The OLPC / OSS Dream
The always-amusing XKCD webcomic illustrates the secret dream on the OLPC project in encouraging children to learn and, in doing so, learn programming:
« June 2008 | Main | August 2008 »
July 30, 2008
digg this
del.icio.us
reddit
Slashdot
Stumble It!Zero to Fame in Three Hours
Today, Twitter launched one person from their normal Internet life to getting news on the California-regional LAist and valleywag blogs, CNet, a top-rated digg story, a google search term all to herself, fan-created artwork, and a skyrocketing number of followers inside Twitter. In three hours, with one twitter.
Read that again. Three hours, and a message that fit in under 140 characters. If there was ever an idea that was "Made to Stick", (pardon -- deep apologies, really -- for the truly tasteless pun.)
What created this success?
Some things we can learn from - it was relevant to very current events, it was personal and engaging, and funny. While not all messages can or should be funny, pulling out an emotion helps.
A rundown of the event, the C|net article, and a more serious VentureBeat story on the power of flickr.
Let me share the "message" that, stuck er... was so popular to propel the twitterer via a massive viral spread to Internet fame:
I am totally serious. My Ob/Gyn was IN my vagina and an earthquake started rattling the room!
Yes, that one deeply personal update about bad timing at an OB/Gyn put MissRFTC on the Internet Map. If only we all could create such a ... compelling message.
July 28, 2008
The world's (new) cheapest laptop (for today)
Following fast on the heels of the CherryPal low-cost computer is the Impulse, advertised at $130 for each laptop (but you have to buy 100 at a time). It has, frankly, unexciting specs and wifi is an optional dongle (ew). Network World has more details on the Inspire, but more importantly, the blossoming 4PC market:
The laptop hints towards a trend of lowering PC prices. Last week, a company called CherryPal introduced a $249 mini-desktop, also running a 400MHz processor, with 256M bytes of RAM and 4G bytes of internal flash storage.
Now, of the current contenders in the sub-$500 laptop market, Asus' Eee still seems to be the clear winner in the open market. The OLPC XO-1 was way out in front, and (I posit) still maintains an advantage thanks to some of their tech (screen, wifi, battery life, ruggedness).
However, it's just a matter of time before one manufacturer decides to not be braindead and balance the price, power, performance and portability all together, and basically win the market. This of course seriously endangers the One Laptop Per Child project, if other, more logistically adept companies produce a system comparable in utility to the XO (especially if the XO sheds all of its value-adds in its move from Sugar to Windows XP).
Of course, it's possible that the best thing to fulfill OLPC's goal is for this exact thing to happen.
July 27, 2008
Slashdotted
I seem to be up at Slashdot.org again with the OLPC; this time with my OLPCNews piece combing through the video of Windows on the OLPC (also published on JonCamfield.com. The comments are a lot better this time around, with a few complaints about my clear anti-MS bias.
You know when slashdotters are complaining about your anti-Microsoft bias, you must be doing something right.
To amend my comments and clear any doubt of where my bias lies; I added this as a comment to my XP/XO comparison:
Now, I could be totally wrong; but I somehow don't think that a few months of hard work by the folks over at Microsoft somehow magically turned XP into a low-end-computer miracle working educational platform and also solved all of the various XP security holes at the same time.Also, I'll freely admit it -- I'm biased here. I think XP is the end result of a bad marketing plan and will be a continuance of the same in the ICT4D world -- not a disaster, per se, but far from a success. The fact that countries have been asking "does it run Office?" means that it is being marketed as a laptop, not as an educational tool with a lot of power and flexibility built in. So people are treating it like, frankly, an IT project and not an educational project. At that point, why bother with this funky little thing? Get a beige-box system; or an Asus. Or focus on local telecenters instead of 1:1 computing programs.
Do countries have the right to choose to run XP on the OLPC? Sure. Do I think it's a good idea? Not really. So yes, this is a biased article, but I hope I presented what appear to be the most grievous problems in using XP as an educational tool, compared to the current Sugar release (The joyride releases are much more responsive and with a cleaner, more consistent UI).
July 19, 2008
OLPC GrassCon
Watching the GrassCon
July 15, 2008
mBanking: Fears and Promises
This morning's Technology Salon covered the legal hurdles facing mBanking - using your cell phone to interact with your bank account - in developing world scenarios.
