Nature[1]. From the abstract, it is more likely that the mass of laws and policies that followed the ratification of the 18 amendment included an understanding that industrial alcohol should be exempt but rendered purposely undrinkable - which it already was.

I am personally against prohibition, and against the government attempting to mandate morality; but I fear that this article is popular because of its anti-government attitudes - but you could only come to the conclusion that there was ever a "Federal Poisoning Program" if you give perfect foresight agency to every policy the government passes....

[1] http://www.nature.com/nature/journal/v463/n7279/full/463299a...

This is (wonderfully) reminiscent of Harry Potter and the Methods of Rationality[1].

Both are well worth the read.

[1]http://hpmor.com/

Consolidation is a ways off. The large beer producers have acquired a few craft producers in the last few years, but the artisan beverage market will continue to grow in both number of firms and amount of production.

I think it will be at least another 10 years until we start seeing large craft brewers purchasing smaller craft brewers. This has already happened in coffee with Blue Bottle (venture backed) purchasing Tonx Coffee and Handsome Roasters. Artisan Coffee has had ~10 years of growth on Craft Beer.

Agreed.

I'm not sure how much I agree; wine and coffee work quite differently than distilled spirits.

In wine, you have never needed to own the vineyards to be a vintner - as far back as medieval Europe, we had merchants purchasing grapes from various regions and blending their own house wines. There is still skill in matching various grapes, yeast, water, and oak - much more so than simply bottling someone else's spirit.

It is similar in coffee; individual farmers tend to the crops and produce the green coffee (sometimes a commodity, sometimes a specialty product). Roasters purchase the green coffee from a middleman who takes care of sourcing, shipping, and storage. Often this is done after a sample has been sent and roasted and the price has been negotiated. The roaster is then responsible for the majority of the final flavor profile of the bean. His (or her) level of roast, airflow, timing, etc, will determine how the coffee should be brewed, and what notes will be present and what will be absent.

So again, it's a little different in coffee than spirits... the grapes and green coffee beans are closer to the raw grain used in making whiskey and rye, than to the finished spirit. There is no deception in selling wine made by another vineyards grapes or coffee purchased through a middleman.

As a side note, all of the largest artisan coffee producers (ie Stumptown and Blue Bottle) have started Direct Trade relationships with coffee farms, and pay a price premium to the farmer to increase the quality on the farm side - investment like this has yet to happen in the distilling industry.

Yes, that is what I meant :)

If we take this a little further though, part of craft production is the curation; being willing to dump batches that don't meet your quality standard.

Any producer who knowingly ships product below their quality standard is no longer engaded an artisan production.

Unless this is the 'No true Scotsman' fallacy...

Some of that is true; but there are historical antecedents that explain the use of hops and malts...

My company[1] builds quality control tools for artisan beverage producers; we know that craft production is a batch process relying on agricultural grains with inherent variation.

That doesn't mean it shouldn't be produced.

Craft production is undoubtedly higher quality than any continuous production - it just takes more skill (or software tools..) to do correctly and consistently.

Furthermore, the use of certain ingredients to reduce noticeable inconsistencies came at a time before process control was a known science. For example, hops were originally added to beer being shipped from England to India (hence the name India Pale Ale or IPA). Now, that practice didn't survive because of quality control - it has survived because people like the taste of IPAs and there is demand in the market for that flavor profile.

Consumers don't think of skill of production when rating beers or whiskeys; they judge only the resulting flavor profile and are influenced by contextual clues such as age statements or 'hand made batch numbered artisan spirit' promises.

Side-by-side comparisons are often misleading in hedonic sensory evaluation. Even trained individuals have a tendency to select the sweeter and stronger flavored (intense) product.

Bitter and woody notes are important balances in higher quality bourbons, but are often perceived as negatives when doing comparative evaluation.

This is likely from your description, as the higher end bourbons tend to be wheated and lighter, but with a more prominent woody body from their longer barrel ageing.

Production practice of Vodka matters less, because the point of vodka is "a flavorless spirit". If you're dealing with flavor vodka, your no longer dealing with a craft/artisan spirit anyway.

My company[1] works in this space, building tools for artisan brewers, distillers, and coffee roasters. I know this market quite well!

This is a known practice, and one that a lot of people are unhappy about. We suspect that the TTB will be cracking down quite harshly on this in the near future; forcing a named distiller statement on the label.

