1. …doing stuff…
2. Decide to create new space
3. Attempt to create space, discover that you need an administrator to do it
4. Find out who the administrator is
5. Ask them to do it
6. Wait

Alternative workflow #1: approval only

For the non-administrator:

1. …doing stuff
2. Decide to create a new Confluence space
3. Enter the form, fill out all the details, press Ok.
4. (Confluence sends a “Is this ok?” email to administrator)
5. Wait

Alternative workflow #3: public until reverted

Most users aren’t destructive. And especially in professional environments, virtually all users can be trusted to act in good faith. (Managers strangely predict “chaos” if many users are authorised). So, let them do it:

1. Decide to create Confluence space
2. Create Confluence space
3. Work in new space

1. Click Buy it Now
2. Sign in
3. Click Commit to Buy
4. Click Pay Now
5. Click Continue
6. Sign in again (Paypal this time)
7. Click Continue (confirming payment type?)
8. Click Confirm Payment

1. They’re delivered as unstructured narratives.
2. They mix identifying information (“carrots”) with process information (“thinly sliced”)
3. They’re overly specific (3/4 of a teaspoon, does it matter?)
4. They have too many ingredients, and you don’t know which ones you can leave out
5. They always have one or two ingredients you wouldn’t have lying around. Shrimp paste?

I can’t fix all of that. But here goes:

Cooking for engineers: spicy cauliflower and almonds

A cauliflower dish I made up the other night, seen by an engineer.

Recipe for non-engineers

Here’s what a conventional version of that dish might look like:

Ingredients:

1/2 cauliflower, chopped

1 bok choi

1 tbsp coriander, chopped

1/2 cup slivered almonds

1 tbsp ground turmeric

1 tbsp mustard seeds

1-4 tsp chilli powder, to taste

Put 2cm of water in a saucepan and bring it to the boil. Simmer the cauliflower 2 minutes, then drain, discarding the water. Add half the sesame oil, and return the cauliflower to the saucepan.

Meanwhile, heat the rest of the sesame oil in a frying pan. When hot, add the slivered almonds and remove from the heat, stirring continuously. Once browned, add the almonds to the frying cauliflower, and add the spices. After a few minutes, add the bok choi. Stir frequently until the bok choi is soft, then serve with rice and roti bread.

Principles of cooking for engineers

Here are the rules:

1. One column per preparation dish (saucepan, mixing bowl, tray…)
2. Red arrows show cooking. Maybe thicker arrows for hotter.
3. Ingredients start off to the sides. Name of thing in bold. Quantities as rough as appropriate, in parentheses.
4. Thin blue arrows show transfer of stuff.
5. Text between two black lines means “keep doing the thing above until this happens”
6. Stuff you need to do (processes) in square boxes.
7. Collapse stuff down wherever appropriate. It’s a communication tool, not an exhaustive process analysis.

Explaining step 7: the diagram above could have shown a colander as another column, with cauliflower transferred from the saucepan to the colander and back. But why would you do that?

Discussion

Normal recipe format works pretty well for most people, but I find it takes multiple readings before I can start. Effectively, I’m constructing this kind of flow chart in my head, working out what gets transferred from where, to where. I hope that a refined version of this diagramming methodology would allow you to confidently dive straight in, with no nasty surprises.

On the downside, diagrams are hard to manage. I realise I forgot an ingredient (crushed garlic and ginger paste), but it’s time-consuming to modify the diagram and re-upload.

And lastly, yes, it’s a bit facile to call this “cooking for engineers”. Probably real engineers want precise measurements, correct use of flow control symbols and so forth.

1. Automatically create a label for every group that I receive mail from. Don’t wait for me to do it.
2. Make these labels more than just a label. Give them options like “skip the inbox” or “delete on sight”.
3. Have a view which shows all mail from all mailing lists, with some smart options to make this even more useful.

For extra credit:

4. Detect other sources of regular mail which are not “mailing lists” as such. Newsletters from the bank. Quarterly updates from my alma mater. Treat them exactly the same.

Most of the features are already there, but it’s so tedious having to set up a special rule and label for every single list.

1. There are consumer electronics for auction, usually at big discounts.
2. It costs a certain amount to make a bid, regardless of whether that bid is ultimately successful. For bidrivals.com, it’s 40 British pence.
3. Every bid raises the price by a fixed amount. In this example, by 1c. It also extends the auction to last another 15 seconds or so.
4. If you “win” the auction you must then buy the item at the final price.
5. It’s not a lottery. Because they say so.

At first glance, the auction looks great – buy a phone for $20! Buy a plasma tv for $1.53!

But not so fast. A couple of things that are not obvious to the beginner:

1. Every dollar of the final price represents $40 in bidding fees. A $1000 TV selling for $1 is a big loss for the site. The same TV selling for $25 is a small profit. Sold for $1000 it’s a $40,000 profit.
2. You can use a site-provided bot (“bidbot”, “bidbutler”…) to bid on your behalf. If two people do this simultaneously, they’ll both lose a lot of money with no apparent gain.

So, is it a scam? Well, there are really two quesions:

1. If the site is running completely as described, legitimately, and not using shill bidders (bidding on their own auctions), is this an honest way to make a living – and should you participate?
2. How do you know if a site is legitimate? Is it likely to be?

Is this honest?

I see very little to distinguish these penny auctions from gambling:

• When you bid, whether you win or not depends entirely on whether anyone else bids in the next 15 seconds. Assuming you’re bidding on an item which is clearly a bargain (eg, $5 for a TV), then the normal considerations of auctions do not apply: any rational person would bid if they could so for free.
• The house take is enormous. Frighteningly so. For example, imagine on average the site sells items at a 65% discount from RRP, and bids cost 40 times as much as the amount they increase the value by. This means that it costs on average (100-65)x40 to win a $100 item (whose value is now $65), or in other words (100-65)x40/65=$21.50 to win one dollar’s worth of value. By comparison, a skilled blackjack player in a casino can pay as little as $1.01 to win $1′s worth of value.
• Most bids give no return to the bidder. This means that even if you don’t want to call it “gambling”, it should still be regulated, as the potential for dishonesty is great. You don’t want to be bidding for a dead donkey.

There is quite a bit of antipathy towards these sites: Washington Post, Jeff Atwood, Ed Oswald.

Can you trust them?

There are two main risks:

• The site may use “shill bidders” to bid on items that would otherwise go for a low price. This could prevent you ever winning, or cause you to spend far more than you want.
• Even if you “win”, the site may never ship. The whole thing could be a scam.

Fortunately, there are sites on the look out for this kind of thing, such as pennyauctionwatch.com. There is evidence of dodgy sites, such as fake testimonials.

So, what are the incentives for a site to use shill bidding? Well, as we saw above, the difference between a $1000 item selling for $1 and $25 doesn’t look like much, but it’s the difference between breaking even and posting a big loss. Imagine there is fairly steady bidding activity, but there are just a few gaps before that $25 mark. If the site could shill just a few times, they would massively increase their profitability.

But how much should they shill? Consider two strategies:

1. bid whenever the time gets to 1 second; or
2. bid immediately after anyone else bids.

In 1), the shill bids guarantee almost any asking price, as long as there is still some demand. This has the potential to greatly increase profit, and decrease variance.
In 2), half the bids end up being shill bids. This causes two problems: first, you’re directly losing one bid fee for every shill bid. Second, by inflating the price, you’re accelerating reaching the point at which people no longer want to bid, because the prize at stake is shrinking. So if people might normally bid strongly up to half the value of the item, then shilling along the way is just replacing paying bids with free ones. You might even decrease the final sale value, and every dollar of sale value lost is $40 of bidding fees lost.

Conclusion: shill bidding seems likely to occur, in small doses, because the incentive is just so strong.

Can you beat them?

Probably not. It’s been tried. To beat it:

• You have to find a site that is not a complete scam.
• You have to find a site that is completely honest. Even a little bit of shill bidding will crush you.
• You have to defeat an absolutely incredible house take of 95% (remember, normal house take for gambling ranges from 1% to 5%).
• You have to know enough about the auctions and your fellow users to make you fairly confident that no one will bid in the next 15 seconds. In the $1000 TV at $25 case (40c to bid), you need there to be a greater than 1/250 chance that you will win the auction with this one bid. Sound easy? Think: if that were the case, how did the price get to $25? It would stop, on average, at $2.50.

If it is beatable, you’d think people would have done it. And, since they’re capped at 4 wins per month generally, there would not be much harm in them sharing their secret. Unless they have a network of penny auction-beating bots. There’s a thought.

But just in case you wanted to try:

• Compare sites. Find a safe one that appears to be losing money.
• Collect lots of data. Try GreaseMonkey.
• Find the right time of day, with the least competition.
• Track all the auctions, pick individual moments and place bids.
• Don’t try and win a specific auction. Bid any time your positive expectation on that bid is positive. The moment could pass.
• Consider the effect of distractions. A good moment might be when several auctions are closing at the same time. You could even engineer that by bidding on several simultaneously.
• Consider using several accounts to bid with, to drive off other bidders. If you know they’re paying attention and will react appropriately, that is. I’m thinking you bid with a group, gradually spacing their bids further apart and hoping you can sneak a 15 second gap through.

More reading:

• http://en.wikipedia.org/wiki/Penny_auction
• http://pennyauctionwatch.com/directory/
• http://www.webscriptsdirectory.com/PHP/Auctions/BPPennyAuction—Penny-auction-script-like-Swoopo—PHP-L5831/ – run your own!
• http://penny-auction-insider.blogspot.com/ – apologist for penny auctions?