Cell Phone in rural India
The three general models presented provide a good structure for seeing how mBanking is being approached. First is banks adding mobile services as value-adds to their current customers. This is seen often in more developed situations, and is not intended to target the unbanked or new customers. Second are banks specifically seeking new customers (who do not currently have any bank account) through tailored mobile banking. Finally, we see Mobile Network Operators - MNOs - adding mBanking features to their services.
The technical problems will not surprise anyone who's dealt with mobile phones - differing technical requirements and interoperability (of the phones and of the networks). Basic (and therefore, cheaper and more common) phones do not have much to offer for security -- obviously a tough issue if there's to be reliable financial transactions. More advanced phones have the unfortunate downside of being more susceptible to virii and malware. Plus most banks are not particularly invested in having a shared platform, as they would prefer to create proprietary solutions discouraging users from switching banking providers.
The regulatory and legal problems are largely unsurprising as well - how can you "prove" in court that the requested mBanking transaction was approved by the user - is an SMS admissible evidence? Does the country even have an eSignature law to enable electronic authorization? There are massive consumer fraud protection issues, banking regulations (does a mobile provider have to maintain liquidity in case of an "mbank" run?), not to mention which department or ministry of the government will oversee the laws - will it be the technology/IT ministry or the Finance ministry? Once you also enable cross-border transactions, you have to also be very careful of anti-money-laundering (AML) rules, which is all the more difficult with the specific base of the pyramid target market who could best benefit from mBanking, but who may not have a verifiable home address in any normal sense of the concept.
The silver lining to all these problems is that most of them can be addressed through decent regulations - but naturally that presents its own challenges. Regardless there's strong value (e.g. tax revenue for the government) in formalizing the multi-billion-dollar informal economies that mBanking could help with, so there's pressure to create technology neutral but bank-policy-regulated laws that balance usefulness, consumer protection, and limiting monopoly growth, but whether those will overcome the resistance to other problems, including created currencies like cell phone minutes as value (or Final Fantasy codes).
There are huge promises in mBanking for the BoP market, without a doubt. Artisans, small farmers and the like could get paid remotely for their work, and could get commissions or advances for specific tasks. However, the technology and the regulatory framework need to be, in combination, bulletproof and safe for consumers to use without fear. In the case of MNOs being the bank; they must be required to act like a bank in terms of insurance and liquidity. In the case where the MNO is merely a passthrough to the bank, both the MNO and the banks must be able to prevent fraud and crime, provide risk management strategies for themselves and their customers, and find some reasonable transaction security. The governments must be involved creating a strict but not stifling regulatory framework that protects against predatory practices, fraud, and overly rampant capitalism and monopolistic tendencies.
You can read my full notes from the event at Wiki4Dev
July 14, 2008
OLPCs with CellModems
Someone's stealing my best ideas
Wayan found a gem in the Times of India article on OLPC and World Bank funding:
Later this year, the XO laptops are expected to hit the retail stores. Sources say Reliance Communications, which partnered OLPC Foundation to conduct an XO pilot project in Maharashtra last year, is looking at retailing these laptops bundled with its CDMA modems.
An OLPC Entrepreneur?
The same entrepreneurial idea can feed development, using the OLPC technology instead of (or possibly in addition to) cell phones and PVs. Set up a group of in-country micro-lenders who can walk someone through the usage of the OLPC XO laptops, evaluate requests for laptop loans with local situational and social knowledge, and help with initial setup. Provide micro-loans to individuals with an idea of how to use the laptop in a way that could generate enough revenues for repayment and self-employment. Work with local social customs and systems to find the best way to create social pressure for loan repayment (only x amount of money is available on a rotating basis?), as well as adapt to local markets and needs.
Can the OLPC turn a BoP profit?
Below are a few ideas (presuming some form of Internet, probably cell-phone-network enabled) that could combine the OLPC, community development, and education with making a bit of profit. There are a million other possible things to do with the laptop, using its built in hardware and software tools as well as adding other open-source software to it, so this is by definition an incomplete list. Only local agents can really know what the local demand for OLPC-related services would be, so take these as very basic, generic ideas:
- Youth could create radio programs with local advertising -- youth gain experience in writing, public speaking, budgeting, aspects of radio operation (physics lesson on radio waves?), as well as marketing. Local industries could advertise goods during their radio program, and this isn't even getting into the FOPSE (For-profit Social Enterprises) possibilities like the LapDesk.