That said... not all of the whiskey, rye, or bourbon that MGP produces is bad - some of its aged production is quite good, and companies like Angels Envy which further age it (basically double oak or quarter casking) can produce a unique and high quality product.

My problem with this is from the deception; imagine if it turned out that Dogfish Head and Stone brewing were actually buying their beer from MillerCoors! This is an unfair practice that puts actual craft distillers (some of my clients!) at a huge disadvantage. A named distillery requirement would go a long way to correcting this.

[1] www.Gastrograph.com

Shiny has a full time team working startup hours on it - it's out of beta and has enterprise support (Shiny server pro). It's come a long way to put it mildly; and we've never had a problem using it in production.

We could interface R with another language if we needed to... but we're a data-science company, and unlike many other co's or startups out there, we hire data-scientists who write features and applications directly. R makes this possible.

We have an Android App as part of our services - we use RServe as the API interface so we can call R functions from the App.

To your other point - web development skills are most definitely required, as are good engineering practices, and a stable stack. Shiny is just like any other framework; it takes a range of skills to create an application people want to use and pay for.

We use R and Shiny server pro for our back-end. It's hosted on AWS EB with our static site on S3. We love it - the data scientists can write and push features directly, and everything (including our API's) are easily maintainable in R. I've been shouting the praise of R and it's usefulness as a production ready language for a long time.

Contact me with questions about data science, start ups, or R - if you have a data backed start up, R + Shiny is the way to go.

For anyone interested in quantitatively measuring the flavor profile of your coffee, and seeing how dosing and other changes effect your perceived quality of your daily (hourly?) ritual....

My startup [0] makes an Android App just for the occasion! We run statistical quality control and flavor profiling for coffee roasters, beer brewers, and distillers, so your reviews may actually help your favorite coffee companies make better roasting and sourcing decisions.

[0] www.Gastrograph.com [1] https://play.google.com/store/apps/details?id=com.gastrograp...

All of our API's are built in R using rserve, and we share code and functions through an internal package hosted on bitbucket.

You can call R functions on a server from Android and return the results in a list or array - makes for some really cool internal API magic between our devs and data-scientists :)

My startups [1] does flavor profiling and statistical quality control for beer and bourbon producers - it's a fun job!

Our entire back-end is built in R, mostly within the Hadly-verse, and we use Shiny [2] as our web framework.

Our team works a bit differently than most, I suspect; our data-scientists build features and analysis directly in R, and then add the functionality to our Shiny Platform. Our "real devs" are all server + DB, or Android guys. This has created a great development system where all of the "cool findings" and 'awesome visualizations' are immediately implemented in our system, and made available for our clients!

[1] www.Gastrograph.com [2] http://shiny.rstudio.com/

---- EDIT ---

Edited to add; R is a great language and is 100% suitable for production systems. Its older than Python(!), and, with some experience, can be made in to high performance code.

Gotta love China; I had Wasp Larvae when I was living in Yunnan, and loved it!

The numbing was probably from Sichuan Pepper Corn [1] not the bees themselves!

[1] https://en.wikipedia.org/wiki/Sichuan_pepper

My company[1] did a bit of flavor profiling research for a few Ento farmers and startups. Once you get over your aversion to the idea of eating crickets or mealworms, their actually quite good! The crickets always reminded me of french fries (and I don't even eat fries....).

We have a vegetarian on our team who felt fine eating them. No pain, no harm.

Lobsters on the other hand... jury is out on weather they feel pain.

[1] www.Gastrograph.com

Can you talk a bit about your feature selection or models?

I run a statistical quality control company using machine learning, and picking up on flaws with tiny probabilities (one batch in every twenty or thirty million) might benefit from similar techniques!

Discretionary spending is different than fixed costs and contractual liabilities. I am saying that you should think of costumer acquisition as an upfront expense that leads to future profits - if the business is growing, and has reasonable margins.

For example, Amazon has 'razor thin' margins and is operating at a slight loss. Does anyone really argue that they are not 'Profitable'?

A company can be spending more than they make, and still be profitable (a la Amazon) - hence the "plowing money back into the business" comment above. A company in its growth stage believes there is significant profits to be made by spending on customer acquisition, and generally is making an informed decision not to pad the coffers and horde money (a la Apple).