## Estimating Uber Driver Pay, Net of Expenses: \$12.76 in Boston

[Note: there is now an updated version of the study, with slightly different hourly wage reported in Table 6. I’ve updated my post to reflect that, and attached a new version of the spreadsheet. 2/5/15]

Uber just put out a study by economists Jonathan Hall and Alan Krueger with average hourly pay rates for its drivers. They neglected to subtract all the expenses (gas, wear and tear, insurance, etc.).

As a consumer, I’m a big fan of Uber. Much of the organized resistance to Uber comes from the taxi/limo industry, which generally provides worse service, at a higher price, while exploiting its workers more than Uber does.

Even so, I am curious about whether Uber drivers make decent money. The drivers I’ve talked to seem to be aware of their gas expenses but only vaguely account for the capital costs of the wear and tear on their vehicles. So, I did a little back-of-the-envelope (er, spreadsheet) calculation to get a rough estimate, using Uber’s Boston data and a few assumptions that I document below.

Basically, I’ve tried to back out from their data an estimate of the number of hourly miles per driver. Then we can multiply that by a standard per-mile allowance that’s supposed to estimate total costs of operating a vehicle.

• Reported mean hourly earnings for Boston: \$20.29/hour
• Since Uber keeps 20% of fares, that means customers paid \$25.36/hour
• Reported median trips/hour = 1.67, with \$2/trip charge = \$3.34
• Estimated 30 minutes per hour on the meter, with \$0.21/minute charge = \$6.30
• That leaves \$15.72 in mileage charges, at \$1.20/mile = 13.1 miles hourly
• The Federal goverment lets you deduct expenses for operating a vehicle for business purposes at 57.5 cents per mile. That would come to \$7.53/hour.
• Net pay is then \$20.29-\$7.53 = \$12.76/hour

The report promises that Uber will provide some more detailed analysis of expenses in the future. I look forward to it. It mistakenly suggests that these expenses should be considered “net of taxes”. That’s a mistake because you only get to deduct them from income if you treat the whole revenue of \$20.29 as your income. It may be that 57.5 cents per mile is an overestimate. In that case, the actual expenses would be lower, and net income higher, plus the driver would get a little tax windfall of paying taxes on the lower income calculated after subtracting 57.5 cents/mile rather than the actual somewhat higher income.

How you view \$12.76/hour net of expenses depends on your perspective. It would be tough to live on that hourly wage in the Boston area, especially with no fringe benefits. On the other hand, it’s more than the minimum wage, and much better working conditions (set your own hours, partially control your surroundings).

BTW, I did the same calculations for the other cities they mention in their report. (spreadsheet). Not surprisingly, I get an estimate of more miles per hour (17.76) in LA, where things are more spread out, and fewer (9.47) in DC. Estimated net hourly wages range from \$6.90 in LA to \$24.29 in NY. NY may be misleading because I think they have some different arrangement in NY.

## Reasons for Replication in HCI Research

At the CHI conference last week, Max Wilson organized the RepliCHI workshop (I helped a little, but really it was Max). There were a bunch of presentations reflecting on the challenges of doing replication work, and lots of discussion about the roles and value of replication in HCI research.

Previously, including in the call for position papers for the workshop, we had offered a taxonomy that included direct replication, replicate+extend, conceptual replication, and applied case studies. Aside from everyone having difficulty assigning particular studies to these categories, these categories fail to highlight the purposes of replication studies, which I think are:

1. testing the validity or generality of previous findings;
2. calibrating or testing a measurement instrument;
3. teaching an investigator methods or findings in a way that they’ll really remember.

All of the distinctions we seem to want to make in a taxonomy of replication types revolve around what differs between replicating study and original and what inferences the investigator will make. So let’s tackle those separately.