- The OLPC could be used as a traveling/home-visit cybercafe and "digital office" (some tasks might require a portable printer as well) to provide services like:
- Letter/resume transcription and/or typing
- Contact (skype/voip with family abroad?)
- Interaction with eGovernment services
- Access to current market prices for locally produced goods
- Manage an eBay store of artesania / handcrafts
- Remote basic medicine and consultation with urban-based doctors
- Of course, email/chat/web surfing/entertainment and the like if there's a demand for such services
- Schools (or other groups) could offer the public training and adult education -- the laptop is built to support education; so it's an ideal machine to support training in basic computer skills (typing, mousing, etc.); literacy and numeracy, and so on.
So I hope that the Times of India article has their facts straight, and I hope someone's reading -- and implementing -- our thoughts here and at OLPCNews.com.
OLPC joyride no so joyful (or, what Jon did all day Sunday)
So, this weekend I thought it'd be a great time to upgrade to the latest joyride builds, which are rumored to have solved the earlier record problems, and hopefully the SD card corruption issues as well. This is supposed to be an only-mildly-painful experience, with a few command line tricks, a few boot tricks, and so on. Nothing serious.
XKCD on upgrading difficulties (and sharks)
And, in fact, it updated just fine. I wasn't too impressed at the increased speed that's been promised, but the UI had a few enhancements, and I was looking forward to improved power management.
Now, I use my OLPC for about 4 main tasks - taking notes, checking email/Internet, playing Implode, and watching movies on planes (and TED talks on the Metro!). All tasks I find the XO to excel at, frankly - it's light yet rugged enough to always carry, and with the great wifi reception and sun-readable screen, it's rare that I can't make full use of the XO; whether I'm at the park, home, or near anyone with an open wifi connection.
Booting into the new joyride build I find many improvements to the UI and network views, some that were interesting but neither exciting nor bad, and that mplayer doesn't work for me. In fact, it causes a hard crash. Same with Record, Measure, TamTam.... basically anything fancy. The activity loads, but whenever it tries to actually play the video or activate the camera or mic, it freezes. The wifi lights continue to flicker, and the battery light will change state, but no keyboard or mouse response.
I thought this was an unfortunate joyride issue, and restored back to 703 using olpc-update; same problem. I then did a secure reflash from USB (twice!) and tried an olpc-update from there to restore functionality, but no luck. Booting into open firmware and running test /camera provided a moving camera image, so it's not (thankfully!) a hardware problem; and I don't think it's the activities themselves (I downloaded a Record from the Activities page to test). If I had to guess, I'd think it was a driver/module problem or at the X level, but I'm not familiar enough with Linux or Sugar to really diagnose or fix that.
I continued to muck around in discussion with some folks on the irc.freenode.net #OLPC-help channel, and sent an email into the support gang at laptop.org. cjb suggested I try yet another reflash, starting from a completely blank slate - reformatted USB stick, dowloading a new image, everything. The second time I tried that, it magically took and started working. Continued discussion with the support gang indicated that sometimes the reflashing doesn't fully take, especially when going backwards in builds.
Needless to say, I might wait a while before going on another joyride.
July 08, 2008
Many Paths Leading to the "Base of the Pyramid"
Many Paths to the Base of the Pyramid
(on Flickr by Merkur*)
"Despite its flaws, Just Another Emperor does a superb job of fulfilling Edward's main intent - deflating the hype around philanthrocapitalism without denying it its place as a tool for combating poverty. Edwards reminds us that the free market cannot solve all social ills and inequalities. While noting the benefits of approaches championed by social entrepreneurs and venture philanthropists, he suggests that these movements complement - rather than replace - non-market-based approaches to poverty and sustainability."
Edwards makes the strong argument that
"The philanthrocapitalists love handing out new prizes-for building private spaceships and electric cars, sequencing the human genome, and ending global warming-but not for the [Swan Lake Fire Department] Ladies Auxiliary or reviving New Orleans."
The NextBillion writers respond to his basic point - market solutions only address a limited scope of attractive projects:
For example, a market-oriented development model might address the issue of rural access to energy by promoting an enterprise that sells low cost solar panels. A more traditional civil society organization might work to build a grassroots network of groups that call governments out for favoring urban over rural populations in the distribution of public resources.