What Changes and What Stays the Same

The ideal of a “direct” replication can never be realized. Some follow-on studies are closer to the original than others, but there is always some difference on at least one of the following dimensions:

• Investigator.
• People whose behavior is observed and/or manipulated. Examples would be switching from Korean to U.S. subjects, or college students to Mechanical Turkers.
• Time when the behavior occurs. In HCI, the phenomena we are studying are changing frequently as technologies change and social norms evolve.
• Context in which the behavior occurs, including experimental procedures for experiments or variations in physical and social environment or tasks for naturally occurring behavior. For example, at the workshop, a couple of the presentations involved experiments with either a search task or reading news articles, where it did not make sense to reuse the original tasks because of the passage of time. Another example would be moving from lab to field setting.

Inferences From Confirmatory Results

If the replicating study results match the original, we infer that the original result is reliable and generalizes to the new context. Generally, the more that has changed between original and replicating study, the greater the evidence the new study provides for robustness of the original finding (provided they match). There is always the possibility that multiple difference have had effects that canceled each other out rather than none of them changing the results. Some people at the workshop, notably Noam Tractinsky, strongly argued for changing things in small increments, to avoid that possibility.

If there was some reason to doubt the robustness (reliability or generality) of the original findings, and those findings have important implications for theory or practice, then a confirmatory result may be a significant contribution to the field. No further “extension” or follow-on experiment is necessary. Sometimes, however, the test of robustness will be more impressive if it is conducted several times with somewhat differing populations or conditions. Several workshop participants were concerned that some CHI reviewers and ACs do not value studies whose main contribution is to check the robustness of a previous result. For example, Bonnie John was concerned that reviewers for a paper this year (not hers) had understood and appreciated the contribution of a submission very clearly but had rejected it because it only confirmed a previous finding, even though the reviewers acknowledged and agreed with the authors’ argument that there were reasons to doubt the previous finding. Perhaps the underlying problem is that the CHI community is insufficiently skeptical of results that have been published at the conference. I think it’s fine that we publish studies, ethnographic and experimental, with small samples, but we should be far more skeptical than we are of the robustness of the findings from those studies.

If the measurement instruments (survey items; eye tracking device; counts of FB likes, qualitative coding categories) have changed, from matching results in the replication we also infer that the instruments are valid. This is the main reason to replicate first in the replicate+extend paradigm. In this case, the results of a follow-up study using the calibrated instruments are usually the main contribution of the research. In some situations, however, simply testing the robustness of a measurement instrument (e.g., a shorter version of a questionnaire scale trying to measure the same construct previously measured with a larger number of questions) may be a valuable methodological contribution.

Most workshop participants felt that research contributions following the replicate+extend model are well appreciated by CHI reviewers. One of the presentations at the workshop described a project that followed this model, and it had been accepted for publication/presentation at the conference.  I wish, however, that reviewers would be more critical of papers that omit the instrument calibration phase. The problem here is not rejection of studies that include replication but acceptance of papers that do not.

If the replicating study results contradict the original results, we do not know exactly why; any of the things that differ between original and replicating studies could be the cause, as could errors in conducting or reporting of the initial results. The more that has changed between the original and replicating study, the less we learn from contradictory results. This is another reason for some investigators to prefer changing things in small increments. Generally, contradictory results require follow-on studies to try to refine the original results by characterizing better the conditions under which they apply. That’s the situation we find ourselves with the NewsCube replication study that we presented at the workshop and as a work-in-progress paper at the main conference.

## Personalized Filters Yes; Bubbles No

On Thursday, I gave the closing keynote for the UMAP conference (User Modeling and Personalization) in Girona, Spain (Catalunya). I had planned to talk about by work directed towards creating balanced news aggregators that people will prefer to use over unbalanced ones (see project site). But then Eli Pariser’s TED Talk  and book on “Filter Bubbles” starting getting a lot of attention.  He’s started a trend of a little parlor game where a group of friends all try the same search on Google and look in horror when they see that they get different results. So I decided to broaden my focus a little beyond news aggregators. I titled the talk, “Personalized Filters Yes; Bubbles No.”

As you can perhaps guess from my title, I agree with some of Pariser’s concerns about bubbles. But I think he’s on the wrong track in attributing those concerns to personalization. Most of his concerns, I argue, come from badly implemented personalization, not from personalization itself. I’ve posted a copy of my slides and notes. For anyone who wants a summary of my arguments, here goes.

His first concern I summarize as “Trapped in the Old You”. I argued that personalization systems that try to maximize your long-term clickthrough rates will naturally try to explore to see if you like things, not just give you more of the same. This is the whole point of the algorithmic work on multi-armed bandit models, for example. Moreover, once our personalization systems take into account that  our interests and tastes may change over time, and that there is declining marginal utility, eventually, for more of the same (consider the 7,000th episode of Star Trek; even I stopped watching). Personalization systems that are designed to optimize some measure of user satisfaction (such as click-throughs or purchases or dwell time or ratings) are going to be designed to give you serendipitous experiences, introducing you to things you like that you didn’t know you would like. Moreover, even today’s systems often do that pretty well, in part because when they optimize on matching on one dimension (say topic) they end up giving us some diversity in another dimension that matters to us (say, political ideology). From introspection, I think most people can recall times when automated personalization systems did introduce them to something new that became a favorite.

His second concern I summarize as, “Reinforcing Your Baser Instincts”. Here, too, good personalization systems should take into account the difference between short-term and long-term preferences (entertainment vs. education, for example). We will need delayed indicators of long-term value, such as measuring which words from articles we end up using in our own messages and blog posts, or explicit user feedback after some delay (the next day or week). It may also be helpful to offer features that people can opt in to that nudge them toward their better selves (their long-term preferences). Here, I gave examples of nudges toward balanced news reading that you can see if you look at the slides.

His third concern I summarize as, “Fragmenting Society”, but there are two somewhat separable sub-elements of this. One is the need for common reference points, so that we can have something to discuss with colleagues at the water cooler or strangers on the subway. Here, I think if individuals value having these common reference points, then it will get baked into the personalized information streams they consume in a natural way. That is, they’ll click on some popular things that aren’t inherently interesting to them, and the automated personalization algorithms will infer that they have some interest in those things. Perhaps better would be for personalization algorithms to try to learn a model that assumes individual utility is a combination of personal match and wanting what everyone else is getting, with the systems learning the right mix of the two for the individual, or the individual actually getting a slider bar to control the mix.

The second sub-concern is fragmenting of the global village into polarized tribes.  Here it’s an open question whether personalization will lead to such polarization. It hinges on whether the network fractures into cliques with very little overlap or permeability. But the small-world properties of random graphs suggest that even a few brokers, or a little brokering by a lot of people, may be enough to keep average shortest path short. Individual preferences would have to be strongly in favor of insularity within groups in order to get an outcome of real fragmentation. It turns out that people’s preferences with respect to political information, as concluded by my former doctoral student Kelly Garrett, is that they like confirmatory information but have at best a mild aversion to challenge. Moreover, some people prefer a mix of challenging and confirmatory information and everyone wants challenging information sometimes (like when they know they’re going to have to defend their position at an upcoming family gathering.) Thus, it’s not clear that personalization is going to lead us to political fragmentation, or any other kind. Other forces in society may or may not be doing that, but probably not personalization. Despite that, I do think that it’s a good idea to include perspective-taking features in our personalization interfaces, features that make it easy to see what other people are seeing. My slides include a nice example of this from the ConsiderIt work of Travis Kriplean, a PhD student at the University of Washington.

The final point I’d like to bring up is that personalization broadens the set of things that are seen by *someone*. That means that more things have a chance to get spread virally, and eventually reach a broader audience than would be possible if everyone saw the same set of things. Instead of being horrified by the parlor game showing that we get different search results than our friends do, we should delight in the possibility that our friends will be able to tell us something different.

Overall, we should be pushing for better personalization, and transparent personalization, not concluding that personalization per se is a bad thing.