Naturally both Edwards' favored civil society, as well as social entrepreneurship, are both tethered to traditional development models through grants and policies, so neither is wholly ground-breaking.
Edwards notes that "There is no place for triumphalism in this conversation." - I think that can go deeper; there is no place for triumphalism in development approaches. In some cases, governments and educational ministries/departments will be the best able to address a problem, sometimes civil society, communities, and NGOs, and sometimes the marketplace itself. One of the long-standing problems I've had with traditional development models is the one-size-fits-all approach it is prone to. Combine that approach with lenders without accountability for bad loans, and you have a collection of projects; some good, some disastrous failures, but all being pursued in a one-minded way. Perhaps big infrastructure projects should sometimes be government-funded, but sometimes perhaps the private sector can also help (and provide long-term sustainability to boot!) Sometimes local NGOs are the only way to achieve long-term change from the bottom up. There's a place - a potentially large and powerful place - for the private sector to address, help, and still profit from the base of the pyramid. That doesn't mean that all development work should, or even can be private-sector led.
There's a lot to change globally to create a better world, and we should work on providing an inclusive model to work with and support anyone who thinks they can lend a hand instead of fighting turf wars for the dominance of any one approach.
Could Many Give Many?
Participate (by our rules)
I wrote about this general problem first in a long and academic paper when OLPC was still selling in only lots of a million laptops and only to governments. I railed on OLPC for missing the importance of the small but well supported projects in favor of unmanageably huge (but big-number) projects, and proposed a solution -- peer networks of small schools, governments, and any other interested parties banding together to be able to meet the minimum order.
The original problem has evaporated as the former launch partners never came through with their promises, as Nicholas Negroponte discussed at TED last winter, and the project has moved to smaller installations which have been difficult enough to manage without an implementation strategy.
Production scales, why not giving?
This is all a long-winded way to get back to our current problem; individual funders and NPOs/NGOs who aren't able to or are even interested in enough laptops to use the Give Many program, but a growing support network for grassroots OLPC programs in the "Post 1CC" era.
Also, we have the potential for arbitrage courtesy of eBay to catch some of the slack. OLPC XO-1s have been selling at close to $300 on eBay since since Day 2 of the original G1G1 (actually the price has been in decline since G1G1 it seems).
At Give 1000, the cost per laptop is $260 - that's just 10 programs wanting 100 laptops each. $26,000 isn't chump change, but it's cheaper than any other way to get 100 laptops for a school, community or organization. We know that Ken Hargesheimer wants 150 XOs, and I'm sure there could be homes for 850 other XO laptops at $260 each (with some additional shipping overhead to re-distribute the lots). And of course, until G1G1 returns (rumored to be September) you could potentially eBay a few unwanted XOs...
Is this realistic? Are there enough interested-but-frustrated parties to assemble an order for 1000 XOs? Would OLPC take offense at this slightly tricky order? Is the random 3-6month order delay acceptable for the potential buyers?
eBay Market Data
In thinking about eBay in my post on tricky ways to "Give Many" OLPC XO laptops, I was reminded about something that has bugged me for a very long time about eBay.
eBay's data goldmine
And their data on these market prices is good, and well-accepted. It's standard practice in the nonprofits I've worked with to check the "fair market value" for donated technology by searching for something similar on eBay in order to provide a tax receipt to the donor. And naturally the price itself is the market price because, simply put, eBay is the market for these goods.
It looks like they do provide a set of this data (for a price) as part of the eBay Market Data Program:
The eBay Market Data Program offers rich consumer insight data about what is purchased on eBay and who is purchasing it. Access to this business intelligence can help you make effective buying and selling decisions for your business, regardless of whether you do business on eBay or outside of the eBay Marketplace.
Now, I'm not sure what the next step really is. I'd love to know if eBay would make a subset of this market data (scrubbing personal information out, naturally, but price, sale date, item description, and bidding history) available for free. I'm sure some stats folks would love it, as would market researchers and social scientists in general. I'm sure there are millions of interesting questions to ask and visualizations to create using such data, and there would still be a value-add for the paid version. They have a free non-commercial research API available through Data Unison, it seems, but I can't tell how much data is available through the system (it seems to only cover the past 30 days).
July 02, 2008
Simple Global Pleasures
You might remember the Youtube video of this guy named Matt who did this silly dance and captured it on video everywhere he went a few years ago?
Well.. he's back, with friends.
It's a good video to watch when you worry about things like war, unfair trade practices, poor foreign policy, dictatorship, and so on -- it reminds you that people are globally friendly, silly, happy folk if given a chance. Which is always true, but not always easy to remember.
July 01, 2008
Technology Transformations for the Base of the Pyramid by Al Hammond
BoP Spending
It was a fascinating discussion. I'm a strong supporter of sustainability (otherwise, why bother?) and Al Hammond gives a passionate and convincing argument for the central role of business in creating sustainable solutions. Talking with him beforehand, he mentioned (paraphrasing heavily) comparing the measurable benefits of the past five decades of foreign aid versus the last decade of private sector mobile phone rollouts -- the long-term benefits greatly favor the mobile phones.
I can only imagine that once mBanking really gets rolling, all doubt will be erased that the cell phone has helped the Next Four Billion more than 2.3 trillion dollars in aid. My mind quickly out-paces itself when I begin to ponder mBanking benefits for everyone from rural artisania workers able to take and receive payments for commissioned artwork to p2p payment systems to direct-to-market agricultural benefits...
Now, I have a few outstanding doubts about some parts of these two plans - some scalability and malicious-user problems with the Vietnam model, and some privacy and franchise-enforcement questions with the healthcare idea. Now, I have fewer doubts on both of these concepts, combined, than, say, the OLPC Project (though I strongly believe that a base-of-the-pyramid approach to the OLPC could work well). The huge difference between a BoP, market-driven approach and traditional development is that investors bear the brunt of failed projects, a pleasant change from the recipient country being in deeper debt regardless of the outcomes of debt-financed aid projects. I think traditional development will forever have a role in humanitarian and post-conflict aid, but in infrastructure and service creation, the BoP, private-sector approach will prove long-term much stronger than pure-play foreign aid programs, for the simple reason that it applies reasonable risk management to development projects. What a concept!
Without further ado, my full meeting notes after the jump...
The usual apologies for typoes and terseness applies; I was taking notes on my OLPC XO, which though is a technological marvel, is meant for smaller, less gentler hands.
Low-cost Rural Connectivity Model: Vietnam pilot
NextBillion blog post on rural connectivity , NextBillion on mBanking
- Approach to ict: think about scale; therefore focus on mobile companies, only a market driven approaches will scale
- Pilot in three rural communes in Vietnam using wifi and voip to extend network to rural area; pilot then invite mobile companies backed by intel
- wifi backhaul, but wimax and vsat approaches possible
- Sin Mua commune: pop 2203 mountainous, fiber in 12km and decent power
- ptp antenna with omni distribution 7km hop
- meshed wifi APs ; local calls stay local via switch; EVN telecom
- nghia tho commune: ppo 939; further to fiber, simple install for regular electrician not engineer required
- $1000link can go 65km ptp backhaul
- tinho tho third cmmune, more compex net mesh
- performance
- wifi phone to wifi phone 400 meters range (so 800m total between phones with AP in the middle)
- wifi ph to mobile network 500 m
- laptop internet access 1km
- invisible voip system, just a phone
- economics
- $22k/community with labor and nondiscount/mass equip, likely lower number
- capex 14.47/house. 3/21person
- 20% of gsm network cost
- low power, don't need generators for network devices
- Value prop to customers and mobile providors
- 1/2 of calls stay local to a community => free calls in town b/c never hits network => increased penetration and long-term increased revenue b/c of increased penetration and access to non-local calls via switched net
- 3% teledensity in rural vietnam
- mobile companies can make money off these models
- 30% household penetration generating $4 rev/month, networked infrastructure payback is 1.5yrs
- addresses fears of mobile companies against voip as competition
- Addon benefits
- remote management of switches, auto interference detection and channel switching, auto-rollover to other wifi networks
- shut down virus and p2p traffic
- broadband net allows voice, software flexibility
- easy training of local engineers to build out network, sustainable
- 4G network?? wimax doesn't win out in rural areas in tech and loses w/r/t cost; follow on gsm tech in future?
voip questions
- The wifi is essentially open wifi for laptops; also tourist selling point for free wifi?
- Along with wifi and VOIP rollout, pilot places 3 telecenters in each commune, one at the commune headquarters, one at the main healthcare location, one at the school, with hopes that individuals could set up home-based cybercafes
- negotiated with local radio station to use their tower, gave them an omni AP, wanted internet even though they didn't have computers
- so did they get PCs? pilot started 3 wks back, equipment failure from mistake by other cell provider (corporate sabotage?) and weather problems since, so no reports yet.
- telecom ministry hearing demand from other communes
- SMS over voip network?? no, no sms in pilot; a real dual mode wifi+gsm phone in plan for rollout
- $20 phone + $5 wifi chip + $5 biometric reader = $30 sale price at 1M units phone
- sms capability and biometric reader for full support of mbanking with security of transactions
- connectivity platform
- mbanking about to explode to benefit of rural
- then eduction, entertainment = > huge mkts making basic connectivity look small
- get paid for crop via mobile phone?
Scaling Healthcare and ICT
- building infrastructure including healthcare
- 150+B market in developing world currently for healthcare, 1/3-1/2 on pills
- problems with doctor/hospital avoidance, informal providers, fake drugs, inflated pricing
- last mile healthcare - mi farmicita; franchised, tightly managed distribution platform with high tech info / feedback loops
- franchise distro system of pharm via secret shoppers, audits, etc - to weed out anyone not playing by the rules; provide a private sector discipline that doesn't exist - one strike and you're out
- franchise pharmacies as scalable model, low startup cost, generic localy produced drugs, new techs and new distribution channel
- services: drugs, then diagnostics, minor remote surgery (req. broadband), **grows from pharma to accessible medical kn and diagnostics
- future build out of new techs
- diagnostic tech for matching patient and drug, dna analysis unit for $50 ea, 5 minute analysis, 50 cents/test, don't need dr to read it, tests for specific diseases
- remote practice tech, remote reading of sonogram, etc
- future build out of new techs
- Information usage
- walmart logistics, detailed patient records, remote
- 2 way info - collect patient data for health min, marketers, consumer data, biometric data
- incubation, 3rd party insurance to subsidize access to very low income,
- model assumes internet (which can be sold on the side)
- economics
- pharmacy net - 5M investment, 1-2 yr value is 25M
healthcare Q&A
- How scale at right speed to avoid same problems as doc/hospital avoidance? lots of culture/trust issues aswith any medicine/new tech?
- focus on selling pills, build trust and branding thru good product and fair prices; 1 yr to build trust in kenya, build trust gradually and also intro svcs gradually, remote treatment requiers good records
- selling of healthcare info?
- local decisions, incentive to use data for revenue, run tests for health ministry for a fee, sell to marketers, insurance; balance with value prop
- hope towards regulatory rules, system built to pass HIPAA standards
- Other models, like similares in Mexico (untested meds)
- Regulatory and trust problems
- Maintenance costs of the remote practice equipment?
- upkeep of diagnostic - consumables all on sample chip of 5 cent, one tool one test, so low upkeep, designed for bop use, low expertise and cheap
- govt role?
- public sector provision problematic due to high red tape, low trust. so lower hanging fruit is private sector; Philippines law allowing remote pharmacist approval to pill dispensing
- Kenya Mexico, Philippines and India as rollout countries
- high variability of prescription standards across countries (doc, pharmacist, anyone sells) and enforcement and behavior varies from that. education of regulators during scale, but value prop is a massive improvement over current
- telemedicine via fee for service private sector care, but maybe a nonprofit insurance system also?
- Current examples of remote practice include Nighthawk, which has radiology expert across the world for late night xray analysis
Some more questions...
- How resistant is the wifi mesh to malicious usage - i.e. someone who sets up their own AP in the mesh to capture data (e.g. mbanking info, healthcare info?) as it goes through, or to reroute calls?
- Healthcare privacy issues will become paramount since the entire model is built on creating consumer trust; one person who's HIV status is leaked kills the project. It's great that the system is built to conform to HIPAA standards, but I think this decision has to be part of the franchise's strictly managed rules, only sell info in aggregate or at least with patient identification info stripped and in sufficient quantities that small communities don't reveal data by accident.
- The franchise model is great, but problematic with free-wheeling copyright/trademark infringement will produce copycats, will require massive marketing system especially when you have to de-authorize a pharmacy.