At the conference banquet the night before my talk, attendees from different countries were invited to find their compatriots and choose a song to sing for everyone else. (The five Americans sang, “This Land is Your Land”). Inspired by that, I decided to compose a song we could all sing together to close the talk and the conference, and which would reinforce some themes of my talk.  The conference venue was a converted church, and the acoustics were great. Many people sang along. The melody is “Twinkle, Twinkle, Little Star”.

(Update: someone captured it on video: )

The Better Personalization Anthem

User models set me free
as you build the Daily Me

could just be that I’ll want more

UMAP scholars make it so

Words: Paul Resnick and Joe Konstan

## Yelp gets more reviews per reviewer than CitySearch or Yahoo Local

Author attributes it to the fact that reviewers are anonymous at CitySearch and Yahoo local, but build up reputations on Yelp. Of course, there are also other differences between the sites.

Zhongmin Wang (2010) “Anonymity, Social Image, and the Competition for Volunteers: A Case
Study of the Online Market for Reviews,” The B.E. Journal of Economic Analysis & Policy: Vol.
10: Iss. 1 (Contributions), Article 44.
Available at: http://www.bepress.com/bejeap/vol10/iss1/art44

Abstract:
This paper takes a ﬁrst step toward understanding the working of the online market for re-
views.   Most online review ﬁrms rely on unpaid volunteers to write reviews.   Can a for-proﬁt
online review ﬁrm attract productive volunteer reviewers, limit the number of ranting or raving
reviewers, and marginalize fake reviewers?  This paper sheds light on this issue by studying re-
viewer productivity and restaurant ratings at Yelp, where reviewers are encouraged to establish a
social image, and two competing websites, where reviewers are completely anonymous. Using a
dataset of nearly half a million reviewer accounts, we ﬁnd that the number (proportion) of proliﬁc
reviewers on Yelp is an order of magnitude larger than that on either competing site, more produc-
tive reviewers on all three websites are less likely to give an extreme rating, and restaurant ratings
on Yelp tend to be much less extreme than those on either competing site.

## Need recommender systems contest ideas

Do you have an idea or plan for a future challenge/contest that you think could move the field of Recommender Systems forward? I’d love to hear about your idea or plan, even if only in sketch form, and even if you’re not in a position to carry it out yourself. At this year’s RecSys conference in Barcelona, I’ll be moderating a panel titled, “Contests: Way Forward or Detour?” As part of that panel, I’d like to present brief sketches of several contest ideas for the panelists to respond to.

———————-Abstract of the Session
Panelists:
Joseph A. Konstan, University of Minnesota, USA
Andreas Hotho, University of Würzburg, Germany

Contests and challenges have energized researchers and focused attention in many fields recently, including recommender systems. At the 2008 RecSys conference, winners were announced for a contest proposing new startup companies. The 2009 conference featured a panel reflecting on the then recently completed Netflix challenge.

Would additional contests help move the field of recommender systems forward? Or would they just draw attention from the most important problems to problems that are most easily formulated as contests? If contests would be useful, what should the tasks be and how should performance be evaluated? The panel will begin with short presentations by the panelists. Following that, the panelists will respond to brief sketches of possible new contests. In addition to prediction and ranking tasks, tasks might include making creative use of the outputs of a fixed recommender engine, or eliciting inputs for a recommender engine.

## Gerhard Fischer paper at C&T

“Towards an Analytic Framework for Understanding and Supporting Peer-Support Communities in Using and Evolving Software Products” at C&T

Participation in SAP’s online community.

Before/After point system introduced in SAP:
Mean response time decreased (51 min vs. 34 min.)
Mean helper count increased (1.89 vs. 2.02)

Some evidence of gaming of the system—people just ask questions to gain points.

## Karim Lakhani at C&T

Karim Lakhani is giving a great keynote at C&T about tracking innovation. He has worked with MatLab programming contests that have a fascinating format. There’s clear performance outcome; source code of all entries is available to other people; leaders are tracked. Researchers can track which lines of code get reused.

–novel code
–novel combos of others’ code
–NOT borrowed code
–complexity
–NOT conformance

–novel code
–novel combos of others’ code
–borrowed code
–complexity
–conformance

Also did experiments with TopCoder.
One experiment with computational biology contest problem.
Three conditions (random assignment?):
Fully collaborative vs. fully competitive vs. mixed (competitive first week, then all code shared)
Fully collaborative got the best performance
Best performing entries did better than state-of-the-art in computational biology

## Universally Utility-Maximizing Privacy Mechanisms

Interesting paper presentation by Tim Roughgarden.

He gave a nice introduction to the recent literature on provably privacy-preserving mechanisms for publishing statistical summaries such as counts of rows from databases satisfying some property (e.g., income > 100,000). Suppose a mechanism computes the actual count, and then reports something possibly different (e.g., by adding noise). There is a definition of a p-privacy if, for every possible output (count), for any person (row) the ratio of the probability of that output with the row present to the probability of that output with the row omitted is always in the range [p, 1/p]. Intuitively, whatever the actual count, there’s not much revealed about whether any particular person has high income.

One technique that works for counts, LaPlace-p, is to add to the correct count +/- z, where probability of z is 1/2(-lnp)e^^(z/lnp). For any reported count, there’s some confidence interval around it, and the size of that confidence interval is independent of the count. Thus, for reported count 1, you can’t really tell whether the correct count is 1 or 0, and thus you can’t really tell whether a particular person has high income, *even if you have great side information about everyone else in the database*. On the other hand, if the reported count is 27,000, you still can’t tell much about any one person, but you can be pretty sure that the correct count is somewhere around 27,000.

Roughgarden’s paper is about how much value you can get from the best count function (in terms of some loss function comparing true result to reported result) while still preserving the p-privacy requirement. It turns out that a mechanism very much like LaPlace-p, but discretized, works to minimize the expected loss no matter what the user of the count’s priors are about the distribution of correct counts. It is in this sense that it is universal. This requires a little user-specific post-processing of the algorithm’s output, based on the user’s priors about the correct counts. For example, if the reported count is -1, we know that’s not the correct count; it must really be 0 or something positive, and you can back out from the report and the user’s prior beliefs to infer a belief distribution over correct counts.

## Babaioff; Characterizing Truthful Multi-Armed Bandit Mechanisms

Moshe Babaioff presented an interesting paper.

Suppose that you’re conducting an auction for adwords, where you want to rank the bidders based on expected revenue in order to allocate slots and determine prices for slots based on bids. But suppose you don’t know what the clickthrough rate will be for the items.

In a multi-armed bandit model, there are multiple bandit slot machines and you have to decide which arms to pull. There is an explore/exploit tradeoff– you need to explore (experiment) to estimate the clickthrough rates, including some experimentation with those you have low estimates for, in case that estimate is wrong. But over time you switch to more exploitation, where you pull the arm of the highest expected value.

The new twist in this paper is that you want advertisers to truthfully reveal their valuation for a click. If clickthrough rates are known, you can set price essentially using a second-price mechanism based on bid*clickthrough. But if you’re using a multi-armed bandit algorithm to determined clickthrough rates, the correct prices would depend on estimated clickthrough rates that you don’t necessarily see because you don’t test them.

It’s a theory paper. They prove that, with the requirement that the mechanism induce trruthful bidding, there’s always a pure exploration phase, where the selection of the winners can depend on previous clickthroughs but *not* on the bids; and then a pure exploitation phase, where the clickthroughs no longer affect allocation of slots in the next round. The best multi-armed bandit algorithms without the truth-telling requirement don’t have that separation of phases. And, it turns out that the best algorithms without the truth-telling requirement have less “regret” relative to the best you could do if you magically knew the clickthrough rates at the beginning.

So now I’m curious what the best algorithms are without the truth-telling requirement. My guess is that they put more exploration into things that the best estimate so far has higher value for. We actually need to use an algorithm like this for the next version of our “converation double pivots” work on drupal.org, where we’re going to dynamically change the set of recommended items based on a combination of prior generated from recommender algorithms and actual observed clicks. But we don’t have any truthful revelation requirement, so we should be able to use the standard algorithms